Growth-enhancing Effect of Openness to Trade and Migrations: What is the Effective Transmission Channel for Africa?

Growth-enhancing Effect of Openness to Trade and Migrations: What is the Effective Transmission... Abstract This paper investigates the growth-enhancing effect of openness to trade and to migration by focusing on African countries. Relying on robust estimation techniques dealing with both endogeneity and omitted-variables issues, our results put forward the importance of accounting for the type of the partner country. We found evidence that while trade between Africa and industrialised countries has a clear and robust positive impact on Africa’s standards of living, trade with developing countries fails to be growth-enhancing. Moreover, our findings show that migration has no significant effect on per capita income in Africa regardless of the partner. Finally, exploring the trade openness transmission channel, we establish that the growth-enhancing effect of Africa’s trade with industrialised countries mainly occurs through an improvement in total factor productivity. 1. Introduction While a vast literature exists on the link between income and openness to trade,1Frankel and Romer (1999) were the first to offer a convincing causality analysis regarding the income-enhancing effect of trade openness. The authors use the geographic characteristics as an instrument in a gravity-type model to demonstrate a positive effect of trade on per capita income; the main argument being that these factors are plausibly uncorrelated with other determinants of income per person. These findings were confirmed by several subsequent works (see among others Frankel and Rose, 2002; Dollar and Kraay, 2003; Noguer and Siscart, 2005; Freund and Bolaky, 2008), including across different time periods (Irwin and Terviö, 2002). However, consensus is far from clear on this issue. Rodriguez and Rodrik (2000) highlight the non-robust nature of these results once controlled for omitted variables such as distance from the equator or institutions. More recently, Ortega and Peri (2014) go a step further, and argue that the geographical factors used by Frankel and Romer (1999) can also impact income through migration. Geographic characteristics may raise income through the interactions between countries (exchange of ideas, technological diffusion, innovation, investment) and these interactions would be reflected in the mobility of goods (trade) and of people (migration). Thus, trade is not the sole vehicle of globalisation through which interactions between countries promote economic growth. Acknowledging that openness to trade and openness to migration may be both considered as determinants of income,2Ortega and Peri (2014) found evidence of a strong positive effect of openness to migration on long-run per capita income but fail to do so for trade openness. Despite the abundance of the literature, the debate is still open regarding the relationship between income and openness. Indeed, previous studies indiscriminately examine the growth-enhancing effect of openness without accounting for heterogeneity of countries regarding the benefits or costs of openness. This paper fills this gap and focuses on the specific case of Africa. We aim at studying the overall effect of openness on long-term growth in Africa by paying particular attention to the type—African, other developing, developed—of partner countries. To this end, we retain the general trade-growth identification setting of Frankel and Romer (1999),3 and follow Ortega and Peri (2014) in considering that intensity of openness between two countries should be captured by both bilateral trade and bilateral migration. Such a framework is even more relevant in the case of Africa, where openness to global finance is still in its infancy. The choice of Africa and its singularity deserve some comments. Firstly, by scrutinising the architecture of international trade, the case of Africa stands out as unique. As shown in Figure 1, unlike the rest of the world exports of African countries largely focus on commodities, while their imports are dominated by manufactured goods with a similar overall structure to that of developing and industrialised economies. Furthermore, as illustrated by the right side of Figure 2, Africa’s trade (imports and exports) is mainly realised with developed countries. Although this trade orientation could be beneficial for long-term growth in Africa—particularly through improvement in total factor productivity4—this growth is subject to the ups and downs of the terms of trade due to the high concentration of exports on commodities. Figure 1: View largeDownload slide Comparative Structure of International Trade. Notes: The left-hand side (resp. right-hand side) figure reports the percentage of primary commodities and manufactured goods in the total exports (resp. imports) for each region. DE=DevelopingEconomies. Data source: UNCTAD (mean values over the 1995–2014 period) Figure 1: View largeDownload slide Comparative Structure of International Trade. Notes: The left-hand side (resp. right-hand side) figure reports the percentage of primary commodities and manufactured goods in the total exports (resp. imports) for each region. DE=DevelopingEconomies. Data source: UNCTAD (mean values over the 1995–2014 period) Figure 2: View largeDownload slide Openness of Africa (in 2000). Notes: Trade is measured by the sum of imports and exports. Migration from Africa is measured by the stock of African nationals living abroad. Data sources: UNCTAD (trade data) and World Bank (migration data) Figure 2: View largeDownload slide Openness of Africa (in 2000). Notes: Trade is measured by the sum of imports and exports. Migration from Africa is measured by the stock of African nationals living abroad. Data sources: UNCTAD (trade data) and World Bank (migration data) Secondly, statistics on international migration underline that Africa is characterised by (i) strong intra-continental migration (i.e., immigration and emigration within African countries), and (ii) emigration to industrialised OECD countries. As shown in Figure 2, Africa’s openness to migration in 2000 was more than half intra-African, while one-third was directed towards the industrialised OECD countries. This migration structure of Africa can be seen somewhat dichotomous. On the one hand, it may be viewed as detrimental because ‘brain drain’ (emigration of relatively highly educated individuals) could hamper economic development in Africa. On the other hand, it may be considered as an enhancer factor of development in the sense that African nationals living in industrialised countries are vectors of transmission of human and technological capital (education and experience), but also vectors of transmission of financial capital (migrants’ remittances) and better institutions.5 Finally, despite the strong dominance of developed countries in Africa’s trade, some developing economies such as China are gaining more and more market share in Africa since the beginning of the 2000s. If the growth-enhancing effect of openness between Africa and its new developing partners is debatable (see, among others, Lyons and Brown, 2010; He, 2013; Kaplinsky, 2013), these dynamics bring back the old question about the impact of South-South and North-South openness on growth and productivity in the southern countries. Addressing this hot-debated issue is thus worthy of investigation due to the continuously increasing role played by China in African trade. Falling into the strand of the literature initiated by Frankel and Romer (1999) and Romer (1999), our contribution is 3-fold. First, while the previous literature is mainly done at a global level, we pay particular attention to countries’ specificities and heterogeneity in the face of openness by focusing on a panel of African economies. Second, we go further than previous studies by highlighting the importance of the trading partner. We investigate whether the effect of openness to trade and to migration on growth is sensitive to the type (African, other developing, industrialised) of the partner country. In doing so, we also contribute to the very topical debate concerning China-Africa trade links. Third, in addition to the detailed study of the openness-income nexus, we identify the transmission channel through which trade affects growth. Our main results can be summarised as follows. First, we establish a mitigated overall impact of openness on income in Africa. While trade seems to exert a positive effect on income, this impact is not robust to the inclusion of control variables. The influence of immigration is also fragile and depends on the method used to predict the geographic component of openness. Second, we put forward the importance of accounting for the type of the trading partner. Indeed, we found evidence of a clear and robust partner-varying impact of openness for Africa: only trade with industrialised countries has a strong and robust positive impact on income. Compared to Ortega and Peri’s (2014) contribution—which is the closest paper to ours and which insists on the dominant role of migration—we thus rehabilitate the growth-enhancing effect of trade, provided that Africa’s trade partner country is an advanced one. Third, the positive impact of migration from African economies to industrialised countries (emigration for Africa) is not robust. This probably reflects the confrontation between the ‘brain drain’ negative effect and the ‘productivity transfer’ positive impact of emigration for Africa. Moreover, we found that Africa’s openness (both to trade and to migration) with developing and emerging countries—including China—fails to improve per capita income.6 Finally, exploring the openness transmission channel thanks to the income decomposition of Hall and Jones (1999), we establish that the growth-enhancing effect of African trade with industrialised countries mainly occurs through an improvement in total factor productivity. Various sensitivity analyses are provided to assess the robustness of all our findings. The rest of the paper is organised as follows. Section 2 describes our empirical strategy. Section 3 is devoted to the presentation of data. In Section 4, we present and discuss our main results, and provide some robustness checks. Section 5 is dedicated to examining the transmission channel through which openness impacts income. Finally, Section 6 concludes the paper. 2. Empirical strategy Our empirical framework is inspired from Ortega and Peri (2014) which, in turn, extends the specification proposed by Frankel and Romer (1999). To overcome the well-known endogeneity issue in the trade-income relationship, Frankel and Romer (1999) rely on the instrumental variable technique based on a gravity model. They estimate the causal effects of trade on income using cross-country variation in trade flows due to bilateral geography. According to and Ortega specification suffers from a potential omitted-variables problem because trade and migration openness are both influenced by geography. Thus, country’s geographic characteristics can affect income not only through trade but also through migration. Indeed, geographical proximity and accessibility raise income through the interactions between countries (exchange of ideas, technological diffusion, innovation, investment) which would be reflected in the mobility of goods and people (Ortega and Peri, 2014). In other words, trade is not the sole channel through which interactions between countries increase income. Therefore, to fully identify the impact of trade openness, these two vehicles of globalisation should be jointly considered. 2.1 Baseline specification Our empirical model is given by: lnYi=α0+αTTi+αMMi+αPlnPopi+αAlnAreai+β′Xi+ui (1) where Yi denotes per capita income in country i, Ti and Mi represent openness to trade and openness to migration, respectively, Popi and Areai stand for population and area which capture the impact of country size, Xi collects control variables, and ui is the error term. The rationale behind this empirical model is as follows. Classical international trade theory has highlighted that openness to trade increases output through specialisation based on comparative advantages. New trade theory has documented the growth-enhancing role of trade by focusing on the exploitation of increasing returns to scale and network effects (Helpman and Krugman, 1985; and Grossman and Helpman, 1991a, b). The joint impact of trade and migration on income is explained by Ortega and Peri (2014) in a simple multi-country model that features trade and migration flows both across country borders and across regions within the same country. In this model—which extends Alesina et al. (2000)—aggregate production is a function of varieties for intermediate goods and human capital; and each region is endowed with a differentiated good and a differentiated type of labour. Intermediate goods and labour being mobile across regions of different countries but subject to iceberg-type costs, this model derives income per worker as a function of theoretical measures of trade and migration openness which are, respectively, inverse measures of trade and migration costs. Their empirical counterparts are, respectively, trade flows (exports+imports) as share of GDP, and immigration rate (foreign-born) as share of total population. The income-enhancing impact of openness to migration in the theoretical model of Ortega and Peri (2014) operates through an increase in total factor productivity reflecting growing diversity in productive skills caused by immigration. At a first sight, this channel is not very relevant for African economies which are net labour-sending countries: brain drain might negatively affect income per capita by depriving African economies of valuable talents. However, there are many channels through which emigration can promote economic performance in home countries (see Section 5 for an investigation). Foremost, remittances for emigrants can compensate for the loss of workers by enabling households and entrepreneurs to overcome credit constraints and providing an alternative way to finance investment in human and physical capital (Giuliano and Ruiz-Arranz, 2009). Besides, home countries can benefit from human capital of returning migrants (Stark et al., 1997; Beine et al., 2008) and the transfer of knowledge through the diaspora (Ortega and Peri, 2014). Furthermore, since there is strong evidence of the role of institutions in economic development,7 emigration can be profitable to economic growth in the home country by improving the quality of institutions. Indeed, many recent studies in international migration literature highlight the role of emigration in improving institutions (Spilimbergo, 2009; Docquier et al., 2016). Using an international dataset, Spilimbergo (2009) shows that foreign-educated individuals play an important role in fostering democracy in the home country, but only if foreign education is acquired in democratic countries. Based on cross-section and panel analyses for a large sample of developing countries, Docquier et al. (2016) also found that general emigration has a positive effect on the quality of institutions in the home country. Acknowledging the econometric concerns discussed above, Ortega and Peri (2014) propose to instrument both trade openness and openness to migration by their gravity-based predictors. This geography-based prediction of bilateral trade or bilateral migration stock is obtained by estimating the following pseudo-gravity model: lnWij=γ0+γ1lnDistij+γ2lnPopi+γ3lnPopj+γ4lnAreai+γ5lnAreaj+γ6(Landlockedi+Landlockedj)+γ7Borderij+γ8Colonyij+γ9ComLangij+γ10Comcurij+γ11Timeij+γ12lnDistij×Borderij+γ13lnPopi×Borderij+γ14lnPopj×Borderij+γ15lnAreai×Borderij+γ16lnAreaj×Borderij+γ17(Landlockedi+Landlockedj)×Borderij+eij (2) where Wij is either bilateral trade—i.e., the value of trade (exports + imports)8 between countries i and j divided by the GDP of origin country i—or bilateral migration (emigration)—i.e., the stock of migrants born in country j ( i) and living in country i ( j) as share of country i’s population, Distij is the distance between country i and country j, Pop and Area are the same variables defined in (1) and they are included to account for country size, Landlocked is a dummy variable for landlocked countries, Border is a dummy variable to indicate whether countries i and j share a common border, Colony is a dummy for colonial relationship, and ComLang is a dummy for sharing a common official language. Our specification includes an additional variable ( Comcur) compared to Ortega and Peri (2014). This variable aims at capturing the sharing of a common currency and might play an important role since the impact of currency unions on bilateral trade was frequently relayed in related studies (Rose, 2000, 2001; Frankel and Rose, 2002). As argued by the literature on the endogeneity of optimum currency area criteria, sharing a single currency may set motion forces that promote economic integration and then facilitate migration. Following Ortega and Peri (2014), we include time zone differences denoted by Timeij. As mentioned by Head et al. (2009) the impact of time zone differences between the exporting country and its trading partners is ambiguous since two contradictory effects that differ across service subcategories are at play, namely the continuity effect (the ability to operate around the clock) and the synchronisation effect (the need to coordinate during business hours). Since a large part of trade is with immediate neighbours, we finally include interaction terms of border dummy with the distance, population, area and landlocked dummies (Frankel and Romer, 1999). For the sake of completeness and as previously mentioned, it should be emphasised that we mainly focus in this paper on openness in terms of trade and migration. In other words, we do not specifically account for the third dimension of international relationships, namely financial capital mobility. Our choice is justified by the fact that we concentrate on Africa, i.e. on countries for which the financial sector is not highly developed. As it is a well-established fact that weak financial integration does not favour capital mobility,9 we do not include the financial dimension in our estimations. However, although we do not introduce remittances in our gravity model, we fully acknowledge their importance as a source of income growth and we will investigate this transmission channel in Section 5. Once the gravity regressions described by (2) are estimated, we sum them up over partner countries j to obtain the predicted trade and migration openness for each origin country i. More specifically, let Zij be the vector of explanatory variables included in Equation (2), ΓT the vector of coefficients in the bilateral trade regression, and ΓM the corresponding vector for the bilateral migration regression. The gravity-based predictor of trade openness for origin country i, Tˆi, is then obtained by summing bilateral trade up over partner countries j: Tˆi=∑j≠iexp(ΓTZij) (3) Similarly, the gravity-based predictor of migration openness for origin country i, Mˆi, is given by: Mˆi=∑j≠iexp(ΓMZij) (4) 2.2 Identifying partner-varying impact of openness The income-enhancing impact of openness described in Equation (1) is based on the idea that interactions among countries affect income (through trade and migration) in the same way whatever the partner country. However, it is very reasonable to think that the income-enhancing impact of openness (to trade and to migration) depends on the partner country, especially for African economies. First, with regards to openness to trade, new trade theory suggests that a country can obtain advanced technology from its trading partners through trade. If this channel is dominant, countries may benefit more from trading with advanced economies which are more technologically innovative (Yanikkaya, 2003). As a consequence, trade with industrialised countries may be more income-enhancing for Africa than trade with other countries like China and African neighbours. Second, turning to migration openness, its impact should also depend on the partner country. Openness to migration in African countries is mainly characterised by an important intra-continental mobility and emigration to developed countries (Europe, North America). Therefore, because of the channel advocated by Ortega and Peri (2014) —i.e., growing diversity in productive skills—and the aforementioned ambiguous relationship between growth and emigration to developed countries, it is reasonable to think that the effect of intra-African openness to migration (i.e., immigration from African countries) should be different from the impact of African migration with developed countries (which mainly consists in emigration from Africa to advanced countries). To evaluate this partner-varying impact of the two vehicles of globalisation, we consider the following disaggregated model: lnYi=α0+αTSTiS+αMSMiS+αPlnPopi+αAlnAreai+β′Xi+ui (5) where TiS and MiS, respectively, represent trade and migration openness of an African country i with a particular partner country j∈S, S being either the subset of African partners, the subset of partners among developing countries, or the subset of partners among industrialised countries. To better characterise openness with regard to the subset of partners S, openness to trade and to migration are instrumented by estimating the pseudo-gravity model on the considered subset of partners S: lnWijS=γ0S+γ1SlnDistij+γ2SlnPopi+γ3SlnPopj+γ4SlnAreai+γ5SlnAreaj+γ6S(Landlockedi+Landlockedj)+γ7SBorderij+γ8SColonyij+γ9SComLangij+γ10SComcurij+γ11STimeij+γ12SlnDistij×Borderij+γ13SlnPopi×Borderij+γ14SlnPopj×Borderij+γ15SlnAreai×Borderij+γ16SlnAreaj×Borderij+γ17S(Landlockedi+Landlockedj)×Borderij+eijS (6) Based on the estimation of the disaggregated pseudo-gravity model in Equation (6), the gravity-based predictors of trade openness ( TˆiS) and migration openness ( MˆiS) for country i are, respectively, obtained by summing bilateral trade and migration up over partner countries j among the subset S: TˆiS=∑j∈Sexp(ΓTZij) (7) MˆiS=∑j∈Sexp(ΓMZij) (8) 2.3 Control variables and identification The validity of geographically constructed instrumental variables is weakened by the fact that geographical features may influence directly income per capita or indirectly through other channels than openness (Rodriguez and Rodrik, 2000): impacts of location and climate on transport costs, disease burdens, agricultural productivity and natural resources endowment (Gallup et al., 1999), colonial history and institutional quality (Hall and Jones, 1999; Acemoglu et al., 2001). To tackle this econometric issue, we include a set of control variables allowing us to account for all the main potential channels through which geographical features can influence income per capita. More precisely, we consider distance to the equator, the key geographic variable found to increase the odds of European settlements in the country and, therefore, to determine the history of institutions’ quality (Hall and Jones, 1999; Rodriguez and Rodrik, 2000). The choice of this variable deserves some comments. While we do not consider strictly speaking an institutional quality index among our controls in the income model, the distance to equator allows us to account for institutions’ quality in a given country. Specifically, the distance from the equator may be viewed as reflecting the effect of climate or as a proxy for omitted country’s specificities that are correlated with latitude. The underlying idea is that countries which are nearer the equator have generally worse health conditions and institutions (Sachs and Warner, 1997; Hall and Jones, 1999; Easterly and Levine, 2001). Thus, the distance to equator variable allows us to take into account various country’s characteristics among which the quality of institutions in a single proxy which has the advantage of parsimony. Note that, as for remittances and for the sake of completeness, we will investigate further in Section 5 the role of openness to migration in promoting growth. We also consider other controls: a landlocked dummy and distance to the coast to control for transport costs, the percent of land area in geographical tropics to account for agricultural productivity, disease environment (incidence of malaria and yellow fever) that may influence human history, and legal origin from colonial history that matters for economic outcomes (La Porta et al., 1999, 2008). 3. Data Our data are taken from various sources. Data on bilateral trade are collected from the IMF Direction of Trade Statistics (DOTS). The DOTS database contains data on the value of merchandise exports and imports between each country and all its trading partners. The period for which data are available depends on the considered country but for most of them data extend from the 1980s to the present. As in Ortega and Peri (2014), bilateral migration data are taken from Docquier et al. (2010)—a database of bilateral stocks of immigrants (and emigrants) covering the 1990s to 2000s for 194 countries. Data on geographic variables are from the CEPII’s Gravity database described in Head et al. (2010) and from Gallup et al. (1999). We also use the Gravity database for ethnic, linguistic and colonial ties. The real income per person (real PPP-adjusted GDP per person) is collected from the Penn World Tables (version 7.1). Data on nominal GDP and population are taken from the World Bank World Development Indicators (WDI) database. As in Ortega and Peri (2014), we use the database from Acemoglu et al. (2001) for legal origins, oil endowment and disease environment. Paying particular attention to Africa, we consider, in addition to our whole, world sample, a subsample of 52 African countries.10 Table 1 reports some basic descriptive statistics for the main variables considered in the paper. The mean of real GDP per person in the world is $10,732 with a standard deviation of 13,067, while in Africa the mean is only 2,790 with a standard deviation of 4,502. The minimum real GDP per person in the world is in Africa. Average trade share is 54% in the world and 43% in Africa.11 In line with Figure 2, African trade is dominated by trade with industrialised countries, while intra-African trade is very small. Concerning migration (foreign-born) share (per 1,000 population), its mean is 0.05 in the world and 0.03 in Africa, with a standard deviation of 0.08 and 0.06, respectively. Openness to migration in African countries is characterised by an important intra-continental mobility and emigration to industrialised countries. Table 1 also reports descriptive statistics on some of our main control variables (population, area, percent of European descent in 1900). Table 1: Descriptive Statistics Mean Std. dev. Min. Max. N Whole sample Real GDP per person in 2000 (PPP, 2005 USD) 10,732 13,067 180 65,125 187 Trade (in % of GDP) 0.54 0.42 0 2.68 200 Immigration rate 0.05 0.08 0 0.53 200 Population (in thousands) 30,739 119,883 942 1,262,645 200 Area (in sq. kms) 691,427 1,894,387 25 17,075,400 200 Distance to equator (in degrees) 25.79 16.97 0.2 64.18 200 Euro. descent in 1900 (in %) 29.57 41.69 0 100 157 Africa sample Real GDP per person in 2000 (PPP, 2005 USD) 2,790 4,502 180 25,993 52 Trade (in % of GDP) Total 0.43 0.31 0 1.33 52 Intra-African 0.07 0.08 0 0.31 52 With no highly ind. 0.17 0.15 0 0.83 52 With highly ind. 0.25 0.22 0 1 52 Migration rate Total 0.03 0.06 0 0.32 52 Intra-African 0.03 0.06 0 0.31 52 From no highly ind. (including Africa) 0.03 0.06 0 0.31 52 To highly ind. 0.01 0.03 0 0.16 52 Population (in thousands) 15,428 21,909 81.13 122,877 52 Area (in sq. kms) 581,749 643,900 455 2,505,813 52 Distance to equator (in degrees) 13.47 10.05 0.2 36.83 52 Euro. descent in 1900 (in %) 3.64 14.33 0 100 51 Mean Std. dev. Min. Max. N Whole sample Real GDP per person in 2000 (PPP, 2005 USD) 10,732 13,067 180 65,125 187 Trade (in % of GDP) 0.54 0.42 0 2.68 200 Immigration rate 0.05 0.08 0 0.53 200 Population (in thousands) 30,739 119,883 942 1,262,645 200 Area (in sq. kms) 691,427 1,894,387 25 17,075,400 200 Distance to equator (in degrees) 25.79 16.97 0.2 64.18 200 Euro. descent in 1900 (in %) 29.57 41.69 0 100 157 Africa sample Real GDP per person in 2000 (PPP, 2005 USD) 2,790 4,502 180 25,993 52 Trade (in % of GDP) Total 0.43 0.31 0 1.33 52 Intra-African 0.07 0.08 0 0.31 52 With no highly ind. 0.17 0.15 0 0.83 52 With highly ind. 0.25 0.22 0 1 52 Migration rate Total 0.03 0.06 0 0.32 52 Intra-African 0.03 0.06 0 0.31 52 From no highly ind. (including Africa) 0.03 0.06 0 0.31 52 To highly ind. 0.01 0.03 0 0.16 52 Population (in thousands) 15,428 21,909 81.13 122,877 52 Area (in sq. kms) 581,749 643,900 455 2,505,813 52 Distance to equator (in degrees) 13.47 10.05 0.2 36.83 52 Euro. descent in 1900 (in %) 3.64 14.33 0 100 51 Notes: Statistics are reported for the year 2000. N denotes the number of countries. View Large Table 1: Descriptive Statistics Mean Std. dev. Min. Max. N Whole sample Real GDP per person in 2000 (PPP, 2005 USD) 10,732 13,067 180 65,125 187 Trade (in % of GDP) 0.54 0.42 0 2.68 200 Immigration rate 0.05 0.08 0 0.53 200 Population (in thousands) 30,739 119,883 942 1,262,645 200 Area (in sq. kms) 691,427 1,894,387 25 17,075,400 200 Distance to equator (in degrees) 25.79 16.97 0.2 64.18 200 Euro. descent in 1900 (in %) 29.57 41.69 0 100 157 Africa sample Real GDP per person in 2000 (PPP, 2005 USD) 2,790 4,502 180 25,993 52 Trade (in % of GDP) Total 0.43 0.31 0 1.33 52 Intra-African 0.07 0.08 0 0.31 52 With no highly ind. 0.17 0.15 0 0.83 52 With highly ind. 0.25 0.22 0 1 52 Migration rate Total 0.03 0.06 0 0.32 52 Intra-African 0.03 0.06 0 0.31 52 From no highly ind. (including Africa) 0.03 0.06 0 0.31 52 To highly ind. 0.01 0.03 0 0.16 52 Population (in thousands) 15,428 21,909 81.13 122,877 52 Area (in sq. kms) 581,749 643,900 455 2,505,813 52 Distance to equator (in degrees) 13.47 10.05 0.2 36.83 52 Euro. descent in 1900 (in %) 3.64 14.33 0 100 51 Mean Std. dev. Min. Max. N Whole sample Real GDP per person in 2000 (PPP, 2005 USD) 10,732 13,067 180 65,125 187 Trade (in % of GDP) 0.54 0.42 0 2.68 200 Immigration rate 0.05 0.08 0 0.53 200 Population (in thousands) 30,739 119,883 942 1,262,645 200 Area (in sq. kms) 691,427 1,894,387 25 17,075,400 200 Distance to equator (in degrees) 25.79 16.97 0.2 64.18 200 Euro. descent in 1900 (in %) 29.57 41.69 0 100 157 Africa sample Real GDP per person in 2000 (PPP, 2005 USD) 2,790 4,502 180 25,993 52 Trade (in % of GDP) Total 0.43 0.31 0 1.33 52 Intra-African 0.07 0.08 0 0.31 52 With no highly ind. 0.17 0.15 0 0.83 52 With highly ind. 0.25 0.22 0 1 52 Migration rate Total 0.03 0.06 0 0.32 52 Intra-African 0.03 0.06 0 0.31 52 From no highly ind. (including Africa) 0.03 0.06 0 0.31 52 To highly ind. 0.01 0.03 0 0.16 52 Population (in thousands) 15,428 21,909 81.13 122,877 52 Area (in sq. kms) 581,749 643,900 455 2,505,813 52 Distance to equator (in degrees) 13.47 10.05 0.2 36.83 52 Euro. descent in 1900 (in %) 3.64 14.33 0 100 51 Notes: Statistics are reported for the year 2000. N denotes the number of countries. View Large 4. Empirical results 4.1 Gravity estimates for trade and migration Columns (1–4) of Table 2 contain the gravity model estimation results for global openness (trade and migration) as in Ortega and Peri (2014), while columns (5–8) concern openness of Africa to the world. The odd columns display the OLS results, while the even columns are related to the Poisson Pseudo Maximum Likelihood (PPML) non-linear approach. As argued by Silva and Tenreyro (2006), contrary to the log-linearised model estimation by OLS, PPML estimation has two main advantages: it allows to deal with (i) observations of the dependent variable with zero value, and (ii) heteroskedasticity-related issues. We follow the procedure of Silva and Tenreyro (2010) in order to address the identification problem of the (pseudo) maximum likelihood estimates of the Poisson regression models with non-negative values of the dependent variable (bilateral trade or bilateral migration) and a large number of zeros on some regressors. Table 2: Gravity Regression, African Openness with the World All countries African countries Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.85 *** −1.04 *** −1.35 *** −1.40 *** −0.77 *** −0.66 *** −1.52 *** −1.45 *** (0.04) (0.06) (0.04) (0.08) (0.08) (0.11) (0.08) (0.15) Ln pop. origin 0.04 *** −0.08 ** −0.41 *** −0.35 *** 0.05 * −0.16 *** −0.50 *** −0.47 *** (0.02) (0.04) (0.02) (0.07) (0.03) (0.05) (0.04) (0.12) Ln pop. dest. 1.06 *** 0.74 *** 0.65 *** 0.80 *** 1.20 *** 0.97 *** 0.57 *** 0.52 *** (0.01) (0.03) (0.02) (0.08) (0.03) (0.05) (0.04) (0.11) Ln area origin −0.06 *** 0.04 0.19 *** 0.14 *** −0.00 0.12 *** 0.24 *** 0.14 * (0.01) (0.03) (0.02) (0.03) (0.02) (0.04) (0.03) (0.08) Ln area dest. −0.23 *** −0.08 ** −0.08 *** −0.07 −0.33 *** −0.25 *** −0.13 *** −0.01 (0.01) (0.04) (0.02) (0.05) (0.02) (0.06) (0.03) (0.10) Sum landlocked −0.85 *** −0.55 *** −0.28 *** −0.64 *** −0.96 *** −0.96 *** −0.44 *** −0.79 *** (0.04) (0.07) (0.05) (0.13) (0.06) (0.12) (0.08) (0.23) Border −4.60 *** −1.71 −0.17 −5.17 *** 6.21 *** 0.61 0.63 −0.90 (0.87) (1.07) (0.99) (1.69) (1.80) (2.28) (1.68) (2.87) Border*Ln dist. 0.92 *** 0.27 0.07 1.12 *** −0.48 0.40 0.28 0.50 (0.19) (0.29) (0.22) (0.36) (0.42) (0.64) (0.33) (0.60) Border*Ln pop. origin −0.33 *** 0.17 −0.13 0.31 ** −0.18 0.12 0.05 0.31 (0.09) (0.10) (0.09) (0.14) (0.17) (0.23) (0.14) (0.25) Border*Ln pop. dest. −0.18 ** −0.10 −0.25 *** −0.64 *** −0.52 *** −0.69 *** −0.10 −0.20 (0.08) (0.09) (0.09) (0.13) (0.16) (0.22) (0.16) (0.22) Border*Ln area origin 0.01 −0.16 −0.16 −0.46 *** −0.23 −0.44 ** −0.40 *** −0.27 (0.08) (0.10) (0.10) (0.06) (0.17) (0.22) (0.14) (0.21) Border*Ln area dest. −0.04 0.08 0.34 *** 0.31 * 0.22 0.20 0.11 −0.05 (0.09) (0.16) (0.09) (0.17) (0.17) (0.17) (0.14) (0.25) Border*landlocked 0.72 *** 0.65 *** 0.32 ** 0.43 ** 0.86 *** 0.85 *** 0.24 0.47 (0.12) (0.15) (0.12) (0.21) (0.20) (0.31) (0.19) (0.34) Common language 0.67 *** 0.74 *** 0.91 *** 0.93 *** 0.36 *** 0.27 0.65 *** 1.29 *** (0.09) (0.13) (0.10) (0.20) (0.12) (0.21) (0.14) (0.36) Common off. lang. −0.05 −0.03 0.45 *** 0.16 0.47 *** 0.30 0.88 *** 0.11 (0.09) (0.14) (0.10) (0.22) (0.12) (0.22) (0.14) (0.37) Colonial ties 3.16 *** 1.67 *** 1.33 *** 0.95 *** 4.11 *** 2.51 *** 1.24 *** 0.84 * (0.14) (0.14) (0.17) (0.22) (0.18) (0.19) (0.27) (0.44) Origin hegemon −2.29 *** −2.06 *** 0.97 *** 0.46 −3.35 *** −3.07 *** 1.09 *** 0.72 (0.19) (0.21) (0.22) (0.33) (0.29) (0.29) (0.34) (0.62) Time zone diff. 0.13 *** 0.05 *** 0.09 *** −0.03 −0.12 *** −0.00 −0.01 −0.30 *** (0.01) (0.02) (0.01) (0.04) (0.02) (0.04) (0.02) (0.07) Common currency 0.79 *** 0.58 *** 0.99 *** 0.55 ** 0.55 *** −0.19 0.75 *** 0.90 ** (0.12) (0.14) (0.13) (0.25) (0.21) (0.36) (0.21) (0.43) Constant 5.60 *** 0.96 ** −0.79 ** 0.73 −3.72 *** −2.53 *** 0.98 1.49 (0.29) (0.43) (0.33) (0.55) (0.66) (0.92) (0.70) (1.16) Observations 20,980 37,044 8,219 38,612 8,311 16,761 2,426 17,462 R-squared 0.41 0.20 0.44 0.40 0.34 0.19 0.55 0.19 All countries African countries Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.85 *** −1.04 *** −1.35 *** −1.40 *** −0.77 *** −0.66 *** −1.52 *** −1.45 *** (0.04) (0.06) (0.04) (0.08) (0.08) (0.11) (0.08) (0.15) Ln pop. origin 0.04 *** −0.08 ** −0.41 *** −0.35 *** 0.05 * −0.16 *** −0.50 *** −0.47 *** (0.02) (0.04) (0.02) (0.07) (0.03) (0.05) (0.04) (0.12) Ln pop. dest. 1.06 *** 0.74 *** 0.65 *** 0.80 *** 1.20 *** 0.97 *** 0.57 *** 0.52 *** (0.01) (0.03) (0.02) (0.08) (0.03) (0.05) (0.04) (0.11) Ln area origin −0.06 *** 0.04 0.19 *** 0.14 *** −0.00 0.12 *** 0.24 *** 0.14 * (0.01) (0.03) (0.02) (0.03) (0.02) (0.04) (0.03) (0.08) Ln area dest. −0.23 *** −0.08 ** −0.08 *** −0.07 −0.33 *** −0.25 *** −0.13 *** −0.01 (0.01) (0.04) (0.02) (0.05) (0.02) (0.06) (0.03) (0.10) Sum landlocked −0.85 *** −0.55 *** −0.28 *** −0.64 *** −0.96 *** −0.96 *** −0.44 *** −0.79 *** (0.04) (0.07) (0.05) (0.13) (0.06) (0.12) (0.08) (0.23) Border −4.60 *** −1.71 −0.17 −5.17 *** 6.21 *** 0.61 0.63 −0.90 (0.87) (1.07) (0.99) (1.69) (1.80) (2.28) (1.68) (2.87) Border*Ln dist. 0.92 *** 0.27 0.07 1.12 *** −0.48 0.40 0.28 0.50 (0.19) (0.29) (0.22) (0.36) (0.42) (0.64) (0.33) (0.60) Border*Ln pop. origin −0.33 *** 0.17 −0.13 0.31 ** −0.18 0.12 0.05 0.31 (0.09) (0.10) (0.09) (0.14) (0.17) (0.23) (0.14) (0.25) Border*Ln pop. dest. −0.18 ** −0.10 −0.25 *** −0.64 *** −0.52 *** −0.69 *** −0.10 −0.20 (0.08) (0.09) (0.09) (0.13) (0.16) (0.22) (0.16) (0.22) Border*Ln area origin 0.01 −0.16 −0.16 −0.46 *** −0.23 −0.44 ** −0.40 *** −0.27 (0.08) (0.10) (0.10) (0.06) (0.17) (0.22) (0.14) (0.21) Border*Ln area dest. −0.04 0.08 0.34 *** 0.31 * 0.22 0.20 0.11 −0.05 (0.09) (0.16) (0.09) (0.17) (0.17) (0.17) (0.14) (0.25) Border*landlocked 0.72 *** 0.65 *** 0.32 ** 0.43 ** 0.86 *** 0.85 *** 0.24 0.47 (0.12) (0.15) (0.12) (0.21) (0.20) (0.31) (0.19) (0.34) Common language 0.67 *** 0.74 *** 0.91 *** 0.93 *** 0.36 *** 0.27 0.65 *** 1.29 *** (0.09) (0.13) (0.10) (0.20) (0.12) (0.21) (0.14) (0.36) Common off. lang. −0.05 −0.03 0.45 *** 0.16 0.47 *** 0.30 0.88 *** 0.11 (0.09) (0.14) (0.10) (0.22) (0.12) (0.22) (0.14) (0.37) Colonial ties 3.16 *** 1.67 *** 1.33 *** 0.95 *** 4.11 *** 2.51 *** 1.24 *** 0.84 * (0.14) (0.14) (0.17) (0.22) (0.18) (0.19) (0.27) (0.44) Origin hegemon −2.29 *** −2.06 *** 0.97 *** 0.46 −3.35 *** −3.07 *** 1.09 *** 0.72 (0.19) (0.21) (0.22) (0.33) (0.29) (0.29) (0.34) (0.62) Time zone diff. 0.13 *** 0.05 *** 0.09 *** −0.03 −0.12 *** −0.00 −0.01 −0.30 *** (0.01) (0.02) (0.01) (0.04) (0.02) (0.04) (0.02) (0.07) Common currency 0.79 *** 0.58 *** 0.99 *** 0.55 ** 0.55 *** −0.19 0.75 *** 0.90 ** (0.12) (0.14) (0.13) (0.25) (0.21) (0.36) (0.21) (0.43) Constant 5.60 *** 0.96 ** −0.79 ** 0.73 −3.72 *** −2.53 *** 0.98 1.49 (0.29) (0.43) (0.33) (0.55) (0.66) (0.92) (0.70) (1.16) Observations 20,980 37,044 8,219 38,612 8,311 16,761 2,426 17,462 R-squared 0.41 0.20 0.44 0.40 0.34 0.19 0.55 0.19 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Immigration’ reflects migration openness measured by the number of foreign-born living in the country divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large Table 2: Gravity Regression, African Openness with the World All countries African countries Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.85 *** −1.04 *** −1.35 *** −1.40 *** −0.77 *** −0.66 *** −1.52 *** −1.45 *** (0.04) (0.06) (0.04) (0.08) (0.08) (0.11) (0.08) (0.15) Ln pop. origin 0.04 *** −0.08 ** −0.41 *** −0.35 *** 0.05 * −0.16 *** −0.50 *** −0.47 *** (0.02) (0.04) (0.02) (0.07) (0.03) (0.05) (0.04) (0.12) Ln pop. dest. 1.06 *** 0.74 *** 0.65 *** 0.80 *** 1.20 *** 0.97 *** 0.57 *** 0.52 *** (0.01) (0.03) (0.02) (0.08) (0.03) (0.05) (0.04) (0.11) Ln area origin −0.06 *** 0.04 0.19 *** 0.14 *** −0.00 0.12 *** 0.24 *** 0.14 * (0.01) (0.03) (0.02) (0.03) (0.02) (0.04) (0.03) (0.08) Ln area dest. −0.23 *** −0.08 ** −0.08 *** −0.07 −0.33 *** −0.25 *** −0.13 *** −0.01 (0.01) (0.04) (0.02) (0.05) (0.02) (0.06) (0.03) (0.10) Sum landlocked −0.85 *** −0.55 *** −0.28 *** −0.64 *** −0.96 *** −0.96 *** −0.44 *** −0.79 *** (0.04) (0.07) (0.05) (0.13) (0.06) (0.12) (0.08) (0.23) Border −4.60 *** −1.71 −0.17 −5.17 *** 6.21 *** 0.61 0.63 −0.90 (0.87) (1.07) (0.99) (1.69) (1.80) (2.28) (1.68) (2.87) Border*Ln dist. 0.92 *** 0.27 0.07 1.12 *** −0.48 0.40 0.28 0.50 (0.19) (0.29) (0.22) (0.36) (0.42) (0.64) (0.33) (0.60) Border*Ln pop. origin −0.33 *** 0.17 −0.13 0.31 ** −0.18 0.12 0.05 0.31 (0.09) (0.10) (0.09) (0.14) (0.17) (0.23) (0.14) (0.25) Border*Ln pop. dest. −0.18 ** −0.10 −0.25 *** −0.64 *** −0.52 *** −0.69 *** −0.10 −0.20 (0.08) (0.09) (0.09) (0.13) (0.16) (0.22) (0.16) (0.22) Border*Ln area origin 0.01 −0.16 −0.16 −0.46 *** −0.23 −0.44 ** −0.40 *** −0.27 (0.08) (0.10) (0.10) (0.06) (0.17) (0.22) (0.14) (0.21) Border*Ln area dest. −0.04 0.08 0.34 *** 0.31 * 0.22 0.20 0.11 −0.05 (0.09) (0.16) (0.09) (0.17) (0.17) (0.17) (0.14) (0.25) Border*landlocked 0.72 *** 0.65 *** 0.32 ** 0.43 ** 0.86 *** 0.85 *** 0.24 0.47 (0.12) (0.15) (0.12) (0.21) (0.20) (0.31) (0.19) (0.34) Common language 0.67 *** 0.74 *** 0.91 *** 0.93 *** 0.36 *** 0.27 0.65 *** 1.29 *** (0.09) (0.13) (0.10) (0.20) (0.12) (0.21) (0.14) (0.36) Common off. lang. −0.05 −0.03 0.45 *** 0.16 0.47 *** 0.30 0.88 *** 0.11 (0.09) (0.14) (0.10) (0.22) (0.12) (0.22) (0.14) (0.37) Colonial ties 3.16 *** 1.67 *** 1.33 *** 0.95 *** 4.11 *** 2.51 *** 1.24 *** 0.84 * (0.14) (0.14) (0.17) (0.22) (0.18) (0.19) (0.27) (0.44) Origin hegemon −2.29 *** −2.06 *** 0.97 *** 0.46 −3.35 *** −3.07 *** 1.09 *** 0.72 (0.19) (0.21) (0.22) (0.33) (0.29) (0.29) (0.34) (0.62) Time zone diff. 0.13 *** 0.05 *** 0.09 *** −0.03 −0.12 *** −0.00 −0.01 −0.30 *** (0.01) (0.02) (0.01) (0.04) (0.02) (0.04) (0.02) (0.07) Common currency 0.79 *** 0.58 *** 0.99 *** 0.55 ** 0.55 *** −0.19 0.75 *** 0.90 ** (0.12) (0.14) (0.13) (0.25) (0.21) (0.36) (0.21) (0.43) Constant 5.60 *** 0.96 ** −0.79 ** 0.73 −3.72 *** −2.53 *** 0.98 1.49 (0.29) (0.43) (0.33) (0.55) (0.66) (0.92) (0.70) (1.16) Observations 20,980 37,044 8,219 38,612 8,311 16,761 2,426 17,462 R-squared 0.41 0.20 0.44 0.40 0.34 0.19 0.55 0.19 All countries African countries Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.85 *** −1.04 *** −1.35 *** −1.40 *** −0.77 *** −0.66 *** −1.52 *** −1.45 *** (0.04) (0.06) (0.04) (0.08) (0.08) (0.11) (0.08) (0.15) Ln pop. origin 0.04 *** −0.08 ** −0.41 *** −0.35 *** 0.05 * −0.16 *** −0.50 *** −0.47 *** (0.02) (0.04) (0.02) (0.07) (0.03) (0.05) (0.04) (0.12) Ln pop. dest. 1.06 *** 0.74 *** 0.65 *** 0.80 *** 1.20 *** 0.97 *** 0.57 *** 0.52 *** (0.01) (0.03) (0.02) (0.08) (0.03) (0.05) (0.04) (0.11) Ln area origin −0.06 *** 0.04 0.19 *** 0.14 *** −0.00 0.12 *** 0.24 *** 0.14 * (0.01) (0.03) (0.02) (0.03) (0.02) (0.04) (0.03) (0.08) Ln area dest. −0.23 *** −0.08 ** −0.08 *** −0.07 −0.33 *** −0.25 *** −0.13 *** −0.01 (0.01) (0.04) (0.02) (0.05) (0.02) (0.06) (0.03) (0.10) Sum landlocked −0.85 *** −0.55 *** −0.28 *** −0.64 *** −0.96 *** −0.96 *** −0.44 *** −0.79 *** (0.04) (0.07) (0.05) (0.13) (0.06) (0.12) (0.08) (0.23) Border −4.60 *** −1.71 −0.17 −5.17 *** 6.21 *** 0.61 0.63 −0.90 (0.87) (1.07) (0.99) (1.69) (1.80) (2.28) (1.68) (2.87) Border*Ln dist. 0.92 *** 0.27 0.07 1.12 *** −0.48 0.40 0.28 0.50 (0.19) (0.29) (0.22) (0.36) (0.42) (0.64) (0.33) (0.60) Border*Ln pop. origin −0.33 *** 0.17 −0.13 0.31 ** −0.18 0.12 0.05 0.31 (0.09) (0.10) (0.09) (0.14) (0.17) (0.23) (0.14) (0.25) Border*Ln pop. dest. −0.18 ** −0.10 −0.25 *** −0.64 *** −0.52 *** −0.69 *** −0.10 −0.20 (0.08) (0.09) (0.09) (0.13) (0.16) (0.22) (0.16) (0.22) Border*Ln area origin 0.01 −0.16 −0.16 −0.46 *** −0.23 −0.44 ** −0.40 *** −0.27 (0.08) (0.10) (0.10) (0.06) (0.17) (0.22) (0.14) (0.21) Border*Ln area dest. −0.04 0.08 0.34 *** 0.31 * 0.22 0.20 0.11 −0.05 (0.09) (0.16) (0.09) (0.17) (0.17) (0.17) (0.14) (0.25) Border*landlocked 0.72 *** 0.65 *** 0.32 ** 0.43 ** 0.86 *** 0.85 *** 0.24 0.47 (0.12) (0.15) (0.12) (0.21) (0.20) (0.31) (0.19) (0.34) Common language 0.67 *** 0.74 *** 0.91 *** 0.93 *** 0.36 *** 0.27 0.65 *** 1.29 *** (0.09) (0.13) (0.10) (0.20) (0.12) (0.21) (0.14) (0.36) Common off. lang. −0.05 −0.03 0.45 *** 0.16 0.47 *** 0.30 0.88 *** 0.11 (0.09) (0.14) (0.10) (0.22) (0.12) (0.22) (0.14) (0.37) Colonial ties 3.16 *** 1.67 *** 1.33 *** 0.95 *** 4.11 *** 2.51 *** 1.24 *** 0.84 * (0.14) (0.14) (0.17) (0.22) (0.18) (0.19) (0.27) (0.44) Origin hegemon −2.29 *** −2.06 *** 0.97 *** 0.46 −3.35 *** −3.07 *** 1.09 *** 0.72 (0.19) (0.21) (0.22) (0.33) (0.29) (0.29) (0.34) (0.62) Time zone diff. 0.13 *** 0.05 *** 0.09 *** −0.03 −0.12 *** −0.00 −0.01 −0.30 *** (0.01) (0.02) (0.01) (0.04) (0.02) (0.04) (0.02) (0.07) Common currency 0.79 *** 0.58 *** 0.99 *** 0.55 ** 0.55 *** −0.19 0.75 *** 0.90 ** (0.12) (0.14) (0.13) (0.25) (0.21) (0.36) (0.21) (0.43) Constant 5.60 *** 0.96 ** −0.79 ** 0.73 −3.72 *** −2.53 *** 0.98 1.49 (0.29) (0.43) (0.33) (0.55) (0.66) (0.92) (0.70) (1.16) Observations 20,980 37,044 8,219 38,612 8,311 16,761 2,426 17,462 R-squared 0.41 0.20 0.44 0.40 0.34 0.19 0.55 0.19 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Immigration’ reflects migration openness measured by the number of foreign-born living in the country divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large As shown in Table 2, except for some quantitative differences, the two estimators produce broadly similar results about the nature of the relationship between exogenous factors and the endogenous variable. Focusing on the sample as a whole, we found that countries which trade more with each other are those that are geographically closer, speak the same language, have colonial ties and share a common currency. We also note that the intensity of trade between two countries increases with the size of the destination country and time zone differences, but decreases with the size of the country of origin, the surface area of the destination country and landlockedness. The same links are qualitatively observed between the regressors and the intensity of migration across countries. On the whole, our results are in line with those of Ortega and Peri (2014), the main differences being the introduction among regressors of a dummy to capture the sharing of a common currency, the use of the IMF trade database (instead of the NBER-UN dataset) and the sample of countries which is larger in our study. Although these results are qualitatively consistent with expectations, their quantitative interpretation should be done with caution, especially because of the interaction terms that should be considered. For instance, the negative sign associated with the variable reflecting sharing a common border cannot be interpreted as a negative impact of the border because of its interactions with other variables.12 Let us now focus on the specific and interesting case of Africa. Africa being characterised by strong intra-continental migration and emigration in the industrialised world, we first look at migration from the perspective of the destination country (immigration) and consider indiscriminately—as Ortega and Peri (2014)—the partners of Africa. Then, we go a step further with the aim of identifying a possible differentiated impact of openness depending on the partner. In this case, our variable of interest becomes immigration in the context of intra-continental openness and more generally for openness with developing countries. Regarding the relations with industrialised countries, our variable of interest is rather emigration from the perspective of Africa. As shown in Table 2 (columns (5–8)), we found as for the whole sample that geographical, cultural and historical factors largely explain the intensity of bilateral trade. Qualitatively, the results are very close to those obtained for the whole sample with the only difference that in the latter case, the impact of time zone differences is negative. This is not surprising given that the time zone effect depends on the countries and, especially, the type of services (Head et al., 2009). Table 3 reports the results concerning openness between African countries themselves (columns (1–4)), and between Africa and developing countries (columns (5–8)). Overall, these results are consistent with our previous findings. They confirm that a country has more links (trade and migration) with those that are closer geographically and culturally, more populated, with a coastline or with those with which it shares the same currency. These results also corroborate the fact that trade and migration are both explained by the same factors, as pointed out by Ortega and Peri (2014). In most cases, the explanatory power of migration models is higher than that of trade models. Table 3: Gravity Regression, African Openness with Developing Countries Intra-Africa With no industrialised Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.30 *** −1.25 *** −1.50 *** −1.30 *** −1.17 *** −1.19 *** −1.71 *** −1.50 *** (0.17) (0.15) (0.17) (0.34) (0.08) (0.11) (0.07) (0.12) Ln pop. origin −0.01 0.22 −0.88 *** −0.85 *** 0.15 *** −0.07 −0.40 *** −0.47 *** (0.09) (0.15) (0.08) (0.25) (0.03) (0.06) (0.04) (0.11) Ln pop. dest. 0.85 *** 0.90 *** 0.22 ** 0.15 1.09 *** 0.97 *** 0.59 *** 0.93 *** (0.08) (0.09) (0.10) (0.15) (0.03) (0.08) (0.04) (0.12) Ln area origin −0.17 ** −0.22 0.02 0.34 ** −0.08 *** 0.06 0.03 0.17 *** (0.07) (0.16) (0.07) (0.17) (0.02) (0.05) (0.03) (0.05) Ln area dest. −0.15 ** 0.12 0.02 0.04 −0.29 *** −0.22 *** −0.04 −0.03 (0.07) (0.16) (0.07) (0.17) (0.02) (0.08) (0.03) (0.08) Sum landlocked −0.94 *** −0.50 * −0.07 −0.58 ** −0.98 *** −0.85 *** −0.19 *** −0.78 *** (0.13) (0.28) (0.13) (0.28) (0.06) (0.20) (0.07) (0.20) Border 0.19 −6.28 *** −1.25 −1.02 2.94 −3.19 −3.63 *** −3.80 ** (2.13) (2.19) (1.83) (3.41) (1.80) (2.04) (1.12) (1.69) Border*Ln dist. 0.41 1.54 *** 0.18 0.54 −0.08 0.91 0.51 ** 0.99 *** (0.46) (0.53) (0.36) (0.69) (0.43) (0.56) (0.24) (0.37) Border*Ln pop. origin −0.04 −0.32 0.38 ** 0.77 ** −0.26 −0.17 −0.16 0.23 (0.20) (0.22) (0.15) (0.35) (0.17) (0.18) (0.10) (0.16) Border*Ln pop. dest. −0.05 −0.21 0.27 0.22 −0.37 ** −0.31 −0.21 ** −0.68 *** (0.18) (0.19) (0.17) (0.24) (0.16) (0.23) (0.10) (0.15) Border*Ln area origin −0.18 −0.12 −0.04 −0.47 * −0.14 −0.21 −0.00 −0.43 *** (0.20) (0.26) (0.15) (0.28) (0.17) (0.19) (0.11) (0.06) Border*Ln area dest. −0.09 −0.30 −0.06 −0.11 0.17 0.03 0.30 *** 0.26 (0.19) (0.22) (0.15) (0.29) (0.17) (0.17) (0.10) (0.20) Border*landlocked 0.74 *** 0.34 −0.17 0.20 0.86 *** 0.65 * 0.28 * 0.46 * (0.24) (0.34) (0.21) (0.35) (0.21) (0.34) (0.14) (0.27) Common language 0.50 *** 0.65 ** 0.37 ** 1.07 *** 0.43 *** 0.03 0.94 *** 0.93 *** (0.19) (0.27) (0.16) (0.39) (0.12) (0.27) (0.14) (0.25) Common off. lang. 0.72 *** 0.33 0.01 −0.39 0.36 *** 0.51 * 0.39 *** 0.03 (0.19) (0.23) (0.17) (0.40) (0.12) (0.30) (0.14) (0.25) Colonial ties 0.62 −2.41 ** −1.39 −0.69 1.92 *** −0.76 0.92 ** 0.73 * (0.44) (0.99) (1.06) (0.67) (0.51) (0.71) (0.45) (0.41) Origin hegemon −1.41 *** −0.45 2.19 * 0.15 −0.81 0.61 0.71 −0.08 (0.39) (1.12) (1.24) (1.01) (0.56) (0.77) (0.64) (0.65) Time zone diff. −0.43 *** −0.62 *** 0.02 −0.67 *** −0.03 ** 0.09 *** 0.05 ** −0.20 *** (0.09) (0.20) (0.09) (0.21) (0.02) (0.03) (0.02) (0.07) Common currency 0.37 −0.24 0.77 *** 1.08 ** 0.74 *** −0.05 1.30 *** 0.90 *** (0.24) (0.34) (0.20) (0.44) (0.21) (0.37) (0.21) (0.28) Constant 0.94 0.79 3.61 *** 0.90 −0.65 0.94 2.25 *** 0.89 (1.28) (1.14) (1.27) (1.98) (0.67) (0.84) (0.51) (0.78) Observations 1,331 2,450 525 2,550 7,366 15,634 3,760 30,102 R-squared 0.41 0.19 0.67 0.25 0.34 0.09 0.53 0.54 Intra-Africa With no industrialised Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.30 *** −1.25 *** −1.50 *** −1.30 *** −1.17 *** −1.19 *** −1.71 *** −1.50 *** (0.17) (0.15) (0.17) (0.34) (0.08) (0.11) (0.07) (0.12) Ln pop. origin −0.01 0.22 −0.88 *** −0.85 *** 0.15 *** −0.07 −0.40 *** −0.47 *** (0.09) (0.15) (0.08) (0.25) (0.03) (0.06) (0.04) (0.11) Ln pop. dest. 0.85 *** 0.90 *** 0.22 ** 0.15 1.09 *** 0.97 *** 0.59 *** 0.93 *** (0.08) (0.09) (0.10) (0.15) (0.03) (0.08) (0.04) (0.12) Ln area origin −0.17 ** −0.22 0.02 0.34 ** −0.08 *** 0.06 0.03 0.17 *** (0.07) (0.16) (0.07) (0.17) (0.02) (0.05) (0.03) (0.05) Ln area dest. −0.15 ** 0.12 0.02 0.04 −0.29 *** −0.22 *** −0.04 −0.03 (0.07) (0.16) (0.07) (0.17) (0.02) (0.08) (0.03) (0.08) Sum landlocked −0.94 *** −0.50 * −0.07 −0.58 ** −0.98 *** −0.85 *** −0.19 *** −0.78 *** (0.13) (0.28) (0.13) (0.28) (0.06) (0.20) (0.07) (0.20) Border 0.19 −6.28 *** −1.25 −1.02 2.94 −3.19 −3.63 *** −3.80 ** (2.13) (2.19) (1.83) (3.41) (1.80) (2.04) (1.12) (1.69) Border*Ln dist. 0.41 1.54 *** 0.18 0.54 −0.08 0.91 0.51 ** 0.99 *** (0.46) (0.53) (0.36) (0.69) (0.43) (0.56) (0.24) (0.37) Border*Ln pop. origin −0.04 −0.32 0.38 ** 0.77 ** −0.26 −0.17 −0.16 0.23 (0.20) (0.22) (0.15) (0.35) (0.17) (0.18) (0.10) (0.16) Border*Ln pop. dest. −0.05 −0.21 0.27 0.22 −0.37 ** −0.31 −0.21 ** −0.68 *** (0.18) (0.19) (0.17) (0.24) (0.16) (0.23) (0.10) (0.15) Border*Ln area origin −0.18 −0.12 −0.04 −0.47 * −0.14 −0.21 −0.00 −0.43 *** (0.20) (0.26) (0.15) (0.28) (0.17) (0.19) (0.11) (0.06) Border*Ln area dest. −0.09 −0.30 −0.06 −0.11 0.17 0.03 0.30 *** 0.26 (0.19) (0.22) (0.15) (0.29) (0.17) (0.17) (0.10) (0.20) Border*landlocked 0.74 *** 0.34 −0.17 0.20 0.86 *** 0.65 * 0.28 * 0.46 * (0.24) (0.34) (0.21) (0.35) (0.21) (0.34) (0.14) (0.27) Common language 0.50 *** 0.65 ** 0.37 ** 1.07 *** 0.43 *** 0.03 0.94 *** 0.93 *** (0.19) (0.27) (0.16) (0.39) (0.12) (0.27) (0.14) (0.25) Common off. lang. 0.72 *** 0.33 0.01 −0.39 0.36 *** 0.51 * 0.39 *** 0.03 (0.19) (0.23) (0.17) (0.40) (0.12) (0.30) (0.14) (0.25) Colonial ties 0.62 −2.41 ** −1.39 −0.69 1.92 *** −0.76 0.92 ** 0.73 * (0.44) (0.99) (1.06) (0.67) (0.51) (0.71) (0.45) (0.41) Origin hegemon −1.41 *** −0.45 2.19 * 0.15 −0.81 0.61 0.71 −0.08 (0.39) (1.12) (1.24) (1.01) (0.56) (0.77) (0.64) (0.65) Time zone diff. −0.43 *** −0.62 *** 0.02 −0.67 *** −0.03 ** 0.09 *** 0.05 ** −0.20 *** (0.09) (0.20) (0.09) (0.21) (0.02) (0.03) (0.02) (0.07) Common currency 0.37 −0.24 0.77 *** 1.08 ** 0.74 *** −0.05 1.30 *** 0.90 *** (0.24) (0.34) (0.20) (0.44) (0.21) (0.37) (0.21) (0.28) Constant 0.94 0.79 3.61 *** 0.90 −0.65 0.94 2.25 *** 0.89 (1.28) (1.14) (1.27) (1.98) (0.67) (0.84) (0.51) (0.78) Observations 1,331 2,450 525 2,550 7,366 15,634 3,760 30,102 R-squared 0.41 0.19 0.67 0.25 0.34 0.09 0.53 0.54 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Immigration’ reflects migration openness measured by the number of foreign-born living in the country divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large Table 3: Gravity Regression, African Openness with Developing Countries Intra-Africa With no industrialised Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.30 *** −1.25 *** −1.50 *** −1.30 *** −1.17 *** −1.19 *** −1.71 *** −1.50 *** (0.17) (0.15) (0.17) (0.34) (0.08) (0.11) (0.07) (0.12) Ln pop. origin −0.01 0.22 −0.88 *** −0.85 *** 0.15 *** −0.07 −0.40 *** −0.47 *** (0.09) (0.15) (0.08) (0.25) (0.03) (0.06) (0.04) (0.11) Ln pop. dest. 0.85 *** 0.90 *** 0.22 ** 0.15 1.09 *** 0.97 *** 0.59 *** 0.93 *** (0.08) (0.09) (0.10) (0.15) (0.03) (0.08) (0.04) (0.12) Ln area origin −0.17 ** −0.22 0.02 0.34 ** −0.08 *** 0.06 0.03 0.17 *** (0.07) (0.16) (0.07) (0.17) (0.02) (0.05) (0.03) (0.05) Ln area dest. −0.15 ** 0.12 0.02 0.04 −0.29 *** −0.22 *** −0.04 −0.03 (0.07) (0.16) (0.07) (0.17) (0.02) (0.08) (0.03) (0.08) Sum landlocked −0.94 *** −0.50 * −0.07 −0.58 ** −0.98 *** −0.85 *** −0.19 *** −0.78 *** (0.13) (0.28) (0.13) (0.28) (0.06) (0.20) (0.07) (0.20) Border 0.19 −6.28 *** −1.25 −1.02 2.94 −3.19 −3.63 *** −3.80 ** (2.13) (2.19) (1.83) (3.41) (1.80) (2.04) (1.12) (1.69) Border*Ln dist. 0.41 1.54 *** 0.18 0.54 −0.08 0.91 0.51 ** 0.99 *** (0.46) (0.53) (0.36) (0.69) (0.43) (0.56) (0.24) (0.37) Border*Ln pop. origin −0.04 −0.32 0.38 ** 0.77 ** −0.26 −0.17 −0.16 0.23 (0.20) (0.22) (0.15) (0.35) (0.17) (0.18) (0.10) (0.16) Border*Ln pop. dest. −0.05 −0.21 0.27 0.22 −0.37 ** −0.31 −0.21 ** −0.68 *** (0.18) (0.19) (0.17) (0.24) (0.16) (0.23) (0.10) (0.15) Border*Ln area origin −0.18 −0.12 −0.04 −0.47 * −0.14 −0.21 −0.00 −0.43 *** (0.20) (0.26) (0.15) (0.28) (0.17) (0.19) (0.11) (0.06) Border*Ln area dest. −0.09 −0.30 −0.06 −0.11 0.17 0.03 0.30 *** 0.26 (0.19) (0.22) (0.15) (0.29) (0.17) (0.17) (0.10) (0.20) Border*landlocked 0.74 *** 0.34 −0.17 0.20 0.86 *** 0.65 * 0.28 * 0.46 * (0.24) (0.34) (0.21) (0.35) (0.21) (0.34) (0.14) (0.27) Common language 0.50 *** 0.65 ** 0.37 ** 1.07 *** 0.43 *** 0.03 0.94 *** 0.93 *** (0.19) (0.27) (0.16) (0.39) (0.12) (0.27) (0.14) (0.25) Common off. lang. 0.72 *** 0.33 0.01 −0.39 0.36 *** 0.51 * 0.39 *** 0.03 (0.19) (0.23) (0.17) (0.40) (0.12) (0.30) (0.14) (0.25) Colonial ties 0.62 −2.41 ** −1.39 −0.69 1.92 *** −0.76 0.92 ** 0.73 * (0.44) (0.99) (1.06) (0.67) (0.51) (0.71) (0.45) (0.41) Origin hegemon −1.41 *** −0.45 2.19 * 0.15 −0.81 0.61 0.71 −0.08 (0.39) (1.12) (1.24) (1.01) (0.56) (0.77) (0.64) (0.65) Time zone diff. −0.43 *** −0.62 *** 0.02 −0.67 *** −0.03 ** 0.09 *** 0.05 ** −0.20 *** (0.09) (0.20) (0.09) (0.21) (0.02) (0.03) (0.02) (0.07) Common currency 0.37 −0.24 0.77 *** 1.08 ** 0.74 *** −0.05 1.30 *** 0.90 *** (0.24) (0.34) (0.20) (0.44) (0.21) (0.37) (0.21) (0.28) Constant 0.94 0.79 3.61 *** 0.90 −0.65 0.94 2.25 *** 0.89 (1.28) (1.14) (1.27) (1.98) (0.67) (0.84) (0.51) (0.78) Observations 1,331 2,450 525 2,550 7,366 15,634 3,760 30,102 R-squared 0.41 0.19 0.67 0.25 0.34 0.09 0.53 0.54 Intra-Africa With no industrialised Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.30 *** −1.25 *** −1.50 *** −1.30 *** −1.17 *** −1.19 *** −1.71 *** −1.50 *** (0.17) (0.15) (0.17) (0.34) (0.08) (0.11) (0.07) (0.12) Ln pop. origin −0.01 0.22 −0.88 *** −0.85 *** 0.15 *** −0.07 −0.40 *** −0.47 *** (0.09) (0.15) (0.08) (0.25) (0.03) (0.06) (0.04) (0.11) Ln pop. dest. 0.85 *** 0.90 *** 0.22 ** 0.15 1.09 *** 0.97 *** 0.59 *** 0.93 *** (0.08) (0.09) (0.10) (0.15) (0.03) (0.08) (0.04) (0.12) Ln area origin −0.17 ** −0.22 0.02 0.34 ** −0.08 *** 0.06 0.03 0.17 *** (0.07) (0.16) (0.07) (0.17) (0.02) (0.05) (0.03) (0.05) Ln area dest. −0.15 ** 0.12 0.02 0.04 −0.29 *** −0.22 *** −0.04 −0.03 (0.07) (0.16) (0.07) (0.17) (0.02) (0.08) (0.03) (0.08) Sum landlocked −0.94 *** −0.50 * −0.07 −0.58 ** −0.98 *** −0.85 *** −0.19 *** −0.78 *** (0.13) (0.28) (0.13) (0.28) (0.06) (0.20) (0.07) (0.20) Border 0.19 −6.28 *** −1.25 −1.02 2.94 −3.19 −3.63 *** −3.80 ** (2.13) (2.19) (1.83) (3.41) (1.80) (2.04) (1.12) (1.69) Border*Ln dist. 0.41 1.54 *** 0.18 0.54 −0.08 0.91 0.51 ** 0.99 *** (0.46) (0.53) (0.36) (0.69) (0.43) (0.56) (0.24) (0.37) Border*Ln pop. origin −0.04 −0.32 0.38 ** 0.77 ** −0.26 −0.17 −0.16 0.23 (0.20) (0.22) (0.15) (0.35) (0.17) (0.18) (0.10) (0.16) Border*Ln pop. dest. −0.05 −0.21 0.27 0.22 −0.37 ** −0.31 −0.21 ** −0.68 *** (0.18) (0.19) (0.17) (0.24) (0.16) (0.23) (0.10) (0.15) Border*Ln area origin −0.18 −0.12 −0.04 −0.47 * −0.14 −0.21 −0.00 −0.43 *** (0.20) (0.26) (0.15) (0.28) (0.17) (0.19) (0.11) (0.06) Border*Ln area dest. −0.09 −0.30 −0.06 −0.11 0.17 0.03 0.30 *** 0.26 (0.19) (0.22) (0.15) (0.29) (0.17) (0.17) (0.10) (0.20) Border*landlocked 0.74 *** 0.34 −0.17 0.20 0.86 *** 0.65 * 0.28 * 0.46 * (0.24) (0.34) (0.21) (0.35) (0.21) (0.34) (0.14) (0.27) Common language 0.50 *** 0.65 ** 0.37 ** 1.07 *** 0.43 *** 0.03 0.94 *** 0.93 *** (0.19) (0.27) (0.16) (0.39) (0.12) (0.27) (0.14) (0.25) Common off. lang. 0.72 *** 0.33 0.01 −0.39 0.36 *** 0.51 * 0.39 *** 0.03 (0.19) (0.23) (0.17) (0.40) (0.12) (0.30) (0.14) (0.25) Colonial ties 0.62 −2.41 ** −1.39 −0.69 1.92 *** −0.76 0.92 ** 0.73 * (0.44) (0.99) (1.06) (0.67) (0.51) (0.71) (0.45) (0.41) Origin hegemon −1.41 *** −0.45 2.19 * 0.15 −0.81 0.61 0.71 −0.08 (0.39) (1.12) (1.24) (1.01) (0.56) (0.77) (0.64) (0.65) Time zone diff. −0.43 *** −0.62 *** 0.02 −0.67 *** −0.03 ** 0.09 *** 0.05 ** −0.20 *** (0.09) (0.20) (0.09) (0.21) (0.02) (0.03) (0.02) (0.07) Common currency 0.37 −0.24 0.77 *** 1.08 ** 0.74 *** −0.05 1.30 *** 0.90 *** (0.24) (0.34) (0.20) (0.44) (0.21) (0.37) (0.21) (0.28) Constant 0.94 0.79 3.61 *** 0.90 −0.65 0.94 2.25 *** 0.89 (1.28) (1.14) (1.27) (1.98) (0.67) (0.84) (0.51) (0.78) Observations 1,331 2,450 525 2,550 7,366 15,634 3,760 30,102 R-squared 0.41 0.19 0.67 0.25 0.34 0.09 0.53 0.54 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Immigration’ reflects migration openness measured by the number of foreign-born living in the country divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large Finally, Table 4 provides the estimation results concerning openness between Africa and the industrialised countries. Table 4: Gravity Regression, African Openness with Industrialised Countries Trade Emigration OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.88 *** −0.78 *** −1.04 *** −1.96 *** (0.13) (0.13) (0.15) (0.22) Ln pop. origin −0.13 ** −0.28 *** −0.12 ** −0.05 (0.05) (0.08) (0.05) (0.11) Ln pop. dest. 1.50 *** 1.11 *** 0.77 *** 0.33 *** (0.05) (0.07) (0.05) (0.12) Ln area origin 0.01 0.11 ** −0.22 *** −0.46 *** (0.04) (0.05) (0.04) (0.08) Ln area dest. −0.36 *** −0.28 *** −0.05 0.20 (0.04) (0.08) (0.05) (0.15) Sum landlocked −1.07 *** −0.85 *** −0.93 *** −1.05 *** (0.11) (0.13) (0.11) (0.23) Common language 0.83 *** 0.51 0.45 ** 0.49 * (0.26) (0.31) (0.20) (0.27) Common off. lang. 0.06 0.13 1.40 *** 0.77 * (0.24) (0.29) (0.19) (0.40) Colonial ties 2.46 *** 1.41 *** 1.64 *** 1.40 *** (0.20) (0.23) (0.20) (0.33) Origin hegemon −2.59 *** −2.68 *** (0.31) (0.32) Time zone diff. −0.10 *** 0.04 * 0.04 0.10 * (0.03) (0.03) (0.03) (0.06) Constant −1.61 −0.56 −1.95 8.22 *** (1.13) (1.39) (1.36) (1.95) Observations 2,036 2,300 898 1,173 R-squared 0.49 0.35 0.59 0.46 Trade Emigration OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.88 *** −0.78 *** −1.04 *** −1.96 *** (0.13) (0.13) (0.15) (0.22) Ln pop. origin −0.13 ** −0.28 *** −0.12 ** −0.05 (0.05) (0.08) (0.05) (0.11) Ln pop. dest. 1.50 *** 1.11 *** 0.77 *** 0.33 *** (0.05) (0.07) (0.05) (0.12) Ln area origin 0.01 0.11 ** −0.22 *** −0.46 *** (0.04) (0.05) (0.04) (0.08) Ln area dest. −0.36 *** −0.28 *** −0.05 0.20 (0.04) (0.08) (0.05) (0.15) Sum landlocked −1.07 *** −0.85 *** −0.93 *** −1.05 *** (0.11) (0.13) (0.11) (0.23) Common language 0.83 *** 0.51 0.45 ** 0.49 * (0.26) (0.31) (0.20) (0.27) Common off. lang. 0.06 0.13 1.40 *** 0.77 * (0.24) (0.29) (0.19) (0.40) Colonial ties 2.46 *** 1.41 *** 1.64 *** 1.40 *** (0.20) (0.23) (0.20) (0.33) Origin hegemon −2.59 *** −2.68 *** (0.31) (0.32) Time zone diff. −0.10 *** 0.04 * 0.04 0.10 * (0.03) (0.03) (0.03) (0.06) Constant −1.61 −0.56 −1.95 8.22 *** (1.13) (1.39) (1.36) (1.95) Observations 2,036 2,300 898 1,173 R-squared 0.49 0.35 0.59 0.46 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Emigration’ reflects migration openness measured by the number of nationals living abroad divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large Table 4: Gravity Regression, African Openness with Industrialised Countries Trade Emigration OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.88 *** −0.78 *** −1.04 *** −1.96 *** (0.13) (0.13) (0.15) (0.22) Ln pop. origin −0.13 ** −0.28 *** −0.12 ** −0.05 (0.05) (0.08) (0.05) (0.11) Ln pop. dest. 1.50 *** 1.11 *** 0.77 *** 0.33 *** (0.05) (0.07) (0.05) (0.12) Ln area origin 0.01 0.11 ** −0.22 *** −0.46 *** (0.04) (0.05) (0.04) (0.08) Ln area dest. −0.36 *** −0.28 *** −0.05 0.20 (0.04) (0.08) (0.05) (0.15) Sum landlocked −1.07 *** −0.85 *** −0.93 *** −1.05 *** (0.11) (0.13) (0.11) (0.23) Common language 0.83 *** 0.51 0.45 ** 0.49 * (0.26) (0.31) (0.20) (0.27) Common off. lang. 0.06 0.13 1.40 *** 0.77 * (0.24) (0.29) (0.19) (0.40) Colonial ties 2.46 *** 1.41 *** 1.64 *** 1.40 *** (0.20) (0.23) (0.20) (0.33) Origin hegemon −2.59 *** −2.68 *** (0.31) (0.32) Time zone diff. −0.10 *** 0.04 * 0.04 0.10 * (0.03) (0.03) (0.03) (0.06) Constant −1.61 −0.56 −1.95 8.22 *** (1.13) (1.39) (1.36) (1.95) Observations 2,036 2,300 898 1,173 R-squared 0.49 0.35 0.59 0.46 Trade Emigration OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.88 *** −0.78 *** −1.04 *** −1.96 *** (0.13) (0.13) (0.15) (0.22) Ln pop. origin −0.13 ** −0.28 *** −0.12 ** −0.05 (0.05) (0.08) (0.05) (0.11) Ln pop. dest. 1.50 *** 1.11 *** 0.77 *** 0.33 *** (0.05) (0.07) (0.05) (0.12) Ln area origin 0.01 0.11 ** −0.22 *** −0.46 *** (0.04) (0.05) (0.04) (0.08) Ln area dest. −0.36 *** −0.28 *** −0.05 0.20 (0.04) (0.08) (0.05) (0.15) Sum landlocked −1.07 *** −0.85 *** −0.93 *** −1.05 *** (0.11) (0.13) (0.11) (0.23) Common language 0.83 *** 0.51 0.45 ** 0.49 * (0.26) (0.31) (0.20) (0.27) Common off. lang. 0.06 0.13 1.40 *** 0.77 * (0.24) (0.29) (0.19) (0.40) Colonial ties 2.46 *** 1.41 *** 1.64 *** 1.40 *** (0.20) (0.23) (0.20) (0.33) Origin hegemon −2.59 *** −2.68 *** (0.31) (0.32) Time zone diff. −0.10 *** 0.04 * 0.04 0.10 * (0.03) (0.03) (0.03) (0.06) Constant −1.61 −0.56 −1.95 8.22 *** (1.13) (1.39) (1.36) (1.95) Observations 2,036 2,300 898 1,173 R-squared 0.49 0.35 0.59 0.46 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Emigration’ reflects migration openness measured by the number of nationals living abroad divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large These results are also consistent with theoretical predictions. In particular, geographical, cultural and historical factors explain well bilateral emigration from African countries to industrialised economies. It should be noted that, as Africa has no common border with industrialised countries in our sample, the dummy variable for common border and its interactions with other variables do not appear in the results. For the sake of completeness and as a robustness check, we also investigate the relationship between actual and constructed openness. This link is displayed in the corresponding scatterplots reproduced in Figures 3 and 4. Figure 3 depicts overall openness against its gravity-predicted value for the world and for African openness with the world, based on both OLS and non-linear (Poisson) estimations. For both trade and migration, observed openness is highly correlated with its predicted measure, particularly for the African sample. Figure 413 displays the same relationship for the African sample depending on partners: intra-African trade, African trade with low-income countries and with advanced countries, intra-African migration, immigration in Africa from low-income countries including Africa, and emigration from Africa to advanced countries. For each type of disaggregated openness of Africa, there is high correlation between observed openness and its predicted value. In other words, the figure shows that geographic, cultural and historical variables account for the major part of variation in Africa’s openness. On the whole, our gravity-predicted values for openness thus appear to be reasonable proxies for observed values. Figure 3: View largeDownload slide Observed Openness and Predicted Openness. Notes: This figure reports the scatterplots of the relationship between actual and predicted values of trade (left side) and migration (right side). Figure 3: View largeDownload slide Observed Openness and Predicted Openness. Notes: This figure reports the scatterplots of the relationship between actual and predicted values of trade (left side) and migration (right side). Figure 4: View largeDownload slide Observed Openness and Predicted Openness, depending on Partners. Notes: This figure reports the scatterplots of the relationship between actual and predicted values of trade (left side) and migration (right side) for the subsample of African countries. Figure 4: View largeDownload slide Observed Openness and Predicted Openness, depending on Partners. Notes: This figure reports the scatterplots of the relationship between actual and predicted values of trade (left side) and migration (right side) for the subsample of African countries. 4.2 Income and openness Scatterplots of income per capita against openness to trade and to migration reproduced in Figure 5 provide a first insight about the relationship between income per capita and openness. As shown, for the world sample, there is a positive correlation between income per capita and both trade and migration openness. This positive relationship is also present for Africa overall openness. However, at the disaggregated level, the link between income and openness is found to be related to the type of partners. Indeed, while the relationship between income per person and openness is not positive for intra-African trade and is slightly positive for trade with low-income countries (including Africa), income per capita is positively correlated with trade with advanced countries, intra-African migration, migration from low-income countries and emigration to advanced economies. The next steps give econometric estimations of this openness-growth nexus. Figure 5: View largeDownload slide Income and Openness. Notes: This figure reports the scatterplots of income per capita against openness to trade (left side) and to migration (right side). Figure 5: View largeDownload slide Income and Openness. Notes: This figure reports the scatterplots of income per capita against openness to trade (left side) and to migration (right side). 4.2.1 Baseline results Tables 5 and 6 report the two-stage least-squares (2SLS) joint estimates of the impact of openness to trade and to migration on income per person, using their gravity-predicted measures as instruments. Table 5 displays the results using overall measures of openness for the world and African samples. To clearly highlight our contribution, we start by replicating the results of Ortega and Peri (2014) in columns (1–4). In columns (1) and (2) ((3) and (4)), the estimations are based on linear (non-linear, PPML) gravity-predicted openness. Estimations reported in columns (1) and (3) consider as control variables country size (area and population) and distance to equator—which is the key geographic control identified in the literature (Rodriguez and Rodrik, 2000). In column (1) based on linear predicted openness, the coefficient of openness to immigration is significantly positive while the coefficient of openness to trade is not significant. In column (2), using also linear predicted openness and controlling for other geographic/climate and colonial factors, the impact of immigration increases in level and in significance, while the coefficient of trade remains non significant. Using non-linear predicted openness in columns (3) and (4) as instruments, the impact of immigration remains significantly positive when we consider a comprehensive set of control variables. To sum up, results in columns (1–4) of Table 5 highlight a robust, positive effect of openness to immigration on long-run income per capita at the world level, while there is no evidence of growth-enhancing impact of openness to trade. This finding confirms the results obtained by Ortega and Peri (2014). Table 5: Income and Openness, Baseline Specification World openness African openness LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) Trade −1.80 0.44 1.50 * 0.88 2.80 ** 1.60 3.89 * 4.60 (2.18) (0.59) (0.79) (0.70) (1.37) (1.05) (2.00) (4.72) Immig. 7.72 ** 11.95 *** 1.18 6.73 ** 1.13 11.85 * −4.14 24.13 (3.65) (2.59) (1.06) (3.18) (7.72) (6.05) (11.61) (23.83) Ln pop. 0.03 −0.02 −0.04 −0.06 −0.24 −0.21 −0.26 −0.10 (0.12) (0.06) (0.07) (0.06) (0.18) (0.14) (0.22) (0.32) Ln area −0.21 * 0.25 ** −0.10 0.23 ** 0.05 0.34 * 0.06 0.57 (0.11) (0.11) (0.06) (0.09) (0.11) (0.18) (0.14) (0.56) Dist. equator 0.05 *** 0.01 0.04 *** 0.00 0.05 *** 0.00 0.05 *** 0.01 (0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.06) Constant 8.84 *** 6.98 *** 7.16 *** 7.32 *** 5.53 *** 6.93 *** 5.27 *** 5.47 ** (1.21) (0.78) (0.50) (0.74) (0.49) (0.85) (0.81) (2.20) Observations 187 132 187 132 52 45 52 45 Colonial controls No Yes No Yes No Yes No Yes Geo/climate controls No Yes No Yes No Yes No Yes K-P F-stat 0.887 5.842 5.296 5.726 0.812 1.521 0.772 0.285 SW F-stat for Trade 2.062 13.15 10.97 16.99 8.577 16.30 3.279 0.680 SW F-stat for Mig. 5.635 16.59 55.29 8.622 3.176 3.248 2.413 0.698 SY 10% max IV size 7.030 7.030 7.030 7.030 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 3.630 3.630 3.630 3.630 World openness African openness LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) Trade −1.80 0.44 1.50 * 0.88 2.80 ** 1.60 3.89 * 4.60 (2.18) (0.59) (0.79) (0.70) (1.37) (1.05) (2.00) (4.72) Immig. 7.72 ** 11.95 *** 1.18 6.73 ** 1.13 11.85 * −4.14 24.13 (3.65) (2.59) (1.06) (3.18) (7.72) (6.05) (11.61) (23.83) Ln pop. 0.03 −0.02 −0.04 −0.06 −0.24 −0.21 −0.26 −0.10 (0.12) (0.06) (0.07) (0.06) (0.18) (0.14) (0.22) (0.32) Ln area −0.21 * 0.25 ** −0.10 0.23 ** 0.05 0.34 * 0.06 0.57 (0.11) (0.11) (0.06) (0.09) (0.11) (0.18) (0.14) (0.56) Dist. equator 0.05 *** 0.01 0.04 *** 0.00 0.05 *** 0.00 0.05 *** 0.01 (0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.06) Constant 8.84 *** 6.98 *** 7.16 *** 7.32 *** 5.53 *** 6.93 *** 5.27 *** 5.47 ** (1.21) (0.78) (0.50) (0.74) (0.49) (0.85) (0.81) (2.20) Observations 187 132 187 132 52 45 52 45 Colonial controls No Yes No Yes No Yes No Yes Geo/climate controls No Yes No Yes No Yes No Yes K-P F-stat 0.887 5.842 5.296 5.726 0.812 1.521 0.772 0.285 SW F-stat for Trade 2.062 13.15 10.97 16.99 8.577 16.30 3.279 0.680 SW F-stat for Mig. 5.635 16.59 55.29 8.622 3.176 3.248 2.413 0.698 SY 10% max IV size 7.030 7.030 7.030 7.030 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 3.630 3.630 3.630 3.630 Notes: The dependent variable is the log of income per capita. LP (NLP) stands for linear predicted trade and migration based on the OLS (non-linear Poisson, PPML) gravity estimates. Geographic, climate and disease controls are regional dummies for Africa, a landlocked dummy, the percentage of land in the tropics, average distance to the coast and a measure for oil reserves. Colonial history controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 5: Income and Openness, Baseline Specification World openness African openness LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) Trade −1.80 0.44 1.50 * 0.88 2.80 ** 1.60 3.89 * 4.60 (2.18) (0.59) (0.79) (0.70) (1.37) (1.05) (2.00) (4.72) Immig. 7.72 ** 11.95 *** 1.18 6.73 ** 1.13 11.85 * −4.14 24.13 (3.65) (2.59) (1.06) (3.18) (7.72) (6.05) (11.61) (23.83) Ln pop. 0.03 −0.02 −0.04 −0.06 −0.24 −0.21 −0.26 −0.10 (0.12) (0.06) (0.07) (0.06) (0.18) (0.14) (0.22) (0.32) Ln area −0.21 * 0.25 ** −0.10 0.23 ** 0.05 0.34 * 0.06 0.57 (0.11) (0.11) (0.06) (0.09) (0.11) (0.18) (0.14) (0.56) Dist. equator 0.05 *** 0.01 0.04 *** 0.00 0.05 *** 0.00 0.05 *** 0.01 (0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.06) Constant 8.84 *** 6.98 *** 7.16 *** 7.32 *** 5.53 *** 6.93 *** 5.27 *** 5.47 ** (1.21) (0.78) (0.50) (0.74) (0.49) (0.85) (0.81) (2.20) Observations 187 132 187 132 52 45 52 45 Colonial controls No Yes No Yes No Yes No Yes Geo/climate controls No Yes No Yes No Yes No Yes K-P F-stat 0.887 5.842 5.296 5.726 0.812 1.521 0.772 0.285 SW F-stat for Trade 2.062 13.15 10.97 16.99 8.577 16.30 3.279 0.680 SW F-stat for Mig. 5.635 16.59 55.29 8.622 3.176 3.248 2.413 0.698 SY 10% max IV size 7.030 7.030 7.030 7.030 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 3.630 3.630 3.630 3.630 World openness African openness LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) Trade −1.80 0.44 1.50 * 0.88 2.80 ** 1.60 3.89 * 4.60 (2.18) (0.59) (0.79) (0.70) (1.37) (1.05) (2.00) (4.72) Immig. 7.72 ** 11.95 *** 1.18 6.73 ** 1.13 11.85 * −4.14 24.13 (3.65) (2.59) (1.06) (3.18) (7.72) (6.05) (11.61) (23.83) Ln pop. 0.03 −0.02 −0.04 −0.06 −0.24 −0.21 −0.26 −0.10 (0.12) (0.06) (0.07) (0.06) (0.18) (0.14) (0.22) (0.32) Ln area −0.21 * 0.25 ** −0.10 0.23 ** 0.05 0.34 * 0.06 0.57 (0.11) (0.11) (0.06) (0.09) (0.11) (0.18) (0.14) (0.56) Dist. equator 0.05 *** 0.01 0.04 *** 0.00 0.05 *** 0.00 0.05 *** 0.01 (0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.06) Constant 8.84 *** 6.98 *** 7.16 *** 7.32 *** 5.53 *** 6.93 *** 5.27 *** 5.47 ** (1.21) (0.78) (0.50) (0.74) (0.49) (0.85) (0.81) (2.20) Observations 187 132 187 132 52 45 52 45 Colonial controls No Yes No Yes No Yes No Yes Geo/climate controls No Yes No Yes No Yes No Yes K-P F-stat 0.887 5.842 5.296 5.726 0.812 1.521 0.772 0.285 SW F-stat for Trade 2.062 13.15 10.97 16.99 8.577 16.30 3.279 0.680 SW F-stat for Mig. 5.635 16.59 55.29 8.622 3.176 3.248 2.413 0.698 SY 10% max IV size 7.030 7.030 7.030 7.030 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 3.630 3.630 3.630 3.630 Notes: The dependent variable is the log of income per capita. LP (NLP) stands for linear predicted trade and migration based on the OLS (non-linear Poisson, PPML) gravity estimates. Geographic, climate and disease controls are regional dummies for Africa, a landlocked dummy, the percentage of land in the tropics, average distance to the coast and a measure for oil reserves. Colonial history controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 6: Income and Openness, Identifying Partners’ Impact Intra-African Africa with no ind. Africa with ind. LP LP NLP NLP LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 8.28 13.97 0.86 4.86 4.44 3.50 5.15 6.68 ** 2.08 ** 3.42 ** 3.05 *** 4.51 *** (10.26) (13.00) (2.85) (3.90) (2.87) (3.40) (4.60) (2.92) (0.89) (1.38) (0.93) (1.43) Immig. 0.10 3.40 4.94 6.14 5.36 12.33 ** 6.80 6.21 (4.69) (4.56) (5.20) (4.03) (6.00) (5.22) (16.22) (10.34) Emig. 34.52 ** −2.21 16.16 −9.01 (16.45) (18.41) (11.04) (6.72) Ln pop. −0.45 ** −0.39 * −0.34 ** −0.27 ** −0.30 * −0.15 −0.27 −0.24 −0.26 * −0.30 ** −0.25 * −0.30 ** (0.21) (0.22) (0.14) (0.13) (0.16) (0.14) (0.22) (0.19) (0.15) (0.12) (0.15) (0.14) Ln area 0.15 0.27 0.07 0.17 0.09 0.16 0.09 0.21 * 0.29 0.12 0.15 0.08 (0.16) (0.19) (0.10) (0.11) (0.11) (0.10) (0.14) (0.12) (0.19) (0.13) (0.11) (0.10) Dist. equator 0.05 *** 0.06 *** 0.05 *** 0.06 *** 0.06 *** 0.07 *** 0.06 *** 0.07 *** 0.03 * 0.04 *** 0.04 *** 0.05 *** (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.02) (0.02) (0.02) (0.01) (0.01) (0.01) Constant 6.08 *** 5.14 *** 6.65 *** 5.73 *** 5.63 *** 4.93 *** 5.39 *** 4.55 *** 4.86 *** 5.79 *** 5.46 *** 5.85 *** (1.24) (1.25) (0.64) (0.62) (0.71) (0.95) (0.81) (1.06) (1.10) (0.84) (0.52) (0.45) Observations 52 50 52 50 52 50 52 50 52 50 52 50 Colonial/geo controls No Yes No Yes No Yes No Yes No Yes No Yes K-P F-stat 0.953 0.847 6.171 4.468 1.814 2.123 0.225 3.337 3.738 1.695 9.923 4.850 SW F-stat for Trade 1.905 1.751 9.866 8.120 7.467 7.938 1.088 13.35 19.40 11.32 24.52 10.92 SW F-stat for Mig. 8.659 15.56 5.727 7.168 2.139 3.164 0.604 2.382 9.958 3.580 20.74 7.414 SY 10% max IV size 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 SY 25% max IV size 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 Intra-African Africa with no ind. Africa with ind. LP LP NLP NLP LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 8.28 13.97 0.86 4.86 4.44 3.50 5.15 6.68 ** 2.08 ** 3.42 ** 3.05 *** 4.51 *** (10.26) (13.00) (2.85) (3.90) (2.87) (3.40) (4.60) (2.92) (0.89) (1.38) (0.93) (1.43) Immig. 0.10 3.40 4.94 6.14 5.36 12.33 ** 6.80 6.21 (4.69) (4.56) (5.20) (4.03) (6.00) (5.22) (16.22) (10.34) Emig. 34.52 ** −2.21 16.16 −9.01 (16.45) (18.41) (11.04) (6.72) Ln pop. −0.45 ** −0.39 * −0.34 ** −0.27 ** −0.30 * −0.15 −0.27 −0.24 −0.26 * −0.30 ** −0.25 * −0.30 ** (0.21) (0.22) (0.14) (0.13) (0.16) (0.14) (0.22) (0.19) (0.15) (0.12) (0.15) (0.14) Ln area 0.15 0.27 0.07 0.17 0.09 0.16 0.09 0.21 * 0.29 0.12 0.15 0.08 (0.16) (0.19) (0.10) (0.11) (0.11) (0.10) (0.14) (0.12) (0.19) (0.13) (0.11) (0.10) Dist. equator 0.05 *** 0.06 *** 0.05 *** 0.06 *** 0.06 *** 0.07 *** 0.06 *** 0.07 *** 0.03 * 0.04 *** 0.04 *** 0.05 *** (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.02) (0.02) (0.02) (0.01) (0.01) (0.01) Constant 6.08 *** 5.14 *** 6.65 *** 5.73 *** 5.63 *** 4.93 *** 5.39 *** 4.55 *** 4.86 *** 5.79 *** 5.46 *** 5.85 *** (1.24) (1.25) (0.64) (0.62) (0.71) (0.95) (0.81) (1.06) (1.10) (0.84) (0.52) (0.45) Observations 52 50 52 50 52 50 52 50 52 50 52 50 Colonial/geo controls No Yes No Yes No Yes No Yes No Yes No Yes K-P F-stat 0.953 0.847 6.171 4.468 1.814 2.123 0.225 3.337 3.738 1.695 9.923 4.850 SW F-stat for Trade 1.905 1.751 9.866 8.120 7.467 7.938 1.088 13.35 19.40 11.32 24.52 10.92 SW F-stat for Mig. 8.659 15.56 5.727 7.168 2.139 3.164 0.604 2.382 9.958 3.580 20.74 7.414 SY 10% max IV size 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 SY 25% max IV size 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 Notes: The dependent variable is the log of income per capita. LP (NLP) stands for linear predicted trade and migration based on the OLS (non-linear Poisson, PPML) gravity estimates. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 6: Income and Openness, Identifying Partners’ Impact Intra-African Africa with no ind. Africa with ind. LP LP NLP NLP LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 8.28 13.97 0.86 4.86 4.44 3.50 5.15 6.68 ** 2.08 ** 3.42 ** 3.05 *** 4.51 *** (10.26) (13.00) (2.85) (3.90) (2.87) (3.40) (4.60) (2.92) (0.89) (1.38) (0.93) (1.43) Immig. 0.10 3.40 4.94 6.14 5.36 12.33 ** 6.80 6.21 (4.69) (4.56) (5.20) (4.03) (6.00) (5.22) (16.22) (10.34) Emig. 34.52 ** −2.21 16.16 −9.01 (16.45) (18.41) (11.04) (6.72) Ln pop. −0.45 ** −0.39 * −0.34 ** −0.27 ** −0.30 * −0.15 −0.27 −0.24 −0.26 * −0.30 ** −0.25 * −0.30 ** (0.21) (0.22) (0.14) (0.13) (0.16) (0.14) (0.22) (0.19) (0.15) (0.12) (0.15) (0.14) Ln area 0.15 0.27 0.07 0.17 0.09 0.16 0.09 0.21 * 0.29 0.12 0.15 0.08 (0.16) (0.19) (0.10) (0.11) (0.11) (0.10) (0.14) (0.12) (0.19) (0.13) (0.11) (0.10) Dist. equator 0.05 *** 0.06 *** 0.05 *** 0.06 *** 0.06 *** 0.07 *** 0.06 *** 0.07 *** 0.03 * 0.04 *** 0.04 *** 0.05 *** (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.02) (0.02) (0.02) (0.01) (0.01) (0.01) Constant 6.08 *** 5.14 *** 6.65 *** 5.73 *** 5.63 *** 4.93 *** 5.39 *** 4.55 *** 4.86 *** 5.79 *** 5.46 *** 5.85 *** (1.24) (1.25) (0.64) (0.62) (0.71) (0.95) (0.81) (1.06) (1.10) (0.84) (0.52) (0.45) Observations 52 50 52 50 52 50 52 50 52 50 52 50 Colonial/geo controls No Yes No Yes No Yes No Yes No Yes No Yes K-P F-stat 0.953 0.847 6.171 4.468 1.814 2.123 0.225 3.337 3.738 1.695 9.923 4.850 SW F-stat for Trade 1.905 1.751 9.866 8.120 7.467 7.938 1.088 13.35 19.40 11.32 24.52 10.92 SW F-stat for Mig. 8.659 15.56 5.727 7.168 2.139 3.164 0.604 2.382 9.958 3.580 20.74 7.414 SY 10% max IV size 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 SY 25% max IV size 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 Intra-African Africa with no ind. Africa with ind. LP LP NLP NLP LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 8.28 13.97 0.86 4.86 4.44 3.50 5.15 6.68 ** 2.08 ** 3.42 ** 3.05 *** 4.51 *** (10.26) (13.00) (2.85) (3.90) (2.87) (3.40) (4.60) (2.92) (0.89) (1.38) (0.93) (1.43) Immig. 0.10 3.40 4.94 6.14 5.36 12.33 ** 6.80 6.21 (4.69) (4.56) (5.20) (4.03) (6.00) (5.22) (16.22) (10.34) Emig. 34.52 ** −2.21 16.16 −9.01 (16.45) (18.41) (11.04) (6.72) Ln pop. −0.45 ** −0.39 * −0.34 ** −0.27 ** −0.30 * −0.15 −0.27 −0.24 −0.26 * −0.30 ** −0.25 * −0.30 ** (0.21) (0.22) (0.14) (0.13) (0.16) (0.14) (0.22) (0.19) (0.15) (0.12) (0.15) (0.14) Ln area 0.15 0.27 0.07 0.17 0.09 0.16 0.09 0.21 * 0.29 0.12 0.15 0.08 (0.16) (0.19) (0.10) (0.11) (0.11) (0.10) (0.14) (0.12) (0.19) (0.13) (0.11) (0.10) Dist. equator 0.05 *** 0.06 *** 0.05 *** 0.06 *** 0.06 *** 0.07 *** 0.06 *** 0.07 *** 0.03 * 0.04 *** 0.04 *** 0.05 *** (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.02) (0.02) (0.02) (0.01) (0.01) (0.01) Constant 6.08 *** 5.14 *** 6.65 *** 5.73 *** 5.63 *** 4.93 *** 5.39 *** 4.55 *** 4.86 *** 5.79 *** 5.46 *** 5.85 *** (1.24) (1.25) (0.64) (0.62) (0.71) (0.95) (0.81) (1.06) (1.10) (0.84) (0.52) (0.45) Observations 52 50 52 50 52 50 52 50 52 50 52 50 Colonial/geo controls No Yes No Yes No Yes No Yes No Yes No Yes K-P F-stat 0.953 0.847 6.171 4.468 1.814 2.123 0.225 3.337 3.738 1.695 9.923 4.850 SW F-stat for Trade 1.905 1.751 9.866 8.120 7.467 7.938 1.088 13.35 19.40 11.32 24.52 10.92 SW F-stat for Mig. 8.659 15.56 5.727 7.168 2.139 3.164 0.604 2.382 9.958 3.580 20.74 7.414 SY 10% max IV size 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 SY 25% max IV size 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 Notes: The dependent variable is the log of income per capita. LP (NLP) stands for linear predicted trade and migration based on the OLS (non-linear Poisson, PPML) gravity estimates. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large At this stage, it is important to check the relevance of gravity-based instruments since the lack of significance of trade openness may come from a problem of weak instruments. To this end, we implement (i) the Kleibergen and Paap (2006) rk Wald F-stat test (KP test) which tests for the null hypothesis of jointly weak instruments, and (ii) the Sanderson and Windmeijer (2015) F-stat test (SW test) of weak identification for each endogenous regressor separately.14 Doing so allows us to evaluate whether each individual endogenous regressor is well identified separately, by partialling-out the influence of the other endogenous regressors. For the world sample (columns (1–4) of Table 5), the null of (jointly and individual) weak identifications is rejected at conventional level of significance, except for column (1). Particularly, for the most relevant specification relying on non-linear gravity-based instruments and using all control variables, the KP test statistic for jointly weak identification is 5.96, which is above the Stock and Yogo’s (2005) critical value at 25% max IV size (3.63). For this specification, for each endogenous regressor, we reject the null of individual weak identification. The SW test statistic for individual weak identification is 14.60 for openness to trade and 8.63 for openness to immigration; both values being above all Stock and Yogo’s critical values. On the whole, these results indicate that the weak instrument issue is not a severe concern in our estimations. Let us now contrast the previous results with those obtained on the African sample. Columns (5–8) in Table 5 report the impact of African openness with the world. These results differ from those obtained for the world sample, and highlight the relevance of isolating this subgroup of countries. In particular, focusing on non-linear estimation (columns (3) and (4) and (7) and (8)) two main conclusions can be drawn from our findings. First, comparing columns (3) and (7) on the one hand and (4) and (8)—which include all control variables—on the other hand, highlights that the significance of trade is identical between world openness and African openness. Second, contrasting columns (4) and (8) shows that immigration is significant in the full sample while is not the case in the African sample. On the whole, focusing on our variables of interest, the major difference between the full and the African samples concerns the role of immigration, while the impact of trade remains similar. Finally, looking specifically at the African sample, based on both linear and non-linear predicted openness (columns (5) and (7)), the coefficient of trade is significantly positive when we control for country size and distance to equator. However, when we consider all the other controls, there is no evidence of positive and significant impact of trade, and immigration has a positive significant effect only in the linear case. For all specifications, we are not able to reject the null of weak instruments. 4.2.2 Identifying partner-varying impact of openness The fact that results for Africa differ from those obtained at a world level—especially using migration openness—justifies the need to deepen the analysis of the potential growth-enhancing impact of openness in Africa. Indeed, it is reasonable to think that the difficulty to find strong evidence of the influence of openness at the aggregate level in Africa comes from the characteristics of African openness. As aforementioned, trade with advanced economies may have more income-enhancing impact in Africa than trade with other countries like China and African neighbours. With regards to migration, the impact of Africa’s openness to migration with developed countries (net emigration) may differ from that of intra-African migration (immigration and emigration within African countries). To test these conjectures, we estimate the impact of openness on income in Africa depending on the type of partners; the gravity-predicted openness being derived from the estimation of Equation (6). Regarding migration, as a large part of African migration consists in intra-African migration and emigration to developed countries, we consider three cases: (i) inter-African immigration (columns (1–4)) which refers to immigration from African countries, (ii) immigration from developing countries, including African countries (columns (5–8)), and (iii) emigration to industrialised countries (columns (9–12)).15 The corresponding estimation results are reported in Table 6. The results in columns (1–4) show that there is no evidence of growth-enhancing impact of intra-African openness. Neither intra-continental trade nor immigration from African countries significantly influence growth, based on both linear and non-linear gravity-based instruments. In the non-linear case, the hypothesis of (jointly and individual) weak identification of endogenous regressors cannot be retained. The results reported in columns (5–8) also indicate that there is no strong evidence of income-enhancing impact of openness with low-income countries (including African economies). When we only control for country size (population and area) and for distance to equator (columns (5) and (7)), neither trade nor immigration from non-industrialised countries has a significant impact and we cannot reject the weak identification of endogenous regressors. Including other controls in columns (6) and (8), there is a significant positive impact of trade in the linear case (column (6)) and a significant positive effect of immigration in the non-linear case (column (8)); in both cases, only trade is not weakly identified. Turning to openness with advanced economies (trade with industrialised countries and emigration to industrialised economies), the results are reported in columns (9–12) of Table 6. They show overwhelming evidence of a growth-enhancing impact of trade with industrialised countries. When we only control for country size and distance to equator, we found a positive significant role of both trade and emigration on income per capita in the linear case (column (9)) and a positive significant effect of only trade in the non-linear case (column (10)); in both cases, trade and emigration are not weakly identified. In columns (10) and (12), when we include other control variables, there is only a positive significant impact of trade openness with advanced countries, and both endogenous regressors are not weakly identified, except for emigration in column (10). To sum up, in analysing the impact of openness in Africa depending on partners, we found strong evidence that trade with industrialised economies promotes economic development in African countries, while we do not establish a strong impact of openness to migration contrary to Ortega and Peri (2014). In addition to emphasising the interest of accounting for the type of partner countries, these findings corroborate the theoretical intuition that African countries may benefit more from trading with advanced economies which are more technologically innovative (Yanikkaya, 2003). The underlying idea is that if growth is driven by technological progress, trade allows African countries to benefit from the advances in R&D activities of their trading partners. It is worth mentioning that the type of the partner country is linked to the composition of trade flows. Indeed, as shown by Figures 1 and 2, (i) exports of African countries are mainly composed by primary commodities, while they mostly import manufactured goods, and (ii) Africa’s trade is essentially realised with developed countries which principally export manufactured goods. Such goods, which are imported by African countries, are typically more technology and skill intensive than those from developing countries (including Africa).16 In this sense, the development stage of the trading partner and the composition of trade flows are strongly related. As a result, trade with advanced countries which mainly concerns manufactured goods leads to technology transfers that are benefit for Africa, while it is not the case for trade with developing countries that essentially focuses on primary commodities. Besides, our findings also reflect the aforementioned ambiguous relationship between growth and emigration. In other words, we found a compensation between the adverse impact of emigration (through brain drain) and its positive effect (through remittances, human capital of returning migrants, knowledge transfer, improving institutions).17 Furthermore, the absence of significant impact of immigration from African countries may reflect some lack of complementarity in African labour force. 4.3 Sensitivity analyses For the sake of completeness and as robustness checks, we now provide some sensitivity analyses. First, we investigate the robustness of our results to the African countries included in our sample. To this end, we focus on Sub-Saharan Africa (SSA) by excluding North African countries from the sample because of their geographical proximity to Europe.18 As a second robustness check, we test the sensitivity of our results to the choice of the considered year in line with Irwin and Terviö (2002). In addition to our baseline year 2000, two over years are considered, namely 2005 and 2010. Third, we go further than a usual sensitivity analysis and address a hot-debated topic, namely the trade links between China and Africa. This relationship is of particular interest since trade between China and Africa is growing since the early 2000s while it was insignificant in the 1980s and 1990s. China has become the largest trading partner of many African economies including countries where trade was formerly largely oriented toward Europe or America because of colonial or historical ties, such as Angola, Benin, Ethiopia, Gambia, Ghana, Mauritania, Sudan, Togo and Tanzania. Specifically, China that accounted for only 3% of total African trade in 2000 rose to 12% in 2010 and continues to increase its market share on the continent, while the share of trade with traditional partners of Africa as France and the United States is declining. It is thus worthy of interest to investigate whether trade with this ‘new partner’ is growth-enhancing for Africa. It should be mentioned that we focus on the trade side, since emigration of Africans to China is almost zero. Thus, in terms of openness between China and Africa, only the trade channel is relevant to consider. The results of the sensitivity analyses are displayed in Table 7.19 They confirm the evidence of income-enhancing impact of trade with advanced countries. Indeed, in line with our previous findings, Table 7 shows a significant positive impact of trade with industrialised countries for the subsample of SSA economies and whatever the period (year) of analysis. Besides and interestingly, there is no evidence of a strong positive impact of trade with China. Table 7: Sensitivity Analysis 2000, SSA only 2000 2005 2010 African No ind. ind. China African No ind. ind. China African No ind. ind. China (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 6.29 8.38 * 3.74 *** 42.51 1.61 2.40 3.85 *** 19.38 2.40 2.74 4.48 *** 15.83 ** (7.03) (4.77) (1.44) (30.21) (2.59) (1.55) (1.23) (12.11) (2.82) (1.75) (1.61) (8.00) Immig. 5.15 15.55 * 4.51 1.52 4.42 −9.16 (4.13) (9.24) (4.94) (10.88) (5.72) (23.07) Emig. −5.43 −4.00 −4.73 (5.59) (7.05) (7.88) Ln pop. −0.42 −0.36 −0.31 ** −0.04 −0.25 ** −0.26 * −0.26 * −0.07 −0.25 ** −0.29 −0.22 * 0.09 (0.26) (0.26) (0.15) (0.26) (0.12) (0.15) (0.14) (0.20) (0.11) (0.19) (0.13) (0.19) Ln area 0.19 0.24 0.08 −0.02 0.16 0.21 * 0.10 0.02 0.17 0.22 * 0.10 −0.02 (0.17) (0.18) (0.10) (0.16) (0.11) (0.11) (0.10) (0.13) (0.10) (0.13) (0.10) (0.14) Dist. equator 0.03 0.06 0.04 ** 0.07 *** 0.05 *** 0.06 *** 0.04 *** 0.08 *** 0.05 *** 0.04 * 0.03 ** 0.07 *** (0.02) (0.03) (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.01) (0.01) Constant 6.12 *** 4.32 *** 5.99 *** 5.30 *** 6.07 *** 5.47 *** 5.90 *** 5.56 *** 6.19 *** 5.85 *** 6.06 *** 5.47 *** (0.78) (1.56) (0.56) (1.04) (0.61) (0.79) (0.45) (0.90) (0.63) (0.98) (0.45) (0.91) Observations 44 44 44 50 50 50 50 50 50 50 50 50 Colonial/geo controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes K-P F-stat 2.897 1.063 1.856 1.061 6.071 0.872 7.396 2.354 6.007 0.723 5.954 5.498 SW F-stat for Trade 5.730 3.189 4.209 1.061 6.976 4.358 17.35 2.354 7.019 2.789 12.08 5.498 SW F-stat for Mig. 5.885 1.918 11.27 5.089 1.759 8.124 6.033 1.710 10.54 SY 10% max IV size 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 SY 25% max IV size 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 2000, SSA only 2000 2005 2010 African No ind. ind. China African No ind. ind. China African No ind. ind. China (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 6.29 8.38 * 3.74 *** 42.51 1.61 2.40 3.85 *** 19.38 2.40 2.74 4.48 *** 15.83 ** (7.03) (4.77) (1.44) (30.21) (2.59) (1.55) (1.23) (12.11) (2.82) (1.75) (1.61) (8.00) Immig. 5.15 15.55 * 4.51 1.52 4.42 −9.16 (4.13) (9.24) (4.94) (10.88) (5.72) (23.07) Emig. −5.43 −4.00 −4.73 (5.59) (7.05) (7.88) Ln pop. −0.42 −0.36 −0.31 ** −0.04 −0.25 ** −0.26 * −0.26 * −0.07 −0.25 ** −0.29 −0.22 * 0.09 (0.26) (0.26) (0.15) (0.26) (0.12) (0.15) (0.14) (0.20) (0.11) (0.19) (0.13) (0.19) Ln area 0.19 0.24 0.08 −0.02 0.16 0.21 * 0.10 0.02 0.17 0.22 * 0.10 −0.02 (0.17) (0.18) (0.10) (0.16) (0.11) (0.11) (0.10) (0.13) (0.10) (0.13) (0.10) (0.14) Dist. equator 0.03 0.06 0.04 ** 0.07 *** 0.05 *** 0.06 *** 0.04 *** 0.08 *** 0.05 *** 0.04 * 0.03 ** 0.07 *** (0.02) (0.03) (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.01) (0.01) Constant 6.12 *** 4.32 *** 5.99 *** 5.30 *** 6.07 *** 5.47 *** 5.90 *** 5.56 *** 6.19 *** 5.85 *** 6.06 *** 5.47 *** (0.78) (1.56) (0.56) (1.04) (0.61) (0.79) (0.45) (0.90) (0.63) (0.98) (0.45) (0.91) Observations 44 44 44 50 50 50 50 50 50 50 50 50 Colonial/geo controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes K-P F-stat 2.897 1.063 1.856 1.061 6.071 0.872 7.396 2.354 6.007 0.723 5.954 5.498 SW F-stat for Trade 5.730 3.189 4.209 1.061 6.976 4.358 17.35 2.354 7.019 2.789 12.08 5.498 SW F-stat for Mig. 5.885 1.918 11.27 5.089 1.759 8.124 6.033 1.710 10.54 SY 10% max IV size 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 SY 25% max IV size 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 Notes: The dependent variable is the log of income per capita. Predicted trade and migration are obtained from non-linear Poisson gravity estimates. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 7: Sensitivity Analysis 2000, SSA only 2000 2005 2010 African No ind. ind. China African No ind. ind. China African No ind. ind. China (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 6.29 8.38 * 3.74 *** 42.51 1.61 2.40 3.85 *** 19.38 2.40 2.74 4.48 *** 15.83 ** (7.03) (4.77) (1.44) (30.21) (2.59) (1.55) (1.23) (12.11) (2.82) (1.75) (1.61) (8.00) Immig. 5.15 15.55 * 4.51 1.52 4.42 −9.16 (4.13) (9.24) (4.94) (10.88) (5.72) (23.07) Emig. −5.43 −4.00 −4.73 (5.59) (7.05) (7.88) Ln pop. −0.42 −0.36 −0.31 ** −0.04 −0.25 ** −0.26 * −0.26 * −0.07 −0.25 ** −0.29 −0.22 * 0.09 (0.26) (0.26) (0.15) (0.26) (0.12) (0.15) (0.14) (0.20) (0.11) (0.19) (0.13) (0.19) Ln area 0.19 0.24 0.08 −0.02 0.16 0.21 * 0.10 0.02 0.17 0.22 * 0.10 −0.02 (0.17) (0.18) (0.10) (0.16) (0.11) (0.11) (0.10) (0.13) (0.10) (0.13) (0.10) (0.14) Dist. equator 0.03 0.06 0.04 ** 0.07 *** 0.05 *** 0.06 *** 0.04 *** 0.08 *** 0.05 *** 0.04 * 0.03 ** 0.07 *** (0.02) (0.03) (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.01) (0.01) Constant 6.12 *** 4.32 *** 5.99 *** 5.30 *** 6.07 *** 5.47 *** 5.90 *** 5.56 *** 6.19 *** 5.85 *** 6.06 *** 5.47 *** (0.78) (1.56) (0.56) (1.04) (0.61) (0.79) (0.45) (0.90) (0.63) (0.98) (0.45) (0.91) Observations 44 44 44 50 50 50 50 50 50 50 50 50 Colonial/geo controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes K-P F-stat 2.897 1.063 1.856 1.061 6.071 0.872 7.396 2.354 6.007 0.723 5.954 5.498 SW F-stat for Trade 5.730 3.189 4.209 1.061 6.976 4.358 17.35 2.354 7.019 2.789 12.08 5.498 SW F-stat for Mig. 5.885 1.918 11.27 5.089 1.759 8.124 6.033 1.710 10.54 SY 10% max IV size 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 SY 25% max IV size 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 2000, SSA only 2000 2005 2010 African No ind. ind. China African No ind. ind. China African No ind. ind. China (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 6.29 8.38 * 3.74 *** 42.51 1.61 2.40 3.85 *** 19.38 2.40 2.74 4.48 *** 15.83 ** (7.03) (4.77) (1.44) (30.21) (2.59) (1.55) (1.23) (12.11) (2.82) (1.75) (1.61) (8.00) Immig. 5.15 15.55 * 4.51 1.52 4.42 −9.16 (4.13) (9.24) (4.94) (10.88) (5.72) (23.07) Emig. −5.43 −4.00 −4.73 (5.59) (7.05) (7.88) Ln pop. −0.42 −0.36 −0.31 ** −0.04 −0.25 ** −0.26 * −0.26 * −0.07 −0.25 ** −0.29 −0.22 * 0.09 (0.26) (0.26) (0.15) (0.26) (0.12) (0.15) (0.14) (0.20) (0.11) (0.19) (0.13) (0.19) Ln area 0.19 0.24 0.08 −0.02 0.16 0.21 * 0.10 0.02 0.17 0.22 * 0.10 −0.02 (0.17) (0.18) (0.10) (0.16) (0.11) (0.11) (0.10) (0.13) (0.10) (0.13) (0.10) (0.14) Dist. equator 0.03 0.06 0.04 ** 0.07 *** 0.05 *** 0.06 *** 0.04 *** 0.08 *** 0.05 *** 0.04 * 0.03 ** 0.07 *** (0.02) (0.03) (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.01) (0.01) Constant 6.12 *** 4.32 *** 5.99 *** 5.30 *** 6.07 *** 5.47 *** 5.90 *** 5.56 *** 6.19 *** 5.85 *** 6.06 *** 5.47 *** (0.78) (1.56) (0.56) (1.04) (0.61) (0.79) (0.45) (0.90) (0.63) (0.98) (0.45) (0.91) Observations 44 44 44 50 50 50 50 50 50 50 50 50 Colonial/geo controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes K-P F-stat 2.897 1.063 1.856 1.061 6.071 0.872 7.396 2.354 6.007 0.723 5.954 5.498 SW F-stat for Trade 5.730 3.189 4.209 1.061 6.976 4.358 17.35 2.354 7.019 2.789 12.08 5.498 SW F-stat for Mig. 5.885 1.918 11.27 5.089 1.759 8.124 6.033 1.710 10.54 SY 10% max IV size 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 SY 25% max IV size 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 Notes: The dependent variable is the log of income per capita. Predicted trade and migration are obtained from non-linear Poisson gravity estimates. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large 5. Transmission channels 5.1 The channel through which trade with developed countries affects income in Africa While our previous results show that only trade with developed countries positively affects income in Africa, they do not provide information on the channel through which this positive effect operates. This section tackles this issue by providing some insights regarding those transmission channels. 5.1.1 Export and import sides Trade with developed countries may enhance growth in Africa through (i) the exploitation of economies of scale and comparative advantages which are reflected in exports, or (ii) technology transfers captured by imports. To disentangle these two channels, we estimate our regressions by considering two measures of trade openness, namely exports to GDP ratio and imports to GDP ratio (Yanikkaya, 2003).20 The results are reported in Table 8 (columns (1–4)) and show that trade with developed countries promotes growth in Africa through both exports and imports, with a slightly more pronounced impact for imports. Therefore, our findings support the hypothesis that African countries that are more open to trade with developed countries experience higher growth by importing technologically advanced goods, but also by accessing to a larger market allowing to exploit economies of scale and comparative advantages. Our results are in line with Yanikkaya (2003) regarding trade with the U.S., reflecting that imports and exports are complementary. Overall, it should be noted that while this decomposition of trade is useful in attempting to examine the dominant channel, it is often preferable to aggregate exports and imports in particular because they are complementary. Table 8: The Channels of Growth-enhancing Impact of Trade with Developed Countries (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) lnY/L lnY/L lnY/L lnY/L lnY/L α1−αlnK/Y lnH/L lnA lnY/L α1−αlnK/Y lnH/L lnA Trade 2.28 *** 0.01 0.78 ** 2.04 *** 3.54 ** −0.37 0.58 4.00 *** (0.73) (0.30) (0.36) (0.72) (1.52) (0.48) (0.63) (1.54) Exports 6.18 *** 7.11 *** (1.69) (1.96) Imports 7.94 *** 9.38 ** (2.80) (4.54) Ln pop. −0.19 −0.28 * −0.33 * −0.32 * −0.22 * −0.05 −0.00 −0.16 −0.26 ** −0.08 * −0.05 −0.16 (0.16) (0.15) (0.17) (0.17) (0.12) (0.04) (0.06) (0.12) (0.13) (0.05) (0.04) (0.15) Ln area −0.07 0.01 0.16 0.29 * 0.03 −0.01 −0.03 0.04 0.07 −0.00 0.02 0.07 (0.11) (0.11) (0.13) (0.15) (0.08) (0.03) (0.05) (0.08) (0.09) (0.03) (0.03) (0.11) Dist. equator 0.05 *** 0.04 *** 0.04 *** 0.04 *** 0.05 *** −0.00 0.01 ** 0.05 *** 0.05 *** −0.00 0.00 0.05 *** (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.01) (0.01) (0.00) (0.00) (0.01) Constant 6.51 *** 6.27 *** 5.54 *** 4.59 *** −3.92 *** −0.02 −0.90 *** −3.43 *** −4.60 *** 0.26 −0.99 *** −4.42 *** (0.34) (0.35) (0.60) (0.96) (0.35) (0.11) (0.12) (0.37) (0.47) (0.17) (0.18) (0.49) Observations 52 50 52 50 46 46 49 45 45 45 47 44 Colonial/geo controls No Yes No Yes No No No No Yes Yes Yes Yes K-P F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SW F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SY 10% max IV size 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 SY 25% max IV size 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) lnY/L lnY/L lnY/L lnY/L lnY/L α1−αlnK/Y lnH/L lnA lnY/L α1−αlnK/Y lnH/L lnA Trade 2.28 *** 0.01 0.78 ** 2.04 *** 3.54 ** −0.37 0.58 4.00 *** (0.73) (0.30) (0.36) (0.72) (1.52) (0.48) (0.63) (1.54) Exports 6.18 *** 7.11 *** (1.69) (1.96) Imports 7.94 *** 9.38 ** (2.80) (4.54) Ln pop. −0.19 −0.28 * −0.33 * −0.32 * −0.22 * −0.05 −0.00 −0.16 −0.26 ** −0.08 * −0.05 −0.16 (0.16) (0.15) (0.17) (0.17) (0.12) (0.04) (0.06) (0.12) (0.13) (0.05) (0.04) (0.15) Ln area −0.07 0.01 0.16 0.29 * 0.03 −0.01 −0.03 0.04 0.07 −0.00 0.02 0.07 (0.11) (0.11) (0.13) (0.15) (0.08) (0.03) (0.05) (0.08) (0.09) (0.03) (0.03) (0.11) Dist. equator 0.05 *** 0.04 *** 0.04 *** 0.04 *** 0.05 *** −0.00 0.01 ** 0.05 *** 0.05 *** −0.00 0.00 0.05 *** (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.01) (0.01) (0.00) (0.00) (0.01) Constant 6.51 *** 6.27 *** 5.54 *** 4.59 *** −3.92 *** −0.02 −0.90 *** −3.43 *** −4.60 *** 0.26 −0.99 *** −4.42 *** (0.34) (0.35) (0.60) (0.96) (0.35) (0.11) (0.12) (0.37) (0.47) (0.17) (0.18) (0.49) Observations 52 50 52 50 46 46 49 45 45 45 47 44 Colonial/geo controls No Yes No Yes No No No No Yes Yes Yes Yes K-P F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SW F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SY 10% max IV size 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 SY 25% max IV size 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 Notes: In columns (5–12), the explained variables are normalised by the value of the US. The predicted values of trade, exports and imports are those based on the non-linear estimation. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 8: The Channels of Growth-enhancing Impact of Trade with Developed Countries (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) lnY/L lnY/L lnY/L lnY/L lnY/L α1−αlnK/Y lnH/L lnA lnY/L α1−αlnK/Y lnH/L lnA Trade 2.28 *** 0.01 0.78 ** 2.04 *** 3.54 ** −0.37 0.58 4.00 *** (0.73) (0.30) (0.36) (0.72) (1.52) (0.48) (0.63) (1.54) Exports 6.18 *** 7.11 *** (1.69) (1.96) Imports 7.94 *** 9.38 ** (2.80) (4.54) Ln pop. −0.19 −0.28 * −0.33 * −0.32 * −0.22 * −0.05 −0.00 −0.16 −0.26 ** −0.08 * −0.05 −0.16 (0.16) (0.15) (0.17) (0.17) (0.12) (0.04) (0.06) (0.12) (0.13) (0.05) (0.04) (0.15) Ln area −0.07 0.01 0.16 0.29 * 0.03 −0.01 −0.03 0.04 0.07 −0.00 0.02 0.07 (0.11) (0.11) (0.13) (0.15) (0.08) (0.03) (0.05) (0.08) (0.09) (0.03) (0.03) (0.11) Dist. equator 0.05 *** 0.04 *** 0.04 *** 0.04 *** 0.05 *** −0.00 0.01 ** 0.05 *** 0.05 *** −0.00 0.00 0.05 *** (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.01) (0.01) (0.00) (0.00) (0.01) Constant 6.51 *** 6.27 *** 5.54 *** 4.59 *** −3.92 *** −0.02 −0.90 *** −3.43 *** −4.60 *** 0.26 −0.99 *** −4.42 *** (0.34) (0.35) (0.60) (0.96) (0.35) (0.11) (0.12) (0.37) (0.47) (0.17) (0.18) (0.49) Observations 52 50 52 50 46 46 49 45 45 45 47 44 Colonial/geo controls No Yes No Yes No No No No Yes Yes Yes Yes K-P F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SW F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SY 10% max IV size 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 SY 25% max IV size 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) lnY/L lnY/L lnY/L lnY/L lnY/L α1−αlnK/Y lnH/L lnA lnY/L α1−αlnK/Y lnH/L lnA Trade 2.28 *** 0.01 0.78 ** 2.04 *** 3.54 ** −0.37 0.58 4.00 *** (0.73) (0.30) (0.36) (0.72) (1.52) (0.48) (0.63) (1.54) Exports 6.18 *** 7.11 *** (1.69) (1.96) Imports 7.94 *** 9.38 ** (2.80) (4.54) Ln pop. −0.19 −0.28 * −0.33 * −0.32 * −0.22 * −0.05 −0.00 −0.16 −0.26 ** −0.08 * −0.05 −0.16 (0.16) (0.15) (0.17) (0.17) (0.12) (0.04) (0.06) (0.12) (0.13) (0.05) (0.04) (0.15) Ln area −0.07 0.01 0.16 0.29 * 0.03 −0.01 −0.03 0.04 0.07 −0.00 0.02 0.07 (0.11) (0.11) (0.13) (0.15) (0.08) (0.03) (0.05) (0.08) (0.09) (0.03) (0.03) (0.11) Dist. equator 0.05 *** 0.04 *** 0.04 *** 0.04 *** 0.05 *** −0.00 0.01 ** 0.05 *** 0.05 *** −0.00 0.00 0.05 *** (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.01) (0.01) (0.00) (0.00) (0.01) Constant 6.51 *** 6.27 *** 5.54 *** 4.59 *** −3.92 *** −0.02 −0.90 *** −3.43 *** −4.60 *** 0.26 −0.99 *** −4.42 *** (0.34) (0.35) (0.60) (0.96) (0.35) (0.11) (0.12) (0.37) (0.47) (0.17) (0.18) (0.49) Observations 52 50 52 50 46 46 49 45 45 45 47 44 Colonial/geo controls No Yes No Yes No No No No Yes Yes Yes Yes K-P F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SW F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SY 10% max IV size 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 SY 25% max IV size 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 Notes: In columns (5–12), the explained variables are normalised by the value of the US. The predicted values of trade, exports and imports are those based on the non-linear estimation. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large 5.1.2 Output decomposition To reinforce the technology transfer hypothesis, we propose a decomposition of income based on a simple Cobb-Douglas production function in the same spirit as and Jones (1999), Frankel . Specifically, we consider the following function: Yi=Kiα(HiAi)1−α, (9) where Yi stands for output in country i, Ki is the stock of physical capital, Hi denotes the human capital, Ai is a productivity term, and α represents the labour share in income. Rewriting this production function in terms of output per worker and using the logarithmic transformation, we get: lnyi=α(1−α)lnKiYi+lnhi+ln(Ai), (10) where yi=Yi/Li denotes output per worker, and hi=Hi/Li is human capital per worker. Data on stock of physical capital, workers and output are taken from the Penn World Tables. Regarding the level of productivity, it is calculated from the production function assuming that α=1/3 in line with standard neoclassical approach.21 Moreover, following Hall and Jones (1999), we consider human capital as a function of returns to schooling ( Hi=eϕ(Si)Li) as estimated in a Mincerian wage regression. ϕ(S) is a function reflecting the efficiency of a unit of labour with S years of schooling. Data on the number of years of schooling are taken from UNDP HDR statistics that rely on several sources: UNESCO Institute for Statistics, Barro and Lee (2013), ICF Macro Demographic and Health Surveys and UNICEF’s Multple Indicator Cluster. Each of the components on the right-hand side of Equation (10) contributes to the improvement of income per worker. The results of the 2SLS estimation of Equation (10) considering successively each term (respectively the log of income per worker, the log of the capital-output ratio, the log of human capital per worker, and the log of productivity) as the dependent variable are reported in Table 8. In columns (5–8), our results are not controlled by geographical and historical (or colonial) variables, while it is the case in columns (9–12). In line with our previous findings, an increase in trade between Africa and the developed economies significantly contributes to improve income per worker in Africa regardless of the specification. Considering the model without geographical and colonial controls, we found that trade between Africa and the industrialised economies positively affects income per worker in Africa mainly through an improvement in human capital per worker and higher productivity. Specifically, our estimates indicate that a rise of one percentage point (pp) in trade increases the contribution of the intensity of physical capital to income of 3/4 pp whereas it increases the contribution of productivity to income about two pp. These elasticities are close to those of Frankel and Romer (1999) with a more heterogeneous sample. When the complete specification is estimated (columns (9–12)), only productivity appears as the main transmission channel of the positive effect of trade on the standards of living, with an elasticity twice as high as before. Indeed, controlling for geographic and colonial regressors, we found that a one pp increase in trade between Africa and industrialised countries leads to an increase of the contribution of productivity to income per worker of four pp. This relatively high elasticity reflects to some extent the low initial level of productivity in African countries. Furthermore, in all configurations in which trade has a significant influence on income per worker or its components, the distance from equator significantly explains the differences in levels of income per worker between African countries. Moreover, our findings show that trade has not significantly helped to improve the contribution of capital intensity per worker to income. To sum up, if international trade with developed countries increases income in Africa, productivity is clearly the main transmission channel. Figure A1 in Appendix A shows the contents of Africa’s imports in skills and technology intensity. Imports from the developed OECD countries are clearly more equipped in skills and technology intensive, contributing to the improvement of productivity in Africa. Indeed, if the content in low skill of Africa’s imports is proportionally the same regardless of the partner, imports from the developed OECD countries are approximately twice as equipped with middle-skill and high-skill than those from Africa or other developing countries. These facts clearly support our empirical results. The latter also found several anchorage points in the literature even if the specific case of Africa has not been studied before. Indeed, the decisive role of trade openness in improving total factor productivity has been empirically emphasised by Edwards (1998). Miller and Upadhyay (2000) confirmed this result by showing that over and above the positive impact of trade openness on total factor productivity, outward-oriented countries also experience an improvement in total factor productivity. 5.2 Remittances and institutions While we do not find any evidence of growth-enhancing impact of openness to migration in Africa, it is possible that a potential benefice of emigration to developed countries is compensated by the adverse impact of brain drain. According to the literature on international migration, the positive effects of emigration operate through remittances (Giuliano and Ruiz-Arranz, 2009) and by improving political institutions (Spilimbergo, 2009; Docquier et al., 2016). We investigate here the relevance of these two channels by focusing on openness with developed countries, since (i) there is evidence of growth-enhancing effect of trade only with advanced economies, and (ii) most part of remittances received in Africa originate from developed countries. To examine whether remittances from African emigrants in developed countries improve economic growth, we regress output per capita (in log) on total remittances (in % of GDP),22 considering the same controls as above. The corresponding estimation results for the 2000 reference year are reported in column (1) of Table 9. As shown, we confirm the growth-enhancing impact of trade with developed countries, while remittances do not have a significant effect. The non-significance of remittances reflects the ambiguity about the link between remittances and growth. Indeed, on the one hand, remittances are perceived to overcome credit constraints and imperfections in financial markets, leading to higher rates of capital accumulation. On the other hand, if remittances are pure transfers aiming to sustain consumption of family members left behind, they may be detrimental for economic growth due to a moral hazard problem (Chami et al., 2003). Moreover, using remittances for final consumption may cause a Dutch-disease effect (Amuedo-Dorantes and Pozo, 2004). Our findings confirm those of Ahamada and Coulibaly (2013) showing that remittances do not promote growth in sub-Sahara African countries because they are not generally used for investment purposes. Our results are also in line with Bettin and Zazzaro (2012) who propose a new measure of efficiency of the domestic banking system and show that remittances promote growth only in countries whose banks are efficient, which is not the case in Africa. Our findings corroborates those of Konte (2017) as well who aims at understanding the difficulty to find an enhancing effect of remittances on growth despite (i) the important amount of those transfers in several developing countries, and (ii) the literature which has evidenced their positive impact on poverty and inequality. Relying on a recent finite-mixture-of-regressions approach, the author shows that the impact of remittances differs depending on the growth regime to which the considered country belongs. Table 9: Remittances and Institutions (1) (2) (3) (4) lnY/L Polity 2 Political rights Civil liberties Trade 4.20 ** −0.75 −0.42 −0.57 (1.99) (0.57) (0.62) (0.63) Emigration 2.21 4.39 * 3.56 * (2.23) (2.32) (2.09) Remittances 0.01 (0.26) Ln pop. −0.29 * −0.04 −0.01 −0.01 (0.17) (0.03) (0.03) (0.02) Ln area 0.15 (0.19) Dist. equator 0.04 −0.01 ** −0.00 −0.00 (0.03) (0.00) (0.00) (0.00) Constant 5.41 ** 0.76 *** 0.40 ** 0.44 *** (2.28) (0.16) (0.17) (0.16) Observations 48 49 50 50 Colonial/geo controls Yes Yes Yes Yes K-P F-stat 3.646 5.220 5.274 5.274 SW F-stat for Trade 5.501 10.80 10.74 10.74 SW F-stat for Emig. 8.698 8.398 8.398 SW F-stat for Rem. 6.827 SY 10% max IV size 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 (1) (2) (3) (4) lnY/L Polity 2 Political rights Civil liberties Trade 4.20 ** −0.75 −0.42 −0.57 (1.99) (0.57) (0.62) (0.63) Emigration 2.21 4.39 * 3.56 * (2.23) (2.32) (2.09) Remittances 0.01 (0.26) Ln pop. −0.29 * −0.04 −0.01 −0.01 (0.17) (0.03) (0.03) (0.02) Ln area 0.15 (0.19) Dist. equator 0.04 −0.01 ** −0.00 −0.00 (0.03) (0.00) (0.00) (0.00) Constant 5.41 ** 0.76 *** 0.40 ** 0.44 *** (2.28) (0.16) (0.17) (0.16) Observations 48 49 50 50 Colonial/geo controls Yes Yes Yes Yes K-P F-stat 3.646 5.220 5.274 5.274 SW F-stat for Trade 5.501 10.80 10.74 10.74 SW F-stat for Emig. 8.698 8.398 8.398 SW F-stat for Rem. 6.827 SY 10% max IV size 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 Notes: The predicted trade and emigration values are those based on the non-linear estimation. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 25% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 9: Remittances and Institutions (1) (2) (3) (4) lnY/L Polity 2 Political rights Civil liberties Trade 4.20 ** −0.75 −0.42 −0.57 (1.99) (0.57) (0.62) (0.63) Emigration 2.21 4.39 * 3.56 * (2.23) (2.32) (2.09) Remittances 0.01 (0.26) Ln pop. −0.29 * −0.04 −0.01 −0.01 (0.17) (0.03) (0.03) (0.02) Ln area 0.15 (0.19) Dist. equator 0.04 −0.01 ** −0.00 −0.00 (0.03) (0.00) (0.00) (0.00) Constant 5.41 ** 0.76 *** 0.40 ** 0.44 *** (2.28) (0.16) (0.17) (0.16) Observations 48 49 50 50 Colonial/geo controls Yes Yes Yes Yes K-P F-stat 3.646 5.220 5.274 5.274 SW F-stat for Trade 5.501 10.80 10.74 10.74 SW F-stat for Emig. 8.698 8.398 8.398 SW F-stat for Rem. 6.827 SY 10% max IV size 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 (1) (2) (3) (4) lnY/L Polity 2 Political rights Civil liberties Trade 4.20 ** −0.75 −0.42 −0.57 (1.99) (0.57) (0.62) (0.63) Emigration 2.21 4.39 * 3.56 * (2.23) (2.32) (2.09) Remittances 0.01 (0.26) Ln pop. −0.29 * −0.04 −0.01 −0.01 (0.17) (0.03) (0.03) (0.02) Ln area 0.15 (0.19) Dist. equator 0.04 −0.01 ** −0.00 −0.00 (0.03) (0.00) (0.00) (0.00) Constant 5.41 ** 0.76 *** 0.40 ** 0.44 *** (2.28) (0.16) (0.17) (0.16) Observations 48 49 50 50 Colonial/geo controls Yes Yes Yes Yes K-P F-stat 3.646 5.220 5.274 5.274 SW F-stat for Trade 5.501 10.80 10.74 10.74 SW F-stat for Emig. 8.698 8.398 8.398 SW F-stat for Rem. 6.827 SY 10% max IV size 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 Notes: The predicted trade and emigration values are those based on the non-linear estimation. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 25% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Finally, we investigate whether emigration to developed countries helps to promote institutional quality as notably highlighted by Spilimbergo (2009) and Docquier et al. (2016) who put forward the role of transfer of norms. To this end, we regress institutional quality on openness to trade with developed countries and emigration to advanced countries, using the same controls as above which are also perceived as influencing political institutions (Acemoglu et al., 2001). To proxy institutional quality, we rely on (i) the Polity 2 (P2) index drawn from the Polity IV database, and (ii) Political Rights (PR) and Civil Liberties (CL) indexes of Freedom House. P2 is a composite index based on the several concepts of democracy including (i) the presence of institutions and procedures through which citizens can express effective preferences about alternative policies and leaders, (ii) the existence of institutionalised constraints on the exercise of power by the executive, (iii) the guarantee of civil liberties to all citizens in their daily lives and in acts of political participation. P2 ranges from −10 (complete autocracy) to +10 (complete democracy). The PR index is based on a checklist of 10 political rights’ questions that are classified into three subcategories: electoral processes (3 questions), political pluralism and participation (4 questions), and functioning of the government (3 questions). The CL index is based on 15 civil liberties’ questions that are classified into four subcategories: freedom of expression and belief (4 questions), associational and organisational rights (3 questions), rule of law (4 questions), and personal autonomy and individual rights (4 questions). Both PR and CL checklists are translated into a rating ranging from 1 (most free) to 7 (least free). To make our results fully comparable across different measures of institutions—as in Acemoglu et al. (2008)—we normalise each measure on a scale ranging from 0 (complete autocratic/least free) to 1 (complete democratic/most free). The results for the year 2000 reported in Table 9 (columns (2–4)) show a positive influence of emigration to developed countries on political institutions that is significant only at the 10% level for the PR and CL indexes of Freedom House. This quite weak significance can reflect the fact that emigration, by lowering incentives, may delay democratic reforms (Docquier et al., 2016). 6. Conclusion This paper contributes to the debate on the growth-enhancing impact of openness by focusing on Africa. Examining the nature of the relationship both in a South-South and North-South perspective, we jointly consider openness to trade and to migration as vehicles of globalisation; both types of openness being instrumented by gravity-based predictors. Without distinguishing between partners, we show that international openness (both to trade and to migration) has a mitigated impact on per capita income in Africa; the positive effect of trade and migration being sensitive to the inclusion of control variables and the method used (linear or non-linear) to predict the geographic component of openness. In contrast, discriminating according to the partner allows us to obtain interesting and robust findings. Specifically, we establish that trade between Africa and the industrialised countries has a clear and significant positive impact on the standards of living in Africa, and show that such fostering effect operates through an improvement in total factor productivity. This growth-enhancing effect is consistent with the new trade theory suggesting that countries benefit from the advances in technological progress of their trading partners. Conversely, South-South openness fails to improve income in Africa regardless of the configuration (intra-African openness or openness between Africa and other developing countries). Turning to openness to migration, we do not find evidence of any growth-enhancing effect in Africa, whatever the partner country. South-South trade fails to improve the standards of living in Africa, probably because of the similarity of traded products and production techniques, which leave little room for learning between countries and technology transfer. Similarly, homogeneity of skills and institutions is likely to explain the failure of intra-African immigration to improve per capita income. Moreover, the lack of evidence of clear-cut positive effect of emigration from Africa to developed countries may result from two opposite effects at play in the African context: the negative effect of the ‘brain drain’ and the positive impact of the transfer of knowledge. On the whole, while our findings do not attribute a key role to openness to migration, they show that trade increases income in Africa, provided that the trade partner is an advanced country. In other words, our results underline that the characteristics of the trading partner—i.e., whether it is a developed or developing country—strongly matters in the trade-growth relationship. These findings highlight the importance of trade flows and trade-promoting policies between Africa and industrialised economies to foster African growth in a North-South perspective. Acknowledgements We would like to thank Christophe Destais, Sébastien Jean and participants at the EconomiX-CNRS and BETA-CNRS seminars for their remarks. We are also grateful to an anonymous referee and the Editor for very useful suggestions that helped us improve the paper. Footnotes 1 For a survey, see Edwards (1995) and Rodrik (1995) among others. 2 More precisely, openness to trade and openness to migration are jointly introduced in the income equation, being instrumented by the same geographical factors. 3 Recall that this framework is based on the gravity model of trade in which countries’ geographic characteristics are used to obtain instrumental variables estimates of trade effect on income. 4 See among others Edwards (1998) and Miller and Upadhyay (2000). See also our analysis in Section 5. 5 These transmission channels will be investigated in Section 5. 6 It would be interesting to reevaluate this effect in a few years (especially for China), when more—recent—observations will be available to better capture a potential medium to long-term growth-enhancing impact. 7 See the influential papers of Hall and Jones (1999), Acemoglu et al. (2001) and Rodrik et al. (2004). 8 We follow standard practice in measuring trade as imports plus exports divided by GDP (trade intensity ratio). An alternative would have been to examine import and export shares separately but, as shown by Frankel and Romer (1999) among others, the geographic components of the two shares are—not surprisingly—almost identical. Thus, the effects of imports and exports cannot be separated in the sense that the trade intensity ratio does not allow to distinguish countries having similar GDP and characteristics, but different trade-flows composition. However, in combining total trade (imports plus exports) and domestic economy, this ratio provides information regarding country’s involvement in trade—i.e., how important is trade for the considered country—which is the key information for our purpose (for a detailed discussion on trade indicators, see Ng, 2002). Nevertheless, for the sake of completeness and in order to disentangle the two channels through which trade with developed countries may promote growth in Africa—economies of scale and comparative advantages in the one hand, and transfer of technologies in the other hand—we will also consider the two ratios (exports over GDP and imports over GDP) separately (see Section 5). 9 See e.g. Caballero et al. (2008) and Mendoza et al. (2009). 10 The 52 countries are: Angola, Benin, Botswana, Burkina Faso, Burundi, Cabo Verde, Cameroon, Central African Republic, Chad, Comoros, Congo Dem. Rep., Congo Rep., Cote d’Ivoire, Djibouti, Egypt, Equatorial Guinea, Eritrea, Ethiopia, Gabon, Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Libya, Madagascar, Malawi, Mali, Mauritania, Mauritius, Morocco, Mozambique, Namibia, Niger, Nigeria, Rwanda, Sao Tome and Principe, Senegal, Seychelles, Sierra Leone, Somalia, South Africa, South Sudan, Sudan, Swaziland, Tanzania, Togo, Tunisia, Uganda, Zambia, and Zimbabwe. 11 Trade share for each country is calculated as the sum of its observed bilateral trade divided by GDP. Compared to the use of aggregate data, doing so allows us to specifically identity the partner country and, consequently, to account for potential different effects of openness depending on the type of the partner. 12 Frankel and Romer (1999) also emphasise the problem of accuracy in the estimation of this coefficient. 13 To save space, Figure 4 considers only gravity-predicted openness based on non-linear (PPML) estimation. 14 Note that this test constitutes a modification and improvement of the procedure described by Angrist and Pischke (2009). 15 For the sake of completeness and to disentangle immigration from and emigration to different types of countries separately, we have also regressed per capita income of African country i on emigration from this country i to other African countries and developing countries. The results—available upon request to the authors—corroborate those presented in Table 6 as we found that such type of emigration does not robustly impact growth. The case of immigration from advanced countries has not been considered, as such type of migration is very limited and, therefore, negligible. 16 This will be investigated further in Section 5. 17 See Section 5. 18 The six following countries are excluded from the African sample: Algeria, Djibouti, Egypt, Libya, Morocco and Tunisia. 19 To save space, we do not report the corresponding gravity estimates but they are available upon request to the authors. 20 The gravity-predicted values of exports and imports are considered as instruments in the corresponding regressions (see Table A1 in Appendix A). 21 The empirical value of this parameter we get for African countries is 0.314, which is very close to the reference value. 22 Data on bilateral remittances are not available. Remittances are taken from WDI and are defined as the sum of personal transfers and compensation of employees. The predicted values of trade with developed countries and emigration to developed countries are used as instruments. References Acemoglu D. , Johnson S. , Robinson J. ( 2001 ) ‘ The Colonial Origins of Comparative Development: An Empirical Investigation ’, American Economic Review , 91 : 1369 – 1401 . Google Scholar CrossRef Search ADS Acemoglu D. , Johnson S. , Robinson J. A. , Yared P. ( 2008 ) ‘ Income and Democracy ’, American Economic Review , 98 : 808 – 42 . Google Scholar CrossRef Search ADS Ahamada I. , Coulibaly D. ( 2013 ) ‘ Remittances and Growth in Sub-saharan African Countries: Evidence from a Panel Causality Test ’, Journal of International Development , 25 : 310 – 24 . Google Scholar CrossRef Search ADS Alesina A. , Spolaore E. , Wacziarg R. ( 2000 ) ‘ Economic Integration and Political Disintegration ’, American Economic Review , 90 : 1276 – 96 . Google Scholar CrossRef Search ADS Amuedo-Dorantes C. , Pozo S. ( 2004 ) ‘ Workers’ Remittances and the Real Exchange Rate: A Paradox of Gifts ’, World Development , 32 : 1407 – 17 . Google Scholar CrossRef Search ADS Angrist J. D. , Pischke J.-S. ( 2009 ) Mostly Harmless Econometrics: An Empiricist’s Companion . Princeton : Princeton University Press . Barro R. , Lee J. W. ( 2013 ) ‘ A New Data Set of Educational Attainment Dataset in the World, 1950–2010 ’, Journal of Development Economics , 104 : 184 – 198 . Google Scholar CrossRef Search ADS Beine M. , Docquier F. , Rapoport H. ( 2008 ) ‘ Brain Drain and Human Capital Formation in Developing Countries: Winners and Losers ’, Economic Journal , 118 : 631 – 52 . Google Scholar CrossRef Search ADS Bettin G. , Zazzaro A. ( 2012 ) ‘ Remittances and Financial Development: Substitutes or Complements in Economic Growth? ’, Bulletin of Economic Research , 64 ( 4 ): 509 – 36 . Google Scholar CrossRef Search ADS Caballero R. J. , Fahri E. , Gourinchas P. O. ( 2008 ) ‘ An Equilibrium Model of Global Imbalances and Low Interest Rates ’, American Economic Review , 98 ( 1 ): 358 – 93 . Google Scholar CrossRef Search ADS Chami R. , Fullenkampn C. , Jahjah S. 2003 . Are immigrant remittance flows a source of capital for development? IMF Working Paper 189. Docquier F. , Lodigiani E. , Rapoport H. , Schif M. ( 2016 ) ‘ Emigration and Democracy ’, Journal of Development Economics , 120 : 209 – 23 . forthcoming. Google Scholar CrossRef Search ADS Docquier F. , Özden Ç. , Peri G. , 2010 . The wage effects of immigration and emigration. Working Paper 16646. National Bureau of Economic Research. Dollar D. , Kraay A. ( 2003 ) ‘ Institutions, Trade, and Growth ’, Journal of Monetary Economics , 50 : 133 – 62 . Google Scholar CrossRef Search ADS Easterly W. , Levine R. ( 2001 ) ‘ It’s Not Factor Accumulation: Stylized Facts and Growth Models ’, World Bank Economic Review , 15 ( 2 ): 177 – 219 . Google Scholar CrossRef Search ADS Edwards S. ( 1995 ) ‘Trade policy, exchange rates, and growth’, in Dornbusch R. , Edwards S. (eds) , Reform, Recovery, and Growth: Latin America and the Middle East . Chicago : University of Chicago Press , pp. 13 – 49 . Edwards S. ( 1998 ) ‘ Openness, Productivity and Growth: What Do We Really Know? ’, Economic Journal , 108 : 383 – 98 . Google Scholar CrossRef Search ADS Frankel J. A. , Romer D. ( 1999 ) ‘ Does Trade Cause Growth? ’, American Economic Review , 89 : 379 – 99 . Google Scholar CrossRef Search ADS Frankel J. A. , Rose A. ( 2002 ) ‘ An Estimate of the Effect of Common Currencies on Trade and Income ’, Quarterly Journal of Economics , 117 : 437 – 66 . Google Scholar CrossRef Search ADS Freund C. , Bolaky B. ( 2008 ) ‘ Trade, Regulations, and Income ’, Journal of Development Economics , 87 : 309 – 21 . Google Scholar CrossRef Search ADS Gallup J. L. , Sachs J. D. , Mellinger A. D. ( 1999 ) ‘ Geography and Economic Development ’, International Regional Science Review , 22 : 179 – 232 . Google Scholar CrossRef Search ADS Giuliano P. , Ruiz-Arranz M. ( 2009 ) ‘ Remittances, Financial Development and Growth ’, Journal of Development Economics , 90 : 144 – 52 . Google Scholar CrossRef Search ADS Grossman G. , Helpman E. ( 1991 a) Innovation and Growth in the Global Economy . Cambridge, MA : MIT Press . Grossman G. , Helpman E. ( 1991 b) ‘ Trade, Knowledge Spillovers, and Growth ’, European Economic Review , XXXV : 517 – 26 . Google Scholar CrossRef Search ADS Hall R. E. , Jones C. I. ( 1999 ) ‘ Why Do Some Countries Produce so much more Output Per Worker than Others? ’, Quarterly Journal of Economics , 114 : 83 – 116 . Google Scholar CrossRef Search ADS He Y. ( 2013 ) ‘ Does China’s Trade Expansion Help African Development?—An Empirical Estimation ’, China Economic Review , 26 : 28 – 38 . Google Scholar CrossRef Search ADS Head K. , Mayer T. , Ries J. ( 2009 ) ‘ How Remote is the Offshoring Threat? ’, European Economic Review , 53 : 429 – 44 . Google Scholar CrossRef Search ADS Head K. , Mayer T. , Ries J. ( 2010 ) ‘ The Erosion of Colonial Trade Linkages After Independence ’, Journal of International Economics , 81 : 1 – 14 . Google Scholar CrossRef Search ADS Helpman E. , Krugman P. ( 1985 ) Market Structure and Foreign Trade . Cambridge, MA : MIT Press . Irwin D. A. , Terviö M. ( 2002 ) ‘ Does Trade Raise Income?: Evidence from the Twentieth Century ’, Journal of International Economics , 58 : 1 – 18 . Google Scholar CrossRef Search ADS Kaplinsky R. ( 2013 ) ‘ What Contribution Can China Make to Inclusive Growth in Sub-Saharan Africa? ’, Development and Change , 44 : 1295 – 1316 . Google Scholar CrossRef Search ADS Kleibergen F. , Paap R. ( 2006 ) ‘ Generalized Reduced Rank Tests Using the Singular Value Decomposition ’, Journal of Econometrics , 133 : 97 – 126 . Google Scholar CrossRef Search ADS Konte M. ( 2017 ) ‘ Do Remittances Not Promote Growth? A Finite Mixture-of-Regressions Approach ’, Empirical Economics , 1 – 36 . https://doi.org/10.1007/s00181-016-1224-z. La Porta R. , Lopez-de-Silanes F. , Shleifer A. ( 2008 ) ‘ The Economic Consequences of Legal Origins ’, Journal of Economic Literature , 46 : 285 – 332 . Google Scholar CrossRef Search ADS La Porta R. , Lopez-de-Silanes F. , Shleifer A. , Vishny R. ( 1999 ) ‘ The Quality of Government ’, Journal of Law, Economics, & Organization , 15 : 222 – 79 . Google Scholar CrossRef Search ADS Lyons M. , Brown A. ( 2010 ) ‘ Perception of Boom, Bust and the China-Africa Trade in Lomé and Bamako ’, World Development , 38 : 771 – 82 . Google Scholar CrossRef Search ADS Mendoza E. , Quadrini V. , Ríos-Rull J.-V. ( 2009 ) ‘ Financial Integration, Financial Development, and Global Imbalances ’, Journal of Political Economy , 117 ( 3 ): 371 – 416 . Google Scholar CrossRef Search ADS Miller S. M. , Upadhyay M. P. ( 2000 ) ‘ The Effects of Openness, Trade Orientation, and Human Capital on Total Factor Productivity ’, Journal of Development Economics , 50 : 133 – 62 . Ng F. ( 2002 ) ‘Trade indicators and indices’, in Hoekman B. , Mattoo A. , English P. (eds) , Development, Trade and the WTO, A Handbook . Washington, DC : The International Bank for Reconstruction and Development and World Bank , pp. 585 – 88 . Noguer M. , Siscart M. ( 2005 ) ‘ Trade Raises Income: A Precise and Robust Result ’, Journal of International Economics , 65 : 447 – 60 . Google Scholar CrossRef Search ADS Ortega F. , Peri G. ( 2014 ) ‘ Openness and Income: The Roles of Trade and Migration ’, Journal of International Economics , 92 : 213 – 51 . Google Scholar CrossRef Search ADS Rodriguez F. , Rodrik D. ( 2000 ) ‘ Trade Policy and Economic Growth: A Skeptic’s Guide to the Cross-national Evidence ’, NBER Macroeconomics Annual , 15 : 261 – 338 . Google Scholar CrossRef Search ADS Rodrik D. ( 1995 ) ‘Trade and industrial policy reform’, in Behrman J. , Srinivasan T. N. (eds) , Handbook of Development Economics , Vol. 3B . Amsterdam : North-Holland , pp. 2925 – 82 . Rodrik D. , Subramanian A. , Trebbi F. ( 2004 ) ‘ Institutions Rule: The Primacy of Institutions Over Geography and Integration in Economic Development ’, Journal of Economic Growth , 9 : 131 – 65 . Google Scholar CrossRef Search ADS Rose A. ( 2000 ) ‘ One Money, One Market? The Effect on Common Currencies on International Trade ’, Economic Policy , XXX : 7 – 45 . Rose A. ( 2001 ) ‘ Currency Unions and Trade: The Effect is Large ’, Economic Policy , XXXI : 433 – 62 . Sachs J. D. , Warner A. M. ( 1997 ) ‘ Fundamental Sources of Long-run Growth ’, American Economic Review , 87 ( 2 ): 184 – 8 . Sanderson E. , Windmeijer F. ( 2015 ) ‘ A Weak Instrument F-test in Linear IV Models with Multiple Endogenous Variables ’, Journal of Econometrics , 190 ( 2 ): 212 – 21 . Google Scholar CrossRef Search ADS Silva J. M. C. S. , Tenreyro S. ( 2006 ) ‘ The Log of Gravity ’, Review of Economics and Statistics , 88 : 641 – 58 . Google Scholar CrossRef Search ADS Silva J. M. C. S. , Tenreyro S. ( 2010 ) ‘ On the Existence of the Maximum Likelihood Estimates in Poisson Regression ’, Economic Letters , 107 : 310 – 2 . Google Scholar CrossRef Search ADS Spilimbergo A. ( 2009 ) ‘ Foreign Students and Democracy ’, American Economic Review , 99 : 528 – 43 . Google Scholar CrossRef Search ADS Stark O. , Helmenstein C. , Prskawetz A. ( 1997 ) ‘ A Brain Gain with a Brain Drain ’, Economics Letters , 55 : 227 – 34 . Google Scholar CrossRef Search ADS Stock J. H. , Yogo M. ( 2005 ) ‘Testing for weak instruments in linear IV Regression’, in Stock J. H. , Andrews D. W. K. (eds) , Identification and Inference for Econometric Models: Essays in Honor of Thomas J. Rothenberg . Cambridge : Cambridge University Press , pp. 80 – 108 . Chapter 5. Google Scholar CrossRef Search ADS Yanikkaya H. ( 2003 ) ‘ Trade Openness and Economic Growth: A Cross-country Empirical Investigation ’, Journal of Development Economics , 72 : 57 – 89 . Google Scholar CrossRef Search ADS Figure A1: View largeDownload slide Composition of Africa’s Imports. Notes: Data are extracted from UNCTAD and correspond to mean values over the 1995–2014 period. Figure A1: View largeDownload slide Composition of Africa’s Imports. Notes: Data are extracted from UNCTAD and correspond to mean values over the 1995–2014 period. Table A1: Gravity Regression, African Openness with Industrialised Countries Exports Imports OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.90 *** −0.78 *** −0.69 *** −0.77 *** (0.18) (0.15) (0.15) (0.22) Ln pop. origin −0.05 −0.34 ** −0.20 *** −0.20 *** (0.06) (0.14) (0.07) (0.06) Ln pop. dest. 1.43 *** 1.11 *** 1.65 *** 1.12 *** (0.05) (0.10) (0.06) (0.08) Ln area origin −0.03 0.21 *** −0.01 −0.00 (0.04) (0.07) (0.05) (0.04) Ln area dest. −0.34 *** −0.18 −0.43 *** −0.40 *** (0.04) (0.13) (0.05) (0.06) Sum landlocked −1.09 *** −0.82 *** −1.00 *** −0.89 *** (0.12) (0.23) (0.14) (0.11) Common language 0.78 *** 0.74 ** 1.03 *** 0.25 (0.24) (0.38) (0.40) (0.34) Common off. lang. 0.08 −0.12 −0.23 0.38 (0.22) (0.31) (0.39) (0.32) Colonial ties 2.46 *** 1.37 *** 2.77 *** 1.46 *** (0.27) (0.37) (0.25) (0.21) Hegemony −2.08 *** −2.57 *** −3.05 *** −2.80 *** (0.34) (0.40) (0.41) (0.38) Time zone diff. −0.11 *** 0.05 * −0.12 *** 0.02 (0.03) (0.03) (0.04) (0.03) Constant −2.59 ** −2.30 −3.93 ** −0.22 (1.17) (1.93) (1.54) (1.45) Observations 1,892 2,300 1,936 2,300 R-squared 0.47 0.17 0.39 0.44 Exports Imports OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.90 *** −0.78 *** −0.69 *** −0.77 *** (0.18) (0.15) (0.15) (0.22) Ln pop. origin −0.05 −0.34 ** −0.20 *** −0.20 *** (0.06) (0.14) (0.07) (0.06) Ln pop. dest. 1.43 *** 1.11 *** 1.65 *** 1.12 *** (0.05) (0.10) (0.06) (0.08) Ln area origin −0.03 0.21 *** −0.01 −0.00 (0.04) (0.07) (0.05) (0.04) Ln area dest. −0.34 *** −0.18 −0.43 *** −0.40 *** (0.04) (0.13) (0.05) (0.06) Sum landlocked −1.09 *** −0.82 *** −1.00 *** −0.89 *** (0.12) (0.23) (0.14) (0.11) Common language 0.78 *** 0.74 ** 1.03 *** 0.25 (0.24) (0.38) (0.40) (0.34) Common off. lang. 0.08 −0.12 −0.23 0.38 (0.22) (0.31) (0.39) (0.32) Colonial ties 2.46 *** 1.37 *** 2.77 *** 1.46 *** (0.27) (0.37) (0.25) (0.21) Hegemony −2.08 *** −2.57 *** −3.05 *** −2.80 *** (0.34) (0.40) (0.41) (0.38) Time zone diff. −0.11 *** 0.05 * −0.12 *** 0.02 (0.03) (0.03) (0.04) (0.03) Constant −2.59 ** −2.30 −3.93 ** −0.22 (1.17) (1.93) (1.54) (1.45) Observations 1,892 2,300 1,936 2,300 R-squared 0.47 0.17 0.39 0.44 Notes: The dependent variables ‘Exports’ and ‘Imports’ stand, respectively, for the sum of bilateral exports divided by GDP and the sum of bilateral imports divided by GDP. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large Table A1: Gravity Regression, African Openness with Industrialised Countries Exports Imports OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.90 *** −0.78 *** −0.69 *** −0.77 *** (0.18) (0.15) (0.15) (0.22) Ln pop. origin −0.05 −0.34 ** −0.20 *** −0.20 *** (0.06) (0.14) (0.07) (0.06) Ln pop. dest. 1.43 *** 1.11 *** 1.65 *** 1.12 *** (0.05) (0.10) (0.06) (0.08) Ln area origin −0.03 0.21 *** −0.01 −0.00 (0.04) (0.07) (0.05) (0.04) Ln area dest. −0.34 *** −0.18 −0.43 *** −0.40 *** (0.04) (0.13) (0.05) (0.06) Sum landlocked −1.09 *** −0.82 *** −1.00 *** −0.89 *** (0.12) (0.23) (0.14) (0.11) Common language 0.78 *** 0.74 ** 1.03 *** 0.25 (0.24) (0.38) (0.40) (0.34) Common off. lang. 0.08 −0.12 −0.23 0.38 (0.22) (0.31) (0.39) (0.32) Colonial ties 2.46 *** 1.37 *** 2.77 *** 1.46 *** (0.27) (0.37) (0.25) (0.21) Hegemony −2.08 *** −2.57 *** −3.05 *** −2.80 *** (0.34) (0.40) (0.41) (0.38) Time zone diff. −0.11 *** 0.05 * −0.12 *** 0.02 (0.03) (0.03) (0.04) (0.03) Constant −2.59 ** −2.30 −3.93 ** −0.22 (1.17) (1.93) (1.54) (1.45) Observations 1,892 2,300 1,936 2,300 R-squared 0.47 0.17 0.39 0.44 Exports Imports OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.90 *** −0.78 *** −0.69 *** −0.77 *** (0.18) (0.15) (0.15) (0.22) Ln pop. origin −0.05 −0.34 ** −0.20 *** −0.20 *** (0.06) (0.14) (0.07) (0.06) Ln pop. dest. 1.43 *** 1.11 *** 1.65 *** 1.12 *** (0.05) (0.10) (0.06) (0.08) Ln area origin −0.03 0.21 *** −0.01 −0.00 (0.04) (0.07) (0.05) (0.04) Ln area dest. −0.34 *** −0.18 −0.43 *** −0.40 *** (0.04) (0.13) (0.05) (0.06) Sum landlocked −1.09 *** −0.82 *** −1.00 *** −0.89 *** (0.12) (0.23) (0.14) (0.11) Common language 0.78 *** 0.74 ** 1.03 *** 0.25 (0.24) (0.38) (0.40) (0.34) Common off. lang. 0.08 −0.12 −0.23 0.38 (0.22) (0.31) (0.39) (0.32) Colonial ties 2.46 *** 1.37 *** 2.77 *** 1.46 *** (0.27) (0.37) (0.25) (0.21) Hegemony −2.08 *** −2.57 *** −3.05 *** −2.80 *** (0.34) (0.40) (0.41) (0.38) Time zone diff. −0.11 *** 0.05 * −0.12 *** 0.02 (0.03) (0.03) (0.04) (0.03) Constant −2.59 ** −2.30 −3.93 ** −0.22 (1.17) (1.93) (1.54) (1.45) Observations 1,892 2,300 1,936 2,300 R-squared 0.47 0.17 0.39 0.44 Notes: The dependent variables ‘Exports’ and ‘Imports’ stand, respectively, for the sum of bilateral exports divided by GDP and the sum of bilateral imports divided by GDP. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large © The Author(s) 2018. Published by Oxford University Press on behalf of the Centre for the Study of African Economies, all rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of African Economies Oxford University Press

Growth-enhancing Effect of Openness to Trade and Migrations: What is the Effective Transmission Channel for Africa?

Loading next page...
 
/lp/ou_press/growth-enhancing-effect-of-openness-to-trade-and-migrations-what-is-9C0lRbfrhx
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Centre for the Study of African Economies, all rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
0963-8024
eISSN
1464-3723
D.O.I.
10.1093/jae/ejx044
Publisher site
See Article on Publisher Site

Abstract

Abstract This paper investigates the growth-enhancing effect of openness to trade and to migration by focusing on African countries. Relying on robust estimation techniques dealing with both endogeneity and omitted-variables issues, our results put forward the importance of accounting for the type of the partner country. We found evidence that while trade between Africa and industrialised countries has a clear and robust positive impact on Africa’s standards of living, trade with developing countries fails to be growth-enhancing. Moreover, our findings show that migration has no significant effect on per capita income in Africa regardless of the partner. Finally, exploring the trade openness transmission channel, we establish that the growth-enhancing effect of Africa’s trade with industrialised countries mainly occurs through an improvement in total factor productivity. 1. Introduction While a vast literature exists on the link between income and openness to trade,1Frankel and Romer (1999) were the first to offer a convincing causality analysis regarding the income-enhancing effect of trade openness. The authors use the geographic characteristics as an instrument in a gravity-type model to demonstrate a positive effect of trade on per capita income; the main argument being that these factors are plausibly uncorrelated with other determinants of income per person. These findings were confirmed by several subsequent works (see among others Frankel and Rose, 2002; Dollar and Kraay, 2003; Noguer and Siscart, 2005; Freund and Bolaky, 2008), including across different time periods (Irwin and Terviö, 2002). However, consensus is far from clear on this issue. Rodriguez and Rodrik (2000) highlight the non-robust nature of these results once controlled for omitted variables such as distance from the equator or institutions. More recently, Ortega and Peri (2014) go a step further, and argue that the geographical factors used by Frankel and Romer (1999) can also impact income through migration. Geographic characteristics may raise income through the interactions between countries (exchange of ideas, technological diffusion, innovation, investment) and these interactions would be reflected in the mobility of goods (trade) and of people (migration). Thus, trade is not the sole vehicle of globalisation through which interactions between countries promote economic growth. Acknowledging that openness to trade and openness to migration may be both considered as determinants of income,2Ortega and Peri (2014) found evidence of a strong positive effect of openness to migration on long-run per capita income but fail to do so for trade openness. Despite the abundance of the literature, the debate is still open regarding the relationship between income and openness. Indeed, previous studies indiscriminately examine the growth-enhancing effect of openness without accounting for heterogeneity of countries regarding the benefits or costs of openness. This paper fills this gap and focuses on the specific case of Africa. We aim at studying the overall effect of openness on long-term growth in Africa by paying particular attention to the type—African, other developing, developed—of partner countries. To this end, we retain the general trade-growth identification setting of Frankel and Romer (1999),3 and follow Ortega and Peri (2014) in considering that intensity of openness between two countries should be captured by both bilateral trade and bilateral migration. Such a framework is even more relevant in the case of Africa, where openness to global finance is still in its infancy. The choice of Africa and its singularity deserve some comments. Firstly, by scrutinising the architecture of international trade, the case of Africa stands out as unique. As shown in Figure 1, unlike the rest of the world exports of African countries largely focus on commodities, while their imports are dominated by manufactured goods with a similar overall structure to that of developing and industrialised economies. Furthermore, as illustrated by the right side of Figure 2, Africa’s trade (imports and exports) is mainly realised with developed countries. Although this trade orientation could be beneficial for long-term growth in Africa—particularly through improvement in total factor productivity4—this growth is subject to the ups and downs of the terms of trade due to the high concentration of exports on commodities. Figure 1: View largeDownload slide Comparative Structure of International Trade. Notes: The left-hand side (resp. right-hand side) figure reports the percentage of primary commodities and manufactured goods in the total exports (resp. imports) for each region. DE=DevelopingEconomies. Data source: UNCTAD (mean values over the 1995–2014 period) Figure 1: View largeDownload slide Comparative Structure of International Trade. Notes: The left-hand side (resp. right-hand side) figure reports the percentage of primary commodities and manufactured goods in the total exports (resp. imports) for each region. DE=DevelopingEconomies. Data source: UNCTAD (mean values over the 1995–2014 period) Figure 2: View largeDownload slide Openness of Africa (in 2000). Notes: Trade is measured by the sum of imports and exports. Migration from Africa is measured by the stock of African nationals living abroad. Data sources: UNCTAD (trade data) and World Bank (migration data) Figure 2: View largeDownload slide Openness of Africa (in 2000). Notes: Trade is measured by the sum of imports and exports. Migration from Africa is measured by the stock of African nationals living abroad. Data sources: UNCTAD (trade data) and World Bank (migration data) Secondly, statistics on international migration underline that Africa is characterised by (i) strong intra-continental migration (i.e., immigration and emigration within African countries), and (ii) emigration to industrialised OECD countries. As shown in Figure 2, Africa’s openness to migration in 2000 was more than half intra-African, while one-third was directed towards the industrialised OECD countries. This migration structure of Africa can be seen somewhat dichotomous. On the one hand, it may be viewed as detrimental because ‘brain drain’ (emigration of relatively highly educated individuals) could hamper economic development in Africa. On the other hand, it may be considered as an enhancer factor of development in the sense that African nationals living in industrialised countries are vectors of transmission of human and technological capital (education and experience), but also vectors of transmission of financial capital (migrants’ remittances) and better institutions.5 Finally, despite the strong dominance of developed countries in Africa’s trade, some developing economies such as China are gaining more and more market share in Africa since the beginning of the 2000s. If the growth-enhancing effect of openness between Africa and its new developing partners is debatable (see, among others, Lyons and Brown, 2010; He, 2013; Kaplinsky, 2013), these dynamics bring back the old question about the impact of South-South and North-South openness on growth and productivity in the southern countries. Addressing this hot-debated issue is thus worthy of investigation due to the continuously increasing role played by China in African trade. Falling into the strand of the literature initiated by Frankel and Romer (1999) and Romer (1999), our contribution is 3-fold. First, while the previous literature is mainly done at a global level, we pay particular attention to countries’ specificities and heterogeneity in the face of openness by focusing on a panel of African economies. Second, we go further than previous studies by highlighting the importance of the trading partner. We investigate whether the effect of openness to trade and to migration on growth is sensitive to the type (African, other developing, industrialised) of the partner country. In doing so, we also contribute to the very topical debate concerning China-Africa trade links. Third, in addition to the detailed study of the openness-income nexus, we identify the transmission channel through which trade affects growth. Our main results can be summarised as follows. First, we establish a mitigated overall impact of openness on income in Africa. While trade seems to exert a positive effect on income, this impact is not robust to the inclusion of control variables. The influence of immigration is also fragile and depends on the method used to predict the geographic component of openness. Second, we put forward the importance of accounting for the type of the trading partner. Indeed, we found evidence of a clear and robust partner-varying impact of openness for Africa: only trade with industrialised countries has a strong and robust positive impact on income. Compared to Ortega and Peri’s (2014) contribution—which is the closest paper to ours and which insists on the dominant role of migration—we thus rehabilitate the growth-enhancing effect of trade, provided that Africa’s trade partner country is an advanced one. Third, the positive impact of migration from African economies to industrialised countries (emigration for Africa) is not robust. This probably reflects the confrontation between the ‘brain drain’ negative effect and the ‘productivity transfer’ positive impact of emigration for Africa. Moreover, we found that Africa’s openness (both to trade and to migration) with developing and emerging countries—including China—fails to improve per capita income.6 Finally, exploring the openness transmission channel thanks to the income decomposition of Hall and Jones (1999), we establish that the growth-enhancing effect of African trade with industrialised countries mainly occurs through an improvement in total factor productivity. Various sensitivity analyses are provided to assess the robustness of all our findings. The rest of the paper is organised as follows. Section 2 describes our empirical strategy. Section 3 is devoted to the presentation of data. In Section 4, we present and discuss our main results, and provide some robustness checks. Section 5 is dedicated to examining the transmission channel through which openness impacts income. Finally, Section 6 concludes the paper. 2. Empirical strategy Our empirical framework is inspired from Ortega and Peri (2014) which, in turn, extends the specification proposed by Frankel and Romer (1999). To overcome the well-known endogeneity issue in the trade-income relationship, Frankel and Romer (1999) rely on the instrumental variable technique based on a gravity model. They estimate the causal effects of trade on income using cross-country variation in trade flows due to bilateral geography. According to and Ortega specification suffers from a potential omitted-variables problem because trade and migration openness are both influenced by geography. Thus, country’s geographic characteristics can affect income not only through trade but also through migration. Indeed, geographical proximity and accessibility raise income through the interactions between countries (exchange of ideas, technological diffusion, innovation, investment) which would be reflected in the mobility of goods and people (Ortega and Peri, 2014). In other words, trade is not the sole channel through which interactions between countries increase income. Therefore, to fully identify the impact of trade openness, these two vehicles of globalisation should be jointly considered. 2.1 Baseline specification Our empirical model is given by: lnYi=α0+αTTi+αMMi+αPlnPopi+αAlnAreai+β′Xi+ui (1) where Yi denotes per capita income in country i, Ti and Mi represent openness to trade and openness to migration, respectively, Popi and Areai stand for population and area which capture the impact of country size, Xi collects control variables, and ui is the error term. The rationale behind this empirical model is as follows. Classical international trade theory has highlighted that openness to trade increases output through specialisation based on comparative advantages. New trade theory has documented the growth-enhancing role of trade by focusing on the exploitation of increasing returns to scale and network effects (Helpman and Krugman, 1985; and Grossman and Helpman, 1991a, b). The joint impact of trade and migration on income is explained by Ortega and Peri (2014) in a simple multi-country model that features trade and migration flows both across country borders and across regions within the same country. In this model—which extends Alesina et al. (2000)—aggregate production is a function of varieties for intermediate goods and human capital; and each region is endowed with a differentiated good and a differentiated type of labour. Intermediate goods and labour being mobile across regions of different countries but subject to iceberg-type costs, this model derives income per worker as a function of theoretical measures of trade and migration openness which are, respectively, inverse measures of trade and migration costs. Their empirical counterparts are, respectively, trade flows (exports+imports) as share of GDP, and immigration rate (foreign-born) as share of total population. The income-enhancing impact of openness to migration in the theoretical model of Ortega and Peri (2014) operates through an increase in total factor productivity reflecting growing diversity in productive skills caused by immigration. At a first sight, this channel is not very relevant for African economies which are net labour-sending countries: brain drain might negatively affect income per capita by depriving African economies of valuable talents. However, there are many channels through which emigration can promote economic performance in home countries (see Section 5 for an investigation). Foremost, remittances for emigrants can compensate for the loss of workers by enabling households and entrepreneurs to overcome credit constraints and providing an alternative way to finance investment in human and physical capital (Giuliano and Ruiz-Arranz, 2009). Besides, home countries can benefit from human capital of returning migrants (Stark et al., 1997; Beine et al., 2008) and the transfer of knowledge through the diaspora (Ortega and Peri, 2014). Furthermore, since there is strong evidence of the role of institutions in economic development,7 emigration can be profitable to economic growth in the home country by improving the quality of institutions. Indeed, many recent studies in international migration literature highlight the role of emigration in improving institutions (Spilimbergo, 2009; Docquier et al., 2016). Using an international dataset, Spilimbergo (2009) shows that foreign-educated individuals play an important role in fostering democracy in the home country, but only if foreign education is acquired in democratic countries. Based on cross-section and panel analyses for a large sample of developing countries, Docquier et al. (2016) also found that general emigration has a positive effect on the quality of institutions in the home country. Acknowledging the econometric concerns discussed above, Ortega and Peri (2014) propose to instrument both trade openness and openness to migration by their gravity-based predictors. This geography-based prediction of bilateral trade or bilateral migration stock is obtained by estimating the following pseudo-gravity model: lnWij=γ0+γ1lnDistij+γ2lnPopi+γ3lnPopj+γ4lnAreai+γ5lnAreaj+γ6(Landlockedi+Landlockedj)+γ7Borderij+γ8Colonyij+γ9ComLangij+γ10Comcurij+γ11Timeij+γ12lnDistij×Borderij+γ13lnPopi×Borderij+γ14lnPopj×Borderij+γ15lnAreai×Borderij+γ16lnAreaj×Borderij+γ17(Landlockedi+Landlockedj)×Borderij+eij (2) where Wij is either bilateral trade—i.e., the value of trade (exports + imports)8 between countries i and j divided by the GDP of origin country i—or bilateral migration (emigration)—i.e., the stock of migrants born in country j ( i) and living in country i ( j) as share of country i’s population, Distij is the distance between country i and country j, Pop and Area are the same variables defined in (1) and they are included to account for country size, Landlocked is a dummy variable for landlocked countries, Border is a dummy variable to indicate whether countries i and j share a common border, Colony is a dummy for colonial relationship, and ComLang is a dummy for sharing a common official language. Our specification includes an additional variable ( Comcur) compared to Ortega and Peri (2014). This variable aims at capturing the sharing of a common currency and might play an important role since the impact of currency unions on bilateral trade was frequently relayed in related studies (Rose, 2000, 2001; Frankel and Rose, 2002). As argued by the literature on the endogeneity of optimum currency area criteria, sharing a single currency may set motion forces that promote economic integration and then facilitate migration. Following Ortega and Peri (2014), we include time zone differences denoted by Timeij. As mentioned by Head et al. (2009) the impact of time zone differences between the exporting country and its trading partners is ambiguous since two contradictory effects that differ across service subcategories are at play, namely the continuity effect (the ability to operate around the clock) and the synchronisation effect (the need to coordinate during business hours). Since a large part of trade is with immediate neighbours, we finally include interaction terms of border dummy with the distance, population, area and landlocked dummies (Frankel and Romer, 1999). For the sake of completeness and as previously mentioned, it should be emphasised that we mainly focus in this paper on openness in terms of trade and migration. In other words, we do not specifically account for the third dimension of international relationships, namely financial capital mobility. Our choice is justified by the fact that we concentrate on Africa, i.e. on countries for which the financial sector is not highly developed. As it is a well-established fact that weak financial integration does not favour capital mobility,9 we do not include the financial dimension in our estimations. However, although we do not introduce remittances in our gravity model, we fully acknowledge their importance as a source of income growth and we will investigate this transmission channel in Section 5. Once the gravity regressions described by (2) are estimated, we sum them up over partner countries j to obtain the predicted trade and migration openness for each origin country i. More specifically, let Zij be the vector of explanatory variables included in Equation (2), ΓT the vector of coefficients in the bilateral trade regression, and ΓM the corresponding vector for the bilateral migration regression. The gravity-based predictor of trade openness for origin country i, Tˆi, is then obtained by summing bilateral trade up over partner countries j: Tˆi=∑j≠iexp(ΓTZij) (3) Similarly, the gravity-based predictor of migration openness for origin country i, Mˆi, is given by: Mˆi=∑j≠iexp(ΓMZij) (4) 2.2 Identifying partner-varying impact of openness The income-enhancing impact of openness described in Equation (1) is based on the idea that interactions among countries affect income (through trade and migration) in the same way whatever the partner country. However, it is very reasonable to think that the income-enhancing impact of openness (to trade and to migration) depends on the partner country, especially for African economies. First, with regards to openness to trade, new trade theory suggests that a country can obtain advanced technology from its trading partners through trade. If this channel is dominant, countries may benefit more from trading with advanced economies which are more technologically innovative (Yanikkaya, 2003). As a consequence, trade with industrialised countries may be more income-enhancing for Africa than trade with other countries like China and African neighbours. Second, turning to migration openness, its impact should also depend on the partner country. Openness to migration in African countries is mainly characterised by an important intra-continental mobility and emigration to developed countries (Europe, North America). Therefore, because of the channel advocated by Ortega and Peri (2014) —i.e., growing diversity in productive skills—and the aforementioned ambiguous relationship between growth and emigration to developed countries, it is reasonable to think that the effect of intra-African openness to migration (i.e., immigration from African countries) should be different from the impact of African migration with developed countries (which mainly consists in emigration from Africa to advanced countries). To evaluate this partner-varying impact of the two vehicles of globalisation, we consider the following disaggregated model: lnYi=α0+αTSTiS+αMSMiS+αPlnPopi+αAlnAreai+β′Xi+ui (5) where TiS and MiS, respectively, represent trade and migration openness of an African country i with a particular partner country j∈S, S being either the subset of African partners, the subset of partners among developing countries, or the subset of partners among industrialised countries. To better characterise openness with regard to the subset of partners S, openness to trade and to migration are instrumented by estimating the pseudo-gravity model on the considered subset of partners S: lnWijS=γ0S+γ1SlnDistij+γ2SlnPopi+γ3SlnPopj+γ4SlnAreai+γ5SlnAreaj+γ6S(Landlockedi+Landlockedj)+γ7SBorderij+γ8SColonyij+γ9SComLangij+γ10SComcurij+γ11STimeij+γ12SlnDistij×Borderij+γ13SlnPopi×Borderij+γ14SlnPopj×Borderij+γ15SlnAreai×Borderij+γ16SlnAreaj×Borderij+γ17S(Landlockedi+Landlockedj)×Borderij+eijS (6) Based on the estimation of the disaggregated pseudo-gravity model in Equation (6), the gravity-based predictors of trade openness ( TˆiS) and migration openness ( MˆiS) for country i are, respectively, obtained by summing bilateral trade and migration up over partner countries j among the subset S: TˆiS=∑j∈Sexp(ΓTZij) (7) MˆiS=∑j∈Sexp(ΓMZij) (8) 2.3 Control variables and identification The validity of geographically constructed instrumental variables is weakened by the fact that geographical features may influence directly income per capita or indirectly through other channels than openness (Rodriguez and Rodrik, 2000): impacts of location and climate on transport costs, disease burdens, agricultural productivity and natural resources endowment (Gallup et al., 1999), colonial history and institutional quality (Hall and Jones, 1999; Acemoglu et al., 2001). To tackle this econometric issue, we include a set of control variables allowing us to account for all the main potential channels through which geographical features can influence income per capita. More precisely, we consider distance to the equator, the key geographic variable found to increase the odds of European settlements in the country and, therefore, to determine the history of institutions’ quality (Hall and Jones, 1999; Rodriguez and Rodrik, 2000). The choice of this variable deserves some comments. While we do not consider strictly speaking an institutional quality index among our controls in the income model, the distance to equator allows us to account for institutions’ quality in a given country. Specifically, the distance from the equator may be viewed as reflecting the effect of climate or as a proxy for omitted country’s specificities that are correlated with latitude. The underlying idea is that countries which are nearer the equator have generally worse health conditions and institutions (Sachs and Warner, 1997; Hall and Jones, 1999; Easterly and Levine, 2001). Thus, the distance to equator variable allows us to take into account various country’s characteristics among which the quality of institutions in a single proxy which has the advantage of parsimony. Note that, as for remittances and for the sake of completeness, we will investigate further in Section 5 the role of openness to migration in promoting growth. We also consider other controls: a landlocked dummy and distance to the coast to control for transport costs, the percent of land area in geographical tropics to account for agricultural productivity, disease environment (incidence of malaria and yellow fever) that may influence human history, and legal origin from colonial history that matters for economic outcomes (La Porta et al., 1999, 2008). 3. Data Our data are taken from various sources. Data on bilateral trade are collected from the IMF Direction of Trade Statistics (DOTS). The DOTS database contains data on the value of merchandise exports and imports between each country and all its trading partners. The period for which data are available depends on the considered country but for most of them data extend from the 1980s to the present. As in Ortega and Peri (2014), bilateral migration data are taken from Docquier et al. (2010)—a database of bilateral stocks of immigrants (and emigrants) covering the 1990s to 2000s for 194 countries. Data on geographic variables are from the CEPII’s Gravity database described in Head et al. (2010) and from Gallup et al. (1999). We also use the Gravity database for ethnic, linguistic and colonial ties. The real income per person (real PPP-adjusted GDP per person) is collected from the Penn World Tables (version 7.1). Data on nominal GDP and population are taken from the World Bank World Development Indicators (WDI) database. As in Ortega and Peri (2014), we use the database from Acemoglu et al. (2001) for legal origins, oil endowment and disease environment. Paying particular attention to Africa, we consider, in addition to our whole, world sample, a subsample of 52 African countries.10 Table 1 reports some basic descriptive statistics for the main variables considered in the paper. The mean of real GDP per person in the world is $10,732 with a standard deviation of 13,067, while in Africa the mean is only 2,790 with a standard deviation of 4,502. The minimum real GDP per person in the world is in Africa. Average trade share is 54% in the world and 43% in Africa.11 In line with Figure 2, African trade is dominated by trade with industrialised countries, while intra-African trade is very small. Concerning migration (foreign-born) share (per 1,000 population), its mean is 0.05 in the world and 0.03 in Africa, with a standard deviation of 0.08 and 0.06, respectively. Openness to migration in African countries is characterised by an important intra-continental mobility and emigration to industrialised countries. Table 1 also reports descriptive statistics on some of our main control variables (population, area, percent of European descent in 1900). Table 1: Descriptive Statistics Mean Std. dev. Min. Max. N Whole sample Real GDP per person in 2000 (PPP, 2005 USD) 10,732 13,067 180 65,125 187 Trade (in % of GDP) 0.54 0.42 0 2.68 200 Immigration rate 0.05 0.08 0 0.53 200 Population (in thousands) 30,739 119,883 942 1,262,645 200 Area (in sq. kms) 691,427 1,894,387 25 17,075,400 200 Distance to equator (in degrees) 25.79 16.97 0.2 64.18 200 Euro. descent in 1900 (in %) 29.57 41.69 0 100 157 Africa sample Real GDP per person in 2000 (PPP, 2005 USD) 2,790 4,502 180 25,993 52 Trade (in % of GDP) Total 0.43 0.31 0 1.33 52 Intra-African 0.07 0.08 0 0.31 52 With no highly ind. 0.17 0.15 0 0.83 52 With highly ind. 0.25 0.22 0 1 52 Migration rate Total 0.03 0.06 0 0.32 52 Intra-African 0.03 0.06 0 0.31 52 From no highly ind. (including Africa) 0.03 0.06 0 0.31 52 To highly ind. 0.01 0.03 0 0.16 52 Population (in thousands) 15,428 21,909 81.13 122,877 52 Area (in sq. kms) 581,749 643,900 455 2,505,813 52 Distance to equator (in degrees) 13.47 10.05 0.2 36.83 52 Euro. descent in 1900 (in %) 3.64 14.33 0 100 51 Mean Std. dev. Min. Max. N Whole sample Real GDP per person in 2000 (PPP, 2005 USD) 10,732 13,067 180 65,125 187 Trade (in % of GDP) 0.54 0.42 0 2.68 200 Immigration rate 0.05 0.08 0 0.53 200 Population (in thousands) 30,739 119,883 942 1,262,645 200 Area (in sq. kms) 691,427 1,894,387 25 17,075,400 200 Distance to equator (in degrees) 25.79 16.97 0.2 64.18 200 Euro. descent in 1900 (in %) 29.57 41.69 0 100 157 Africa sample Real GDP per person in 2000 (PPP, 2005 USD) 2,790 4,502 180 25,993 52 Trade (in % of GDP) Total 0.43 0.31 0 1.33 52 Intra-African 0.07 0.08 0 0.31 52 With no highly ind. 0.17 0.15 0 0.83 52 With highly ind. 0.25 0.22 0 1 52 Migration rate Total 0.03 0.06 0 0.32 52 Intra-African 0.03 0.06 0 0.31 52 From no highly ind. (including Africa) 0.03 0.06 0 0.31 52 To highly ind. 0.01 0.03 0 0.16 52 Population (in thousands) 15,428 21,909 81.13 122,877 52 Area (in sq. kms) 581,749 643,900 455 2,505,813 52 Distance to equator (in degrees) 13.47 10.05 0.2 36.83 52 Euro. descent in 1900 (in %) 3.64 14.33 0 100 51 Notes: Statistics are reported for the year 2000. N denotes the number of countries. View Large Table 1: Descriptive Statistics Mean Std. dev. Min. Max. N Whole sample Real GDP per person in 2000 (PPP, 2005 USD) 10,732 13,067 180 65,125 187 Trade (in % of GDP) 0.54 0.42 0 2.68 200 Immigration rate 0.05 0.08 0 0.53 200 Population (in thousands) 30,739 119,883 942 1,262,645 200 Area (in sq. kms) 691,427 1,894,387 25 17,075,400 200 Distance to equator (in degrees) 25.79 16.97 0.2 64.18 200 Euro. descent in 1900 (in %) 29.57 41.69 0 100 157 Africa sample Real GDP per person in 2000 (PPP, 2005 USD) 2,790 4,502 180 25,993 52 Trade (in % of GDP) Total 0.43 0.31 0 1.33 52 Intra-African 0.07 0.08 0 0.31 52 With no highly ind. 0.17 0.15 0 0.83 52 With highly ind. 0.25 0.22 0 1 52 Migration rate Total 0.03 0.06 0 0.32 52 Intra-African 0.03 0.06 0 0.31 52 From no highly ind. (including Africa) 0.03 0.06 0 0.31 52 To highly ind. 0.01 0.03 0 0.16 52 Population (in thousands) 15,428 21,909 81.13 122,877 52 Area (in sq. kms) 581,749 643,900 455 2,505,813 52 Distance to equator (in degrees) 13.47 10.05 0.2 36.83 52 Euro. descent in 1900 (in %) 3.64 14.33 0 100 51 Mean Std. dev. Min. Max. N Whole sample Real GDP per person in 2000 (PPP, 2005 USD) 10,732 13,067 180 65,125 187 Trade (in % of GDP) 0.54 0.42 0 2.68 200 Immigration rate 0.05 0.08 0 0.53 200 Population (in thousands) 30,739 119,883 942 1,262,645 200 Area (in sq. kms) 691,427 1,894,387 25 17,075,400 200 Distance to equator (in degrees) 25.79 16.97 0.2 64.18 200 Euro. descent in 1900 (in %) 29.57 41.69 0 100 157 Africa sample Real GDP per person in 2000 (PPP, 2005 USD) 2,790 4,502 180 25,993 52 Trade (in % of GDP) Total 0.43 0.31 0 1.33 52 Intra-African 0.07 0.08 0 0.31 52 With no highly ind. 0.17 0.15 0 0.83 52 With highly ind. 0.25 0.22 0 1 52 Migration rate Total 0.03 0.06 0 0.32 52 Intra-African 0.03 0.06 0 0.31 52 From no highly ind. (including Africa) 0.03 0.06 0 0.31 52 To highly ind. 0.01 0.03 0 0.16 52 Population (in thousands) 15,428 21,909 81.13 122,877 52 Area (in sq. kms) 581,749 643,900 455 2,505,813 52 Distance to equator (in degrees) 13.47 10.05 0.2 36.83 52 Euro. descent in 1900 (in %) 3.64 14.33 0 100 51 Notes: Statistics are reported for the year 2000. N denotes the number of countries. View Large 4. Empirical results 4.1 Gravity estimates for trade and migration Columns (1–4) of Table 2 contain the gravity model estimation results for global openness (trade and migration) as in Ortega and Peri (2014), while columns (5–8) concern openness of Africa to the world. The odd columns display the OLS results, while the even columns are related to the Poisson Pseudo Maximum Likelihood (PPML) non-linear approach. As argued by Silva and Tenreyro (2006), contrary to the log-linearised model estimation by OLS, PPML estimation has two main advantages: it allows to deal with (i) observations of the dependent variable with zero value, and (ii) heteroskedasticity-related issues. We follow the procedure of Silva and Tenreyro (2010) in order to address the identification problem of the (pseudo) maximum likelihood estimates of the Poisson regression models with non-negative values of the dependent variable (bilateral trade or bilateral migration) and a large number of zeros on some regressors. Table 2: Gravity Regression, African Openness with the World All countries African countries Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.85 *** −1.04 *** −1.35 *** −1.40 *** −0.77 *** −0.66 *** −1.52 *** −1.45 *** (0.04) (0.06) (0.04) (0.08) (0.08) (0.11) (0.08) (0.15) Ln pop. origin 0.04 *** −0.08 ** −0.41 *** −0.35 *** 0.05 * −0.16 *** −0.50 *** −0.47 *** (0.02) (0.04) (0.02) (0.07) (0.03) (0.05) (0.04) (0.12) Ln pop. dest. 1.06 *** 0.74 *** 0.65 *** 0.80 *** 1.20 *** 0.97 *** 0.57 *** 0.52 *** (0.01) (0.03) (0.02) (0.08) (0.03) (0.05) (0.04) (0.11) Ln area origin −0.06 *** 0.04 0.19 *** 0.14 *** −0.00 0.12 *** 0.24 *** 0.14 * (0.01) (0.03) (0.02) (0.03) (0.02) (0.04) (0.03) (0.08) Ln area dest. −0.23 *** −0.08 ** −0.08 *** −0.07 −0.33 *** −0.25 *** −0.13 *** −0.01 (0.01) (0.04) (0.02) (0.05) (0.02) (0.06) (0.03) (0.10) Sum landlocked −0.85 *** −0.55 *** −0.28 *** −0.64 *** −0.96 *** −0.96 *** −0.44 *** −0.79 *** (0.04) (0.07) (0.05) (0.13) (0.06) (0.12) (0.08) (0.23) Border −4.60 *** −1.71 −0.17 −5.17 *** 6.21 *** 0.61 0.63 −0.90 (0.87) (1.07) (0.99) (1.69) (1.80) (2.28) (1.68) (2.87) Border*Ln dist. 0.92 *** 0.27 0.07 1.12 *** −0.48 0.40 0.28 0.50 (0.19) (0.29) (0.22) (0.36) (0.42) (0.64) (0.33) (0.60) Border*Ln pop. origin −0.33 *** 0.17 −0.13 0.31 ** −0.18 0.12 0.05 0.31 (0.09) (0.10) (0.09) (0.14) (0.17) (0.23) (0.14) (0.25) Border*Ln pop. dest. −0.18 ** −0.10 −0.25 *** −0.64 *** −0.52 *** −0.69 *** −0.10 −0.20 (0.08) (0.09) (0.09) (0.13) (0.16) (0.22) (0.16) (0.22) Border*Ln area origin 0.01 −0.16 −0.16 −0.46 *** −0.23 −0.44 ** −0.40 *** −0.27 (0.08) (0.10) (0.10) (0.06) (0.17) (0.22) (0.14) (0.21) Border*Ln area dest. −0.04 0.08 0.34 *** 0.31 * 0.22 0.20 0.11 −0.05 (0.09) (0.16) (0.09) (0.17) (0.17) (0.17) (0.14) (0.25) Border*landlocked 0.72 *** 0.65 *** 0.32 ** 0.43 ** 0.86 *** 0.85 *** 0.24 0.47 (0.12) (0.15) (0.12) (0.21) (0.20) (0.31) (0.19) (0.34) Common language 0.67 *** 0.74 *** 0.91 *** 0.93 *** 0.36 *** 0.27 0.65 *** 1.29 *** (0.09) (0.13) (0.10) (0.20) (0.12) (0.21) (0.14) (0.36) Common off. lang. −0.05 −0.03 0.45 *** 0.16 0.47 *** 0.30 0.88 *** 0.11 (0.09) (0.14) (0.10) (0.22) (0.12) (0.22) (0.14) (0.37) Colonial ties 3.16 *** 1.67 *** 1.33 *** 0.95 *** 4.11 *** 2.51 *** 1.24 *** 0.84 * (0.14) (0.14) (0.17) (0.22) (0.18) (0.19) (0.27) (0.44) Origin hegemon −2.29 *** −2.06 *** 0.97 *** 0.46 −3.35 *** −3.07 *** 1.09 *** 0.72 (0.19) (0.21) (0.22) (0.33) (0.29) (0.29) (0.34) (0.62) Time zone diff. 0.13 *** 0.05 *** 0.09 *** −0.03 −0.12 *** −0.00 −0.01 −0.30 *** (0.01) (0.02) (0.01) (0.04) (0.02) (0.04) (0.02) (0.07) Common currency 0.79 *** 0.58 *** 0.99 *** 0.55 ** 0.55 *** −0.19 0.75 *** 0.90 ** (0.12) (0.14) (0.13) (0.25) (0.21) (0.36) (0.21) (0.43) Constant 5.60 *** 0.96 ** −0.79 ** 0.73 −3.72 *** −2.53 *** 0.98 1.49 (0.29) (0.43) (0.33) (0.55) (0.66) (0.92) (0.70) (1.16) Observations 20,980 37,044 8,219 38,612 8,311 16,761 2,426 17,462 R-squared 0.41 0.20 0.44 0.40 0.34 0.19 0.55 0.19 All countries African countries Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.85 *** −1.04 *** −1.35 *** −1.40 *** −0.77 *** −0.66 *** −1.52 *** −1.45 *** (0.04) (0.06) (0.04) (0.08) (0.08) (0.11) (0.08) (0.15) Ln pop. origin 0.04 *** −0.08 ** −0.41 *** −0.35 *** 0.05 * −0.16 *** −0.50 *** −0.47 *** (0.02) (0.04) (0.02) (0.07) (0.03) (0.05) (0.04) (0.12) Ln pop. dest. 1.06 *** 0.74 *** 0.65 *** 0.80 *** 1.20 *** 0.97 *** 0.57 *** 0.52 *** (0.01) (0.03) (0.02) (0.08) (0.03) (0.05) (0.04) (0.11) Ln area origin −0.06 *** 0.04 0.19 *** 0.14 *** −0.00 0.12 *** 0.24 *** 0.14 * (0.01) (0.03) (0.02) (0.03) (0.02) (0.04) (0.03) (0.08) Ln area dest. −0.23 *** −0.08 ** −0.08 *** −0.07 −0.33 *** −0.25 *** −0.13 *** −0.01 (0.01) (0.04) (0.02) (0.05) (0.02) (0.06) (0.03) (0.10) Sum landlocked −0.85 *** −0.55 *** −0.28 *** −0.64 *** −0.96 *** −0.96 *** −0.44 *** −0.79 *** (0.04) (0.07) (0.05) (0.13) (0.06) (0.12) (0.08) (0.23) Border −4.60 *** −1.71 −0.17 −5.17 *** 6.21 *** 0.61 0.63 −0.90 (0.87) (1.07) (0.99) (1.69) (1.80) (2.28) (1.68) (2.87) Border*Ln dist. 0.92 *** 0.27 0.07 1.12 *** −0.48 0.40 0.28 0.50 (0.19) (0.29) (0.22) (0.36) (0.42) (0.64) (0.33) (0.60) Border*Ln pop. origin −0.33 *** 0.17 −0.13 0.31 ** −0.18 0.12 0.05 0.31 (0.09) (0.10) (0.09) (0.14) (0.17) (0.23) (0.14) (0.25) Border*Ln pop. dest. −0.18 ** −0.10 −0.25 *** −0.64 *** −0.52 *** −0.69 *** −0.10 −0.20 (0.08) (0.09) (0.09) (0.13) (0.16) (0.22) (0.16) (0.22) Border*Ln area origin 0.01 −0.16 −0.16 −0.46 *** −0.23 −0.44 ** −0.40 *** −0.27 (0.08) (0.10) (0.10) (0.06) (0.17) (0.22) (0.14) (0.21) Border*Ln area dest. −0.04 0.08 0.34 *** 0.31 * 0.22 0.20 0.11 −0.05 (0.09) (0.16) (0.09) (0.17) (0.17) (0.17) (0.14) (0.25) Border*landlocked 0.72 *** 0.65 *** 0.32 ** 0.43 ** 0.86 *** 0.85 *** 0.24 0.47 (0.12) (0.15) (0.12) (0.21) (0.20) (0.31) (0.19) (0.34) Common language 0.67 *** 0.74 *** 0.91 *** 0.93 *** 0.36 *** 0.27 0.65 *** 1.29 *** (0.09) (0.13) (0.10) (0.20) (0.12) (0.21) (0.14) (0.36) Common off. lang. −0.05 −0.03 0.45 *** 0.16 0.47 *** 0.30 0.88 *** 0.11 (0.09) (0.14) (0.10) (0.22) (0.12) (0.22) (0.14) (0.37) Colonial ties 3.16 *** 1.67 *** 1.33 *** 0.95 *** 4.11 *** 2.51 *** 1.24 *** 0.84 * (0.14) (0.14) (0.17) (0.22) (0.18) (0.19) (0.27) (0.44) Origin hegemon −2.29 *** −2.06 *** 0.97 *** 0.46 −3.35 *** −3.07 *** 1.09 *** 0.72 (0.19) (0.21) (0.22) (0.33) (0.29) (0.29) (0.34) (0.62) Time zone diff. 0.13 *** 0.05 *** 0.09 *** −0.03 −0.12 *** −0.00 −0.01 −0.30 *** (0.01) (0.02) (0.01) (0.04) (0.02) (0.04) (0.02) (0.07) Common currency 0.79 *** 0.58 *** 0.99 *** 0.55 ** 0.55 *** −0.19 0.75 *** 0.90 ** (0.12) (0.14) (0.13) (0.25) (0.21) (0.36) (0.21) (0.43) Constant 5.60 *** 0.96 ** −0.79 ** 0.73 −3.72 *** −2.53 *** 0.98 1.49 (0.29) (0.43) (0.33) (0.55) (0.66) (0.92) (0.70) (1.16) Observations 20,980 37,044 8,219 38,612 8,311 16,761 2,426 17,462 R-squared 0.41 0.20 0.44 0.40 0.34 0.19 0.55 0.19 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Immigration’ reflects migration openness measured by the number of foreign-born living in the country divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large Table 2: Gravity Regression, African Openness with the World All countries African countries Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.85 *** −1.04 *** −1.35 *** −1.40 *** −0.77 *** −0.66 *** −1.52 *** −1.45 *** (0.04) (0.06) (0.04) (0.08) (0.08) (0.11) (0.08) (0.15) Ln pop. origin 0.04 *** −0.08 ** −0.41 *** −0.35 *** 0.05 * −0.16 *** −0.50 *** −0.47 *** (0.02) (0.04) (0.02) (0.07) (0.03) (0.05) (0.04) (0.12) Ln pop. dest. 1.06 *** 0.74 *** 0.65 *** 0.80 *** 1.20 *** 0.97 *** 0.57 *** 0.52 *** (0.01) (0.03) (0.02) (0.08) (0.03) (0.05) (0.04) (0.11) Ln area origin −0.06 *** 0.04 0.19 *** 0.14 *** −0.00 0.12 *** 0.24 *** 0.14 * (0.01) (0.03) (0.02) (0.03) (0.02) (0.04) (0.03) (0.08) Ln area dest. −0.23 *** −0.08 ** −0.08 *** −0.07 −0.33 *** −0.25 *** −0.13 *** −0.01 (0.01) (0.04) (0.02) (0.05) (0.02) (0.06) (0.03) (0.10) Sum landlocked −0.85 *** −0.55 *** −0.28 *** −0.64 *** −0.96 *** −0.96 *** −0.44 *** −0.79 *** (0.04) (0.07) (0.05) (0.13) (0.06) (0.12) (0.08) (0.23) Border −4.60 *** −1.71 −0.17 −5.17 *** 6.21 *** 0.61 0.63 −0.90 (0.87) (1.07) (0.99) (1.69) (1.80) (2.28) (1.68) (2.87) Border*Ln dist. 0.92 *** 0.27 0.07 1.12 *** −0.48 0.40 0.28 0.50 (0.19) (0.29) (0.22) (0.36) (0.42) (0.64) (0.33) (0.60) Border*Ln pop. origin −0.33 *** 0.17 −0.13 0.31 ** −0.18 0.12 0.05 0.31 (0.09) (0.10) (0.09) (0.14) (0.17) (0.23) (0.14) (0.25) Border*Ln pop. dest. −0.18 ** −0.10 −0.25 *** −0.64 *** −0.52 *** −0.69 *** −0.10 −0.20 (0.08) (0.09) (0.09) (0.13) (0.16) (0.22) (0.16) (0.22) Border*Ln area origin 0.01 −0.16 −0.16 −0.46 *** −0.23 −0.44 ** −0.40 *** −0.27 (0.08) (0.10) (0.10) (0.06) (0.17) (0.22) (0.14) (0.21) Border*Ln area dest. −0.04 0.08 0.34 *** 0.31 * 0.22 0.20 0.11 −0.05 (0.09) (0.16) (0.09) (0.17) (0.17) (0.17) (0.14) (0.25) Border*landlocked 0.72 *** 0.65 *** 0.32 ** 0.43 ** 0.86 *** 0.85 *** 0.24 0.47 (0.12) (0.15) (0.12) (0.21) (0.20) (0.31) (0.19) (0.34) Common language 0.67 *** 0.74 *** 0.91 *** 0.93 *** 0.36 *** 0.27 0.65 *** 1.29 *** (0.09) (0.13) (0.10) (0.20) (0.12) (0.21) (0.14) (0.36) Common off. lang. −0.05 −0.03 0.45 *** 0.16 0.47 *** 0.30 0.88 *** 0.11 (0.09) (0.14) (0.10) (0.22) (0.12) (0.22) (0.14) (0.37) Colonial ties 3.16 *** 1.67 *** 1.33 *** 0.95 *** 4.11 *** 2.51 *** 1.24 *** 0.84 * (0.14) (0.14) (0.17) (0.22) (0.18) (0.19) (0.27) (0.44) Origin hegemon −2.29 *** −2.06 *** 0.97 *** 0.46 −3.35 *** −3.07 *** 1.09 *** 0.72 (0.19) (0.21) (0.22) (0.33) (0.29) (0.29) (0.34) (0.62) Time zone diff. 0.13 *** 0.05 *** 0.09 *** −0.03 −0.12 *** −0.00 −0.01 −0.30 *** (0.01) (0.02) (0.01) (0.04) (0.02) (0.04) (0.02) (0.07) Common currency 0.79 *** 0.58 *** 0.99 *** 0.55 ** 0.55 *** −0.19 0.75 *** 0.90 ** (0.12) (0.14) (0.13) (0.25) (0.21) (0.36) (0.21) (0.43) Constant 5.60 *** 0.96 ** −0.79 ** 0.73 −3.72 *** −2.53 *** 0.98 1.49 (0.29) (0.43) (0.33) (0.55) (0.66) (0.92) (0.70) (1.16) Observations 20,980 37,044 8,219 38,612 8,311 16,761 2,426 17,462 R-squared 0.41 0.20 0.44 0.40 0.34 0.19 0.55 0.19 All countries African countries Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.85 *** −1.04 *** −1.35 *** −1.40 *** −0.77 *** −0.66 *** −1.52 *** −1.45 *** (0.04) (0.06) (0.04) (0.08) (0.08) (0.11) (0.08) (0.15) Ln pop. origin 0.04 *** −0.08 ** −0.41 *** −0.35 *** 0.05 * −0.16 *** −0.50 *** −0.47 *** (0.02) (0.04) (0.02) (0.07) (0.03) (0.05) (0.04) (0.12) Ln pop. dest. 1.06 *** 0.74 *** 0.65 *** 0.80 *** 1.20 *** 0.97 *** 0.57 *** 0.52 *** (0.01) (0.03) (0.02) (0.08) (0.03) (0.05) (0.04) (0.11) Ln area origin −0.06 *** 0.04 0.19 *** 0.14 *** −0.00 0.12 *** 0.24 *** 0.14 * (0.01) (0.03) (0.02) (0.03) (0.02) (0.04) (0.03) (0.08) Ln area dest. −0.23 *** −0.08 ** −0.08 *** −0.07 −0.33 *** −0.25 *** −0.13 *** −0.01 (0.01) (0.04) (0.02) (0.05) (0.02) (0.06) (0.03) (0.10) Sum landlocked −0.85 *** −0.55 *** −0.28 *** −0.64 *** −0.96 *** −0.96 *** −0.44 *** −0.79 *** (0.04) (0.07) (0.05) (0.13) (0.06) (0.12) (0.08) (0.23) Border −4.60 *** −1.71 −0.17 −5.17 *** 6.21 *** 0.61 0.63 −0.90 (0.87) (1.07) (0.99) (1.69) (1.80) (2.28) (1.68) (2.87) Border*Ln dist. 0.92 *** 0.27 0.07 1.12 *** −0.48 0.40 0.28 0.50 (0.19) (0.29) (0.22) (0.36) (0.42) (0.64) (0.33) (0.60) Border*Ln pop. origin −0.33 *** 0.17 −0.13 0.31 ** −0.18 0.12 0.05 0.31 (0.09) (0.10) (0.09) (0.14) (0.17) (0.23) (0.14) (0.25) Border*Ln pop. dest. −0.18 ** −0.10 −0.25 *** −0.64 *** −0.52 *** −0.69 *** −0.10 −0.20 (0.08) (0.09) (0.09) (0.13) (0.16) (0.22) (0.16) (0.22) Border*Ln area origin 0.01 −0.16 −0.16 −0.46 *** −0.23 −0.44 ** −0.40 *** −0.27 (0.08) (0.10) (0.10) (0.06) (0.17) (0.22) (0.14) (0.21) Border*Ln area dest. −0.04 0.08 0.34 *** 0.31 * 0.22 0.20 0.11 −0.05 (0.09) (0.16) (0.09) (0.17) (0.17) (0.17) (0.14) (0.25) Border*landlocked 0.72 *** 0.65 *** 0.32 ** 0.43 ** 0.86 *** 0.85 *** 0.24 0.47 (0.12) (0.15) (0.12) (0.21) (0.20) (0.31) (0.19) (0.34) Common language 0.67 *** 0.74 *** 0.91 *** 0.93 *** 0.36 *** 0.27 0.65 *** 1.29 *** (0.09) (0.13) (0.10) (0.20) (0.12) (0.21) (0.14) (0.36) Common off. lang. −0.05 −0.03 0.45 *** 0.16 0.47 *** 0.30 0.88 *** 0.11 (0.09) (0.14) (0.10) (0.22) (0.12) (0.22) (0.14) (0.37) Colonial ties 3.16 *** 1.67 *** 1.33 *** 0.95 *** 4.11 *** 2.51 *** 1.24 *** 0.84 * (0.14) (0.14) (0.17) (0.22) (0.18) (0.19) (0.27) (0.44) Origin hegemon −2.29 *** −2.06 *** 0.97 *** 0.46 −3.35 *** −3.07 *** 1.09 *** 0.72 (0.19) (0.21) (0.22) (0.33) (0.29) (0.29) (0.34) (0.62) Time zone diff. 0.13 *** 0.05 *** 0.09 *** −0.03 −0.12 *** −0.00 −0.01 −0.30 *** (0.01) (0.02) (0.01) (0.04) (0.02) (0.04) (0.02) (0.07) Common currency 0.79 *** 0.58 *** 0.99 *** 0.55 ** 0.55 *** −0.19 0.75 *** 0.90 ** (0.12) (0.14) (0.13) (0.25) (0.21) (0.36) (0.21) (0.43) Constant 5.60 *** 0.96 ** −0.79 ** 0.73 −3.72 *** −2.53 *** 0.98 1.49 (0.29) (0.43) (0.33) (0.55) (0.66) (0.92) (0.70) (1.16) Observations 20,980 37,044 8,219 38,612 8,311 16,761 2,426 17,462 R-squared 0.41 0.20 0.44 0.40 0.34 0.19 0.55 0.19 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Immigration’ reflects migration openness measured by the number of foreign-born living in the country divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large As shown in Table 2, except for some quantitative differences, the two estimators produce broadly similar results about the nature of the relationship between exogenous factors and the endogenous variable. Focusing on the sample as a whole, we found that countries which trade more with each other are those that are geographically closer, speak the same language, have colonial ties and share a common currency. We also note that the intensity of trade between two countries increases with the size of the destination country and time zone differences, but decreases with the size of the country of origin, the surface area of the destination country and landlockedness. The same links are qualitatively observed between the regressors and the intensity of migration across countries. On the whole, our results are in line with those of Ortega and Peri (2014), the main differences being the introduction among regressors of a dummy to capture the sharing of a common currency, the use of the IMF trade database (instead of the NBER-UN dataset) and the sample of countries which is larger in our study. Although these results are qualitatively consistent with expectations, their quantitative interpretation should be done with caution, especially because of the interaction terms that should be considered. For instance, the negative sign associated with the variable reflecting sharing a common border cannot be interpreted as a negative impact of the border because of its interactions with other variables.12 Let us now focus on the specific and interesting case of Africa. Africa being characterised by strong intra-continental migration and emigration in the industrialised world, we first look at migration from the perspective of the destination country (immigration) and consider indiscriminately—as Ortega and Peri (2014)—the partners of Africa. Then, we go a step further with the aim of identifying a possible differentiated impact of openness depending on the partner. In this case, our variable of interest becomes immigration in the context of intra-continental openness and more generally for openness with developing countries. Regarding the relations with industrialised countries, our variable of interest is rather emigration from the perspective of Africa. As shown in Table 2 (columns (5–8)), we found as for the whole sample that geographical, cultural and historical factors largely explain the intensity of bilateral trade. Qualitatively, the results are very close to those obtained for the whole sample with the only difference that in the latter case, the impact of time zone differences is negative. This is not surprising given that the time zone effect depends on the countries and, especially, the type of services (Head et al., 2009). Table 3 reports the results concerning openness between African countries themselves (columns (1–4)), and between Africa and developing countries (columns (5–8)). Overall, these results are consistent with our previous findings. They confirm that a country has more links (trade and migration) with those that are closer geographically and culturally, more populated, with a coastline or with those with which it shares the same currency. These results also corroborate the fact that trade and migration are both explained by the same factors, as pointed out by Ortega and Peri (2014). In most cases, the explanatory power of migration models is higher than that of trade models. Table 3: Gravity Regression, African Openness with Developing Countries Intra-Africa With no industrialised Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.30 *** −1.25 *** −1.50 *** −1.30 *** −1.17 *** −1.19 *** −1.71 *** −1.50 *** (0.17) (0.15) (0.17) (0.34) (0.08) (0.11) (0.07) (0.12) Ln pop. origin −0.01 0.22 −0.88 *** −0.85 *** 0.15 *** −0.07 −0.40 *** −0.47 *** (0.09) (0.15) (0.08) (0.25) (0.03) (0.06) (0.04) (0.11) Ln pop. dest. 0.85 *** 0.90 *** 0.22 ** 0.15 1.09 *** 0.97 *** 0.59 *** 0.93 *** (0.08) (0.09) (0.10) (0.15) (0.03) (0.08) (0.04) (0.12) Ln area origin −0.17 ** −0.22 0.02 0.34 ** −0.08 *** 0.06 0.03 0.17 *** (0.07) (0.16) (0.07) (0.17) (0.02) (0.05) (0.03) (0.05) Ln area dest. −0.15 ** 0.12 0.02 0.04 −0.29 *** −0.22 *** −0.04 −0.03 (0.07) (0.16) (0.07) (0.17) (0.02) (0.08) (0.03) (0.08) Sum landlocked −0.94 *** −0.50 * −0.07 −0.58 ** −0.98 *** −0.85 *** −0.19 *** −0.78 *** (0.13) (0.28) (0.13) (0.28) (0.06) (0.20) (0.07) (0.20) Border 0.19 −6.28 *** −1.25 −1.02 2.94 −3.19 −3.63 *** −3.80 ** (2.13) (2.19) (1.83) (3.41) (1.80) (2.04) (1.12) (1.69) Border*Ln dist. 0.41 1.54 *** 0.18 0.54 −0.08 0.91 0.51 ** 0.99 *** (0.46) (0.53) (0.36) (0.69) (0.43) (0.56) (0.24) (0.37) Border*Ln pop. origin −0.04 −0.32 0.38 ** 0.77 ** −0.26 −0.17 −0.16 0.23 (0.20) (0.22) (0.15) (0.35) (0.17) (0.18) (0.10) (0.16) Border*Ln pop. dest. −0.05 −0.21 0.27 0.22 −0.37 ** −0.31 −0.21 ** −0.68 *** (0.18) (0.19) (0.17) (0.24) (0.16) (0.23) (0.10) (0.15) Border*Ln area origin −0.18 −0.12 −0.04 −0.47 * −0.14 −0.21 −0.00 −0.43 *** (0.20) (0.26) (0.15) (0.28) (0.17) (0.19) (0.11) (0.06) Border*Ln area dest. −0.09 −0.30 −0.06 −0.11 0.17 0.03 0.30 *** 0.26 (0.19) (0.22) (0.15) (0.29) (0.17) (0.17) (0.10) (0.20) Border*landlocked 0.74 *** 0.34 −0.17 0.20 0.86 *** 0.65 * 0.28 * 0.46 * (0.24) (0.34) (0.21) (0.35) (0.21) (0.34) (0.14) (0.27) Common language 0.50 *** 0.65 ** 0.37 ** 1.07 *** 0.43 *** 0.03 0.94 *** 0.93 *** (0.19) (0.27) (0.16) (0.39) (0.12) (0.27) (0.14) (0.25) Common off. lang. 0.72 *** 0.33 0.01 −0.39 0.36 *** 0.51 * 0.39 *** 0.03 (0.19) (0.23) (0.17) (0.40) (0.12) (0.30) (0.14) (0.25) Colonial ties 0.62 −2.41 ** −1.39 −0.69 1.92 *** −0.76 0.92 ** 0.73 * (0.44) (0.99) (1.06) (0.67) (0.51) (0.71) (0.45) (0.41) Origin hegemon −1.41 *** −0.45 2.19 * 0.15 −0.81 0.61 0.71 −0.08 (0.39) (1.12) (1.24) (1.01) (0.56) (0.77) (0.64) (0.65) Time zone diff. −0.43 *** −0.62 *** 0.02 −0.67 *** −0.03 ** 0.09 *** 0.05 ** −0.20 *** (0.09) (0.20) (0.09) (0.21) (0.02) (0.03) (0.02) (0.07) Common currency 0.37 −0.24 0.77 *** 1.08 ** 0.74 *** −0.05 1.30 *** 0.90 *** (0.24) (0.34) (0.20) (0.44) (0.21) (0.37) (0.21) (0.28) Constant 0.94 0.79 3.61 *** 0.90 −0.65 0.94 2.25 *** 0.89 (1.28) (1.14) (1.27) (1.98) (0.67) (0.84) (0.51) (0.78) Observations 1,331 2,450 525 2,550 7,366 15,634 3,760 30,102 R-squared 0.41 0.19 0.67 0.25 0.34 0.09 0.53 0.54 Intra-Africa With no industrialised Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.30 *** −1.25 *** −1.50 *** −1.30 *** −1.17 *** −1.19 *** −1.71 *** −1.50 *** (0.17) (0.15) (0.17) (0.34) (0.08) (0.11) (0.07) (0.12) Ln pop. origin −0.01 0.22 −0.88 *** −0.85 *** 0.15 *** −0.07 −0.40 *** −0.47 *** (0.09) (0.15) (0.08) (0.25) (0.03) (0.06) (0.04) (0.11) Ln pop. dest. 0.85 *** 0.90 *** 0.22 ** 0.15 1.09 *** 0.97 *** 0.59 *** 0.93 *** (0.08) (0.09) (0.10) (0.15) (0.03) (0.08) (0.04) (0.12) Ln area origin −0.17 ** −0.22 0.02 0.34 ** −0.08 *** 0.06 0.03 0.17 *** (0.07) (0.16) (0.07) (0.17) (0.02) (0.05) (0.03) (0.05) Ln area dest. −0.15 ** 0.12 0.02 0.04 −0.29 *** −0.22 *** −0.04 −0.03 (0.07) (0.16) (0.07) (0.17) (0.02) (0.08) (0.03) (0.08) Sum landlocked −0.94 *** −0.50 * −0.07 −0.58 ** −0.98 *** −0.85 *** −0.19 *** −0.78 *** (0.13) (0.28) (0.13) (0.28) (0.06) (0.20) (0.07) (0.20) Border 0.19 −6.28 *** −1.25 −1.02 2.94 −3.19 −3.63 *** −3.80 ** (2.13) (2.19) (1.83) (3.41) (1.80) (2.04) (1.12) (1.69) Border*Ln dist. 0.41 1.54 *** 0.18 0.54 −0.08 0.91 0.51 ** 0.99 *** (0.46) (0.53) (0.36) (0.69) (0.43) (0.56) (0.24) (0.37) Border*Ln pop. origin −0.04 −0.32 0.38 ** 0.77 ** −0.26 −0.17 −0.16 0.23 (0.20) (0.22) (0.15) (0.35) (0.17) (0.18) (0.10) (0.16) Border*Ln pop. dest. −0.05 −0.21 0.27 0.22 −0.37 ** −0.31 −0.21 ** −0.68 *** (0.18) (0.19) (0.17) (0.24) (0.16) (0.23) (0.10) (0.15) Border*Ln area origin −0.18 −0.12 −0.04 −0.47 * −0.14 −0.21 −0.00 −0.43 *** (0.20) (0.26) (0.15) (0.28) (0.17) (0.19) (0.11) (0.06) Border*Ln area dest. −0.09 −0.30 −0.06 −0.11 0.17 0.03 0.30 *** 0.26 (0.19) (0.22) (0.15) (0.29) (0.17) (0.17) (0.10) (0.20) Border*landlocked 0.74 *** 0.34 −0.17 0.20 0.86 *** 0.65 * 0.28 * 0.46 * (0.24) (0.34) (0.21) (0.35) (0.21) (0.34) (0.14) (0.27) Common language 0.50 *** 0.65 ** 0.37 ** 1.07 *** 0.43 *** 0.03 0.94 *** 0.93 *** (0.19) (0.27) (0.16) (0.39) (0.12) (0.27) (0.14) (0.25) Common off. lang. 0.72 *** 0.33 0.01 −0.39 0.36 *** 0.51 * 0.39 *** 0.03 (0.19) (0.23) (0.17) (0.40) (0.12) (0.30) (0.14) (0.25) Colonial ties 0.62 −2.41 ** −1.39 −0.69 1.92 *** −0.76 0.92 ** 0.73 * (0.44) (0.99) (1.06) (0.67) (0.51) (0.71) (0.45) (0.41) Origin hegemon −1.41 *** −0.45 2.19 * 0.15 −0.81 0.61 0.71 −0.08 (0.39) (1.12) (1.24) (1.01) (0.56) (0.77) (0.64) (0.65) Time zone diff. −0.43 *** −0.62 *** 0.02 −0.67 *** −0.03 ** 0.09 *** 0.05 ** −0.20 *** (0.09) (0.20) (0.09) (0.21) (0.02) (0.03) (0.02) (0.07) Common currency 0.37 −0.24 0.77 *** 1.08 ** 0.74 *** −0.05 1.30 *** 0.90 *** (0.24) (0.34) (0.20) (0.44) (0.21) (0.37) (0.21) (0.28) Constant 0.94 0.79 3.61 *** 0.90 −0.65 0.94 2.25 *** 0.89 (1.28) (1.14) (1.27) (1.98) (0.67) (0.84) (0.51) (0.78) Observations 1,331 2,450 525 2,550 7,366 15,634 3,760 30,102 R-squared 0.41 0.19 0.67 0.25 0.34 0.09 0.53 0.54 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Immigration’ reflects migration openness measured by the number of foreign-born living in the country divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large Table 3: Gravity Regression, African Openness with Developing Countries Intra-Africa With no industrialised Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.30 *** −1.25 *** −1.50 *** −1.30 *** −1.17 *** −1.19 *** −1.71 *** −1.50 *** (0.17) (0.15) (0.17) (0.34) (0.08) (0.11) (0.07) (0.12) Ln pop. origin −0.01 0.22 −0.88 *** −0.85 *** 0.15 *** −0.07 −0.40 *** −0.47 *** (0.09) (0.15) (0.08) (0.25) (0.03) (0.06) (0.04) (0.11) Ln pop. dest. 0.85 *** 0.90 *** 0.22 ** 0.15 1.09 *** 0.97 *** 0.59 *** 0.93 *** (0.08) (0.09) (0.10) (0.15) (0.03) (0.08) (0.04) (0.12) Ln area origin −0.17 ** −0.22 0.02 0.34 ** −0.08 *** 0.06 0.03 0.17 *** (0.07) (0.16) (0.07) (0.17) (0.02) (0.05) (0.03) (0.05) Ln area dest. −0.15 ** 0.12 0.02 0.04 −0.29 *** −0.22 *** −0.04 −0.03 (0.07) (0.16) (0.07) (0.17) (0.02) (0.08) (0.03) (0.08) Sum landlocked −0.94 *** −0.50 * −0.07 −0.58 ** −0.98 *** −0.85 *** −0.19 *** −0.78 *** (0.13) (0.28) (0.13) (0.28) (0.06) (0.20) (0.07) (0.20) Border 0.19 −6.28 *** −1.25 −1.02 2.94 −3.19 −3.63 *** −3.80 ** (2.13) (2.19) (1.83) (3.41) (1.80) (2.04) (1.12) (1.69) Border*Ln dist. 0.41 1.54 *** 0.18 0.54 −0.08 0.91 0.51 ** 0.99 *** (0.46) (0.53) (0.36) (0.69) (0.43) (0.56) (0.24) (0.37) Border*Ln pop. origin −0.04 −0.32 0.38 ** 0.77 ** −0.26 −0.17 −0.16 0.23 (0.20) (0.22) (0.15) (0.35) (0.17) (0.18) (0.10) (0.16) Border*Ln pop. dest. −0.05 −0.21 0.27 0.22 −0.37 ** −0.31 −0.21 ** −0.68 *** (0.18) (0.19) (0.17) (0.24) (0.16) (0.23) (0.10) (0.15) Border*Ln area origin −0.18 −0.12 −0.04 −0.47 * −0.14 −0.21 −0.00 −0.43 *** (0.20) (0.26) (0.15) (0.28) (0.17) (0.19) (0.11) (0.06) Border*Ln area dest. −0.09 −0.30 −0.06 −0.11 0.17 0.03 0.30 *** 0.26 (0.19) (0.22) (0.15) (0.29) (0.17) (0.17) (0.10) (0.20) Border*landlocked 0.74 *** 0.34 −0.17 0.20 0.86 *** 0.65 * 0.28 * 0.46 * (0.24) (0.34) (0.21) (0.35) (0.21) (0.34) (0.14) (0.27) Common language 0.50 *** 0.65 ** 0.37 ** 1.07 *** 0.43 *** 0.03 0.94 *** 0.93 *** (0.19) (0.27) (0.16) (0.39) (0.12) (0.27) (0.14) (0.25) Common off. lang. 0.72 *** 0.33 0.01 −0.39 0.36 *** 0.51 * 0.39 *** 0.03 (0.19) (0.23) (0.17) (0.40) (0.12) (0.30) (0.14) (0.25) Colonial ties 0.62 −2.41 ** −1.39 −0.69 1.92 *** −0.76 0.92 ** 0.73 * (0.44) (0.99) (1.06) (0.67) (0.51) (0.71) (0.45) (0.41) Origin hegemon −1.41 *** −0.45 2.19 * 0.15 −0.81 0.61 0.71 −0.08 (0.39) (1.12) (1.24) (1.01) (0.56) (0.77) (0.64) (0.65) Time zone diff. −0.43 *** −0.62 *** 0.02 −0.67 *** −0.03 ** 0.09 *** 0.05 ** −0.20 *** (0.09) (0.20) (0.09) (0.21) (0.02) (0.03) (0.02) (0.07) Common currency 0.37 −0.24 0.77 *** 1.08 ** 0.74 *** −0.05 1.30 *** 0.90 *** (0.24) (0.34) (0.20) (0.44) (0.21) (0.37) (0.21) (0.28) Constant 0.94 0.79 3.61 *** 0.90 −0.65 0.94 2.25 *** 0.89 (1.28) (1.14) (1.27) (1.98) (0.67) (0.84) (0.51) (0.78) Observations 1,331 2,450 525 2,550 7,366 15,634 3,760 30,102 R-squared 0.41 0.19 0.67 0.25 0.34 0.09 0.53 0.54 Intra-Africa With no industrialised Trade Immigration Trade Immigration OLS PPML OLS PPML OLS PPML OLS PPML (1) (2) (3) (4) (5) (6) (7) (8) Ln distance −1.30 *** −1.25 *** −1.50 *** −1.30 *** −1.17 *** −1.19 *** −1.71 *** −1.50 *** (0.17) (0.15) (0.17) (0.34) (0.08) (0.11) (0.07) (0.12) Ln pop. origin −0.01 0.22 −0.88 *** −0.85 *** 0.15 *** −0.07 −0.40 *** −0.47 *** (0.09) (0.15) (0.08) (0.25) (0.03) (0.06) (0.04) (0.11) Ln pop. dest. 0.85 *** 0.90 *** 0.22 ** 0.15 1.09 *** 0.97 *** 0.59 *** 0.93 *** (0.08) (0.09) (0.10) (0.15) (0.03) (0.08) (0.04) (0.12) Ln area origin −0.17 ** −0.22 0.02 0.34 ** −0.08 *** 0.06 0.03 0.17 *** (0.07) (0.16) (0.07) (0.17) (0.02) (0.05) (0.03) (0.05) Ln area dest. −0.15 ** 0.12 0.02 0.04 −0.29 *** −0.22 *** −0.04 −0.03 (0.07) (0.16) (0.07) (0.17) (0.02) (0.08) (0.03) (0.08) Sum landlocked −0.94 *** −0.50 * −0.07 −0.58 ** −0.98 *** −0.85 *** −0.19 *** −0.78 *** (0.13) (0.28) (0.13) (0.28) (0.06) (0.20) (0.07) (0.20) Border 0.19 −6.28 *** −1.25 −1.02 2.94 −3.19 −3.63 *** −3.80 ** (2.13) (2.19) (1.83) (3.41) (1.80) (2.04) (1.12) (1.69) Border*Ln dist. 0.41 1.54 *** 0.18 0.54 −0.08 0.91 0.51 ** 0.99 *** (0.46) (0.53) (0.36) (0.69) (0.43) (0.56) (0.24) (0.37) Border*Ln pop. origin −0.04 −0.32 0.38 ** 0.77 ** −0.26 −0.17 −0.16 0.23 (0.20) (0.22) (0.15) (0.35) (0.17) (0.18) (0.10) (0.16) Border*Ln pop. dest. −0.05 −0.21 0.27 0.22 −0.37 ** −0.31 −0.21 ** −0.68 *** (0.18) (0.19) (0.17) (0.24) (0.16) (0.23) (0.10) (0.15) Border*Ln area origin −0.18 −0.12 −0.04 −0.47 * −0.14 −0.21 −0.00 −0.43 *** (0.20) (0.26) (0.15) (0.28) (0.17) (0.19) (0.11) (0.06) Border*Ln area dest. −0.09 −0.30 −0.06 −0.11 0.17 0.03 0.30 *** 0.26 (0.19) (0.22) (0.15) (0.29) (0.17) (0.17) (0.10) (0.20) Border*landlocked 0.74 *** 0.34 −0.17 0.20 0.86 *** 0.65 * 0.28 * 0.46 * (0.24) (0.34) (0.21) (0.35) (0.21) (0.34) (0.14) (0.27) Common language 0.50 *** 0.65 ** 0.37 ** 1.07 *** 0.43 *** 0.03 0.94 *** 0.93 *** (0.19) (0.27) (0.16) (0.39) (0.12) (0.27) (0.14) (0.25) Common off. lang. 0.72 *** 0.33 0.01 −0.39 0.36 *** 0.51 * 0.39 *** 0.03 (0.19) (0.23) (0.17) (0.40) (0.12) (0.30) (0.14) (0.25) Colonial ties 0.62 −2.41 ** −1.39 −0.69 1.92 *** −0.76 0.92 ** 0.73 * (0.44) (0.99) (1.06) (0.67) (0.51) (0.71) (0.45) (0.41) Origin hegemon −1.41 *** −0.45 2.19 * 0.15 −0.81 0.61 0.71 −0.08 (0.39) (1.12) (1.24) (1.01) (0.56) (0.77) (0.64) (0.65) Time zone diff. −0.43 *** −0.62 *** 0.02 −0.67 *** −0.03 ** 0.09 *** 0.05 ** −0.20 *** (0.09) (0.20) (0.09) (0.21) (0.02) (0.03) (0.02) (0.07) Common currency 0.37 −0.24 0.77 *** 1.08 ** 0.74 *** −0.05 1.30 *** 0.90 *** (0.24) (0.34) (0.20) (0.44) (0.21) (0.37) (0.21) (0.28) Constant 0.94 0.79 3.61 *** 0.90 −0.65 0.94 2.25 *** 0.89 (1.28) (1.14) (1.27) (1.98) (0.67) (0.84) (0.51) (0.78) Observations 1,331 2,450 525 2,550 7,366 15,634 3,760 30,102 R-squared 0.41 0.19 0.67 0.25 0.34 0.09 0.53 0.54 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Immigration’ reflects migration openness measured by the number of foreign-born living in the country divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large Finally, Table 4 provides the estimation results concerning openness between Africa and the industrialised countries. Table 4: Gravity Regression, African Openness with Industrialised Countries Trade Emigration OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.88 *** −0.78 *** −1.04 *** −1.96 *** (0.13) (0.13) (0.15) (0.22) Ln pop. origin −0.13 ** −0.28 *** −0.12 ** −0.05 (0.05) (0.08) (0.05) (0.11) Ln pop. dest. 1.50 *** 1.11 *** 0.77 *** 0.33 *** (0.05) (0.07) (0.05) (0.12) Ln area origin 0.01 0.11 ** −0.22 *** −0.46 *** (0.04) (0.05) (0.04) (0.08) Ln area dest. −0.36 *** −0.28 *** −0.05 0.20 (0.04) (0.08) (0.05) (0.15) Sum landlocked −1.07 *** −0.85 *** −0.93 *** −1.05 *** (0.11) (0.13) (0.11) (0.23) Common language 0.83 *** 0.51 0.45 ** 0.49 * (0.26) (0.31) (0.20) (0.27) Common off. lang. 0.06 0.13 1.40 *** 0.77 * (0.24) (0.29) (0.19) (0.40) Colonial ties 2.46 *** 1.41 *** 1.64 *** 1.40 *** (0.20) (0.23) (0.20) (0.33) Origin hegemon −2.59 *** −2.68 *** (0.31) (0.32) Time zone diff. −0.10 *** 0.04 * 0.04 0.10 * (0.03) (0.03) (0.03) (0.06) Constant −1.61 −0.56 −1.95 8.22 *** (1.13) (1.39) (1.36) (1.95) Observations 2,036 2,300 898 1,173 R-squared 0.49 0.35 0.59 0.46 Trade Emigration OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.88 *** −0.78 *** −1.04 *** −1.96 *** (0.13) (0.13) (0.15) (0.22) Ln pop. origin −0.13 ** −0.28 *** −0.12 ** −0.05 (0.05) (0.08) (0.05) (0.11) Ln pop. dest. 1.50 *** 1.11 *** 0.77 *** 0.33 *** (0.05) (0.07) (0.05) (0.12) Ln area origin 0.01 0.11 ** −0.22 *** −0.46 *** (0.04) (0.05) (0.04) (0.08) Ln area dest. −0.36 *** −0.28 *** −0.05 0.20 (0.04) (0.08) (0.05) (0.15) Sum landlocked −1.07 *** −0.85 *** −0.93 *** −1.05 *** (0.11) (0.13) (0.11) (0.23) Common language 0.83 *** 0.51 0.45 ** 0.49 * (0.26) (0.31) (0.20) (0.27) Common off. lang. 0.06 0.13 1.40 *** 0.77 * (0.24) (0.29) (0.19) (0.40) Colonial ties 2.46 *** 1.41 *** 1.64 *** 1.40 *** (0.20) (0.23) (0.20) (0.33) Origin hegemon −2.59 *** −2.68 *** (0.31) (0.32) Time zone diff. −0.10 *** 0.04 * 0.04 0.10 * (0.03) (0.03) (0.03) (0.06) Constant −1.61 −0.56 −1.95 8.22 *** (1.13) (1.39) (1.36) (1.95) Observations 2,036 2,300 898 1,173 R-squared 0.49 0.35 0.59 0.46 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Emigration’ reflects migration openness measured by the number of nationals living abroad divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large Table 4: Gravity Regression, African Openness with Industrialised Countries Trade Emigration OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.88 *** −0.78 *** −1.04 *** −1.96 *** (0.13) (0.13) (0.15) (0.22) Ln pop. origin −0.13 ** −0.28 *** −0.12 ** −0.05 (0.05) (0.08) (0.05) (0.11) Ln pop. dest. 1.50 *** 1.11 *** 0.77 *** 0.33 *** (0.05) (0.07) (0.05) (0.12) Ln area origin 0.01 0.11 ** −0.22 *** −0.46 *** (0.04) (0.05) (0.04) (0.08) Ln area dest. −0.36 *** −0.28 *** −0.05 0.20 (0.04) (0.08) (0.05) (0.15) Sum landlocked −1.07 *** −0.85 *** −0.93 *** −1.05 *** (0.11) (0.13) (0.11) (0.23) Common language 0.83 *** 0.51 0.45 ** 0.49 * (0.26) (0.31) (0.20) (0.27) Common off. lang. 0.06 0.13 1.40 *** 0.77 * (0.24) (0.29) (0.19) (0.40) Colonial ties 2.46 *** 1.41 *** 1.64 *** 1.40 *** (0.20) (0.23) (0.20) (0.33) Origin hegemon −2.59 *** −2.68 *** (0.31) (0.32) Time zone diff. −0.10 *** 0.04 * 0.04 0.10 * (0.03) (0.03) (0.03) (0.06) Constant −1.61 −0.56 −1.95 8.22 *** (1.13) (1.39) (1.36) (1.95) Observations 2,036 2,300 898 1,173 R-squared 0.49 0.35 0.59 0.46 Trade Emigration OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.88 *** −0.78 *** −1.04 *** −1.96 *** (0.13) (0.13) (0.15) (0.22) Ln pop. origin −0.13 ** −0.28 *** −0.12 ** −0.05 (0.05) (0.08) (0.05) (0.11) Ln pop. dest. 1.50 *** 1.11 *** 0.77 *** 0.33 *** (0.05) (0.07) (0.05) (0.12) Ln area origin 0.01 0.11 ** −0.22 *** −0.46 *** (0.04) (0.05) (0.04) (0.08) Ln area dest. −0.36 *** −0.28 *** −0.05 0.20 (0.04) (0.08) (0.05) (0.15) Sum landlocked −1.07 *** −0.85 *** −0.93 *** −1.05 *** (0.11) (0.13) (0.11) (0.23) Common language 0.83 *** 0.51 0.45 ** 0.49 * (0.26) (0.31) (0.20) (0.27) Common off. lang. 0.06 0.13 1.40 *** 0.77 * (0.24) (0.29) (0.19) (0.40) Colonial ties 2.46 *** 1.41 *** 1.64 *** 1.40 *** (0.20) (0.23) (0.20) (0.33) Origin hegemon −2.59 *** −2.68 *** (0.31) (0.32) Time zone diff. −0.10 *** 0.04 * 0.04 0.10 * (0.03) (0.03) (0.03) (0.06) Constant −1.61 −0.56 −1.95 8.22 *** (1.13) (1.39) (1.36) (1.95) Observations 2,036 2,300 898 1,173 R-squared 0.49 0.35 0.59 0.46 Notes: The dependent variable ‘Trade’ refers to trade openness measured by the sum of bilateral exports and imports divided by GDP. The dependent variable ‘Emigration’ reflects migration openness measured by the number of nationals living abroad divided by the total population. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large These results are also consistent with theoretical predictions. In particular, geographical, cultural and historical factors explain well bilateral emigration from African countries to industrialised economies. It should be noted that, as Africa has no common border with industrialised countries in our sample, the dummy variable for common border and its interactions with other variables do not appear in the results. For the sake of completeness and as a robustness check, we also investigate the relationship between actual and constructed openness. This link is displayed in the corresponding scatterplots reproduced in Figures 3 and 4. Figure 3 depicts overall openness against its gravity-predicted value for the world and for African openness with the world, based on both OLS and non-linear (Poisson) estimations. For both trade and migration, observed openness is highly correlated with its predicted measure, particularly for the African sample. Figure 413 displays the same relationship for the African sample depending on partners: intra-African trade, African trade with low-income countries and with advanced countries, intra-African migration, immigration in Africa from low-income countries including Africa, and emigration from Africa to advanced countries. For each type of disaggregated openness of Africa, there is high correlation between observed openness and its predicted value. In other words, the figure shows that geographic, cultural and historical variables account for the major part of variation in Africa’s openness. On the whole, our gravity-predicted values for openness thus appear to be reasonable proxies for observed values. Figure 3: View largeDownload slide Observed Openness and Predicted Openness. Notes: This figure reports the scatterplots of the relationship between actual and predicted values of trade (left side) and migration (right side). Figure 3: View largeDownload slide Observed Openness and Predicted Openness. Notes: This figure reports the scatterplots of the relationship between actual and predicted values of trade (left side) and migration (right side). Figure 4: View largeDownload slide Observed Openness and Predicted Openness, depending on Partners. Notes: This figure reports the scatterplots of the relationship between actual and predicted values of trade (left side) and migration (right side) for the subsample of African countries. Figure 4: View largeDownload slide Observed Openness and Predicted Openness, depending on Partners. Notes: This figure reports the scatterplots of the relationship between actual and predicted values of trade (left side) and migration (right side) for the subsample of African countries. 4.2 Income and openness Scatterplots of income per capita against openness to trade and to migration reproduced in Figure 5 provide a first insight about the relationship between income per capita and openness. As shown, for the world sample, there is a positive correlation between income per capita and both trade and migration openness. This positive relationship is also present for Africa overall openness. However, at the disaggregated level, the link between income and openness is found to be related to the type of partners. Indeed, while the relationship between income per person and openness is not positive for intra-African trade and is slightly positive for trade with low-income countries (including Africa), income per capita is positively correlated with trade with advanced countries, intra-African migration, migration from low-income countries and emigration to advanced economies. The next steps give econometric estimations of this openness-growth nexus. Figure 5: View largeDownload slide Income and Openness. Notes: This figure reports the scatterplots of income per capita against openness to trade (left side) and to migration (right side). Figure 5: View largeDownload slide Income and Openness. Notes: This figure reports the scatterplots of income per capita against openness to trade (left side) and to migration (right side). 4.2.1 Baseline results Tables 5 and 6 report the two-stage least-squares (2SLS) joint estimates of the impact of openness to trade and to migration on income per person, using their gravity-predicted measures as instruments. Table 5 displays the results using overall measures of openness for the world and African samples. To clearly highlight our contribution, we start by replicating the results of Ortega and Peri (2014) in columns (1–4). In columns (1) and (2) ((3) and (4)), the estimations are based on linear (non-linear, PPML) gravity-predicted openness. Estimations reported in columns (1) and (3) consider as control variables country size (area and population) and distance to equator—which is the key geographic control identified in the literature (Rodriguez and Rodrik, 2000). In column (1) based on linear predicted openness, the coefficient of openness to immigration is significantly positive while the coefficient of openness to trade is not significant. In column (2), using also linear predicted openness and controlling for other geographic/climate and colonial factors, the impact of immigration increases in level and in significance, while the coefficient of trade remains non significant. Using non-linear predicted openness in columns (3) and (4) as instruments, the impact of immigration remains significantly positive when we consider a comprehensive set of control variables. To sum up, results in columns (1–4) of Table 5 highlight a robust, positive effect of openness to immigration on long-run income per capita at the world level, while there is no evidence of growth-enhancing impact of openness to trade. This finding confirms the results obtained by Ortega and Peri (2014). Table 5: Income and Openness, Baseline Specification World openness African openness LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) Trade −1.80 0.44 1.50 * 0.88 2.80 ** 1.60 3.89 * 4.60 (2.18) (0.59) (0.79) (0.70) (1.37) (1.05) (2.00) (4.72) Immig. 7.72 ** 11.95 *** 1.18 6.73 ** 1.13 11.85 * −4.14 24.13 (3.65) (2.59) (1.06) (3.18) (7.72) (6.05) (11.61) (23.83) Ln pop. 0.03 −0.02 −0.04 −0.06 −0.24 −0.21 −0.26 −0.10 (0.12) (0.06) (0.07) (0.06) (0.18) (0.14) (0.22) (0.32) Ln area −0.21 * 0.25 ** −0.10 0.23 ** 0.05 0.34 * 0.06 0.57 (0.11) (0.11) (0.06) (0.09) (0.11) (0.18) (0.14) (0.56) Dist. equator 0.05 *** 0.01 0.04 *** 0.00 0.05 *** 0.00 0.05 *** 0.01 (0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.06) Constant 8.84 *** 6.98 *** 7.16 *** 7.32 *** 5.53 *** 6.93 *** 5.27 *** 5.47 ** (1.21) (0.78) (0.50) (0.74) (0.49) (0.85) (0.81) (2.20) Observations 187 132 187 132 52 45 52 45 Colonial controls No Yes No Yes No Yes No Yes Geo/climate controls No Yes No Yes No Yes No Yes K-P F-stat 0.887 5.842 5.296 5.726 0.812 1.521 0.772 0.285 SW F-stat for Trade 2.062 13.15 10.97 16.99 8.577 16.30 3.279 0.680 SW F-stat for Mig. 5.635 16.59 55.29 8.622 3.176 3.248 2.413 0.698 SY 10% max IV size 7.030 7.030 7.030 7.030 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 3.630 3.630 3.630 3.630 World openness African openness LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) Trade −1.80 0.44 1.50 * 0.88 2.80 ** 1.60 3.89 * 4.60 (2.18) (0.59) (0.79) (0.70) (1.37) (1.05) (2.00) (4.72) Immig. 7.72 ** 11.95 *** 1.18 6.73 ** 1.13 11.85 * −4.14 24.13 (3.65) (2.59) (1.06) (3.18) (7.72) (6.05) (11.61) (23.83) Ln pop. 0.03 −0.02 −0.04 −0.06 −0.24 −0.21 −0.26 −0.10 (0.12) (0.06) (0.07) (0.06) (0.18) (0.14) (0.22) (0.32) Ln area −0.21 * 0.25 ** −0.10 0.23 ** 0.05 0.34 * 0.06 0.57 (0.11) (0.11) (0.06) (0.09) (0.11) (0.18) (0.14) (0.56) Dist. equator 0.05 *** 0.01 0.04 *** 0.00 0.05 *** 0.00 0.05 *** 0.01 (0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.06) Constant 8.84 *** 6.98 *** 7.16 *** 7.32 *** 5.53 *** 6.93 *** 5.27 *** 5.47 ** (1.21) (0.78) (0.50) (0.74) (0.49) (0.85) (0.81) (2.20) Observations 187 132 187 132 52 45 52 45 Colonial controls No Yes No Yes No Yes No Yes Geo/climate controls No Yes No Yes No Yes No Yes K-P F-stat 0.887 5.842 5.296 5.726 0.812 1.521 0.772 0.285 SW F-stat for Trade 2.062 13.15 10.97 16.99 8.577 16.30 3.279 0.680 SW F-stat for Mig. 5.635 16.59 55.29 8.622 3.176 3.248 2.413 0.698 SY 10% max IV size 7.030 7.030 7.030 7.030 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 3.630 3.630 3.630 3.630 Notes: The dependent variable is the log of income per capita. LP (NLP) stands for linear predicted trade and migration based on the OLS (non-linear Poisson, PPML) gravity estimates. Geographic, climate and disease controls are regional dummies for Africa, a landlocked dummy, the percentage of land in the tropics, average distance to the coast and a measure for oil reserves. Colonial history controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 5: Income and Openness, Baseline Specification World openness African openness LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) Trade −1.80 0.44 1.50 * 0.88 2.80 ** 1.60 3.89 * 4.60 (2.18) (0.59) (0.79) (0.70) (1.37) (1.05) (2.00) (4.72) Immig. 7.72 ** 11.95 *** 1.18 6.73 ** 1.13 11.85 * −4.14 24.13 (3.65) (2.59) (1.06) (3.18) (7.72) (6.05) (11.61) (23.83) Ln pop. 0.03 −0.02 −0.04 −0.06 −0.24 −0.21 −0.26 −0.10 (0.12) (0.06) (0.07) (0.06) (0.18) (0.14) (0.22) (0.32) Ln area −0.21 * 0.25 ** −0.10 0.23 ** 0.05 0.34 * 0.06 0.57 (0.11) (0.11) (0.06) (0.09) (0.11) (0.18) (0.14) (0.56) Dist. equator 0.05 *** 0.01 0.04 *** 0.00 0.05 *** 0.00 0.05 *** 0.01 (0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.06) Constant 8.84 *** 6.98 *** 7.16 *** 7.32 *** 5.53 *** 6.93 *** 5.27 *** 5.47 ** (1.21) (0.78) (0.50) (0.74) (0.49) (0.85) (0.81) (2.20) Observations 187 132 187 132 52 45 52 45 Colonial controls No Yes No Yes No Yes No Yes Geo/climate controls No Yes No Yes No Yes No Yes K-P F-stat 0.887 5.842 5.296 5.726 0.812 1.521 0.772 0.285 SW F-stat for Trade 2.062 13.15 10.97 16.99 8.577 16.30 3.279 0.680 SW F-stat for Mig. 5.635 16.59 55.29 8.622 3.176 3.248 2.413 0.698 SY 10% max IV size 7.030 7.030 7.030 7.030 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 3.630 3.630 3.630 3.630 World openness African openness LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) Trade −1.80 0.44 1.50 * 0.88 2.80 ** 1.60 3.89 * 4.60 (2.18) (0.59) (0.79) (0.70) (1.37) (1.05) (2.00) (4.72) Immig. 7.72 ** 11.95 *** 1.18 6.73 ** 1.13 11.85 * −4.14 24.13 (3.65) (2.59) (1.06) (3.18) (7.72) (6.05) (11.61) (23.83) Ln pop. 0.03 −0.02 −0.04 −0.06 −0.24 −0.21 −0.26 −0.10 (0.12) (0.06) (0.07) (0.06) (0.18) (0.14) (0.22) (0.32) Ln area −0.21 * 0.25 ** −0.10 0.23 ** 0.05 0.34 * 0.06 0.57 (0.11) (0.11) (0.06) (0.09) (0.11) (0.18) (0.14) (0.56) Dist. equator 0.05 *** 0.01 0.04 *** 0.00 0.05 *** 0.00 0.05 *** 0.01 (0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.06) Constant 8.84 *** 6.98 *** 7.16 *** 7.32 *** 5.53 *** 6.93 *** 5.27 *** 5.47 ** (1.21) (0.78) (0.50) (0.74) (0.49) (0.85) (0.81) (2.20) Observations 187 132 187 132 52 45 52 45 Colonial controls No Yes No Yes No Yes No Yes Geo/climate controls No Yes No Yes No Yes No Yes K-P F-stat 0.887 5.842 5.296 5.726 0.812 1.521 0.772 0.285 SW F-stat for Trade 2.062 13.15 10.97 16.99 8.577 16.30 3.279 0.680 SW F-stat for Mig. 5.635 16.59 55.29 8.622 3.176 3.248 2.413 0.698 SY 10% max IV size 7.030 7.030 7.030 7.030 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 3.630 3.630 3.630 3.630 Notes: The dependent variable is the log of income per capita. LP (NLP) stands for linear predicted trade and migration based on the OLS (non-linear Poisson, PPML) gravity estimates. Geographic, climate and disease controls are regional dummies for Africa, a landlocked dummy, the percentage of land in the tropics, average distance to the coast and a measure for oil reserves. Colonial history controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 6: Income and Openness, Identifying Partners’ Impact Intra-African Africa with no ind. Africa with ind. LP LP NLP NLP LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 8.28 13.97 0.86 4.86 4.44 3.50 5.15 6.68 ** 2.08 ** 3.42 ** 3.05 *** 4.51 *** (10.26) (13.00) (2.85) (3.90) (2.87) (3.40) (4.60) (2.92) (0.89) (1.38) (0.93) (1.43) Immig. 0.10 3.40 4.94 6.14 5.36 12.33 ** 6.80 6.21 (4.69) (4.56) (5.20) (4.03) (6.00) (5.22) (16.22) (10.34) Emig. 34.52 ** −2.21 16.16 −9.01 (16.45) (18.41) (11.04) (6.72) Ln pop. −0.45 ** −0.39 * −0.34 ** −0.27 ** −0.30 * −0.15 −0.27 −0.24 −0.26 * −0.30 ** −0.25 * −0.30 ** (0.21) (0.22) (0.14) (0.13) (0.16) (0.14) (0.22) (0.19) (0.15) (0.12) (0.15) (0.14) Ln area 0.15 0.27 0.07 0.17 0.09 0.16 0.09 0.21 * 0.29 0.12 0.15 0.08 (0.16) (0.19) (0.10) (0.11) (0.11) (0.10) (0.14) (0.12) (0.19) (0.13) (0.11) (0.10) Dist. equator 0.05 *** 0.06 *** 0.05 *** 0.06 *** 0.06 *** 0.07 *** 0.06 *** 0.07 *** 0.03 * 0.04 *** 0.04 *** 0.05 *** (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.02) (0.02) (0.02) (0.01) (0.01) (0.01) Constant 6.08 *** 5.14 *** 6.65 *** 5.73 *** 5.63 *** 4.93 *** 5.39 *** 4.55 *** 4.86 *** 5.79 *** 5.46 *** 5.85 *** (1.24) (1.25) (0.64) (0.62) (0.71) (0.95) (0.81) (1.06) (1.10) (0.84) (0.52) (0.45) Observations 52 50 52 50 52 50 52 50 52 50 52 50 Colonial/geo controls No Yes No Yes No Yes No Yes No Yes No Yes K-P F-stat 0.953 0.847 6.171 4.468 1.814 2.123 0.225 3.337 3.738 1.695 9.923 4.850 SW F-stat for Trade 1.905 1.751 9.866 8.120 7.467 7.938 1.088 13.35 19.40 11.32 24.52 10.92 SW F-stat for Mig. 8.659 15.56 5.727 7.168 2.139 3.164 0.604 2.382 9.958 3.580 20.74 7.414 SY 10% max IV size 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 SY 25% max IV size 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 Intra-African Africa with no ind. Africa with ind. LP LP NLP NLP LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 8.28 13.97 0.86 4.86 4.44 3.50 5.15 6.68 ** 2.08 ** 3.42 ** 3.05 *** 4.51 *** (10.26) (13.00) (2.85) (3.90) (2.87) (3.40) (4.60) (2.92) (0.89) (1.38) (0.93) (1.43) Immig. 0.10 3.40 4.94 6.14 5.36 12.33 ** 6.80 6.21 (4.69) (4.56) (5.20) (4.03) (6.00) (5.22) (16.22) (10.34) Emig. 34.52 ** −2.21 16.16 −9.01 (16.45) (18.41) (11.04) (6.72) Ln pop. −0.45 ** −0.39 * −0.34 ** −0.27 ** −0.30 * −0.15 −0.27 −0.24 −0.26 * −0.30 ** −0.25 * −0.30 ** (0.21) (0.22) (0.14) (0.13) (0.16) (0.14) (0.22) (0.19) (0.15) (0.12) (0.15) (0.14) Ln area 0.15 0.27 0.07 0.17 0.09 0.16 0.09 0.21 * 0.29 0.12 0.15 0.08 (0.16) (0.19) (0.10) (0.11) (0.11) (0.10) (0.14) (0.12) (0.19) (0.13) (0.11) (0.10) Dist. equator 0.05 *** 0.06 *** 0.05 *** 0.06 *** 0.06 *** 0.07 *** 0.06 *** 0.07 *** 0.03 * 0.04 *** 0.04 *** 0.05 *** (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.02) (0.02) (0.02) (0.01) (0.01) (0.01) Constant 6.08 *** 5.14 *** 6.65 *** 5.73 *** 5.63 *** 4.93 *** 5.39 *** 4.55 *** 4.86 *** 5.79 *** 5.46 *** 5.85 *** (1.24) (1.25) (0.64) (0.62) (0.71) (0.95) (0.81) (1.06) (1.10) (0.84) (0.52) (0.45) Observations 52 50 52 50 52 50 52 50 52 50 52 50 Colonial/geo controls No Yes No Yes No Yes No Yes No Yes No Yes K-P F-stat 0.953 0.847 6.171 4.468 1.814 2.123 0.225 3.337 3.738 1.695 9.923 4.850 SW F-stat for Trade 1.905 1.751 9.866 8.120 7.467 7.938 1.088 13.35 19.40 11.32 24.52 10.92 SW F-stat for Mig. 8.659 15.56 5.727 7.168 2.139 3.164 0.604 2.382 9.958 3.580 20.74 7.414 SY 10% max IV size 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 SY 25% max IV size 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 Notes: The dependent variable is the log of income per capita. LP (NLP) stands for linear predicted trade and migration based on the OLS (non-linear Poisson, PPML) gravity estimates. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 6: Income and Openness, Identifying Partners’ Impact Intra-African Africa with no ind. Africa with ind. LP LP NLP NLP LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 8.28 13.97 0.86 4.86 4.44 3.50 5.15 6.68 ** 2.08 ** 3.42 ** 3.05 *** 4.51 *** (10.26) (13.00) (2.85) (3.90) (2.87) (3.40) (4.60) (2.92) (0.89) (1.38) (0.93) (1.43) Immig. 0.10 3.40 4.94 6.14 5.36 12.33 ** 6.80 6.21 (4.69) (4.56) (5.20) (4.03) (6.00) (5.22) (16.22) (10.34) Emig. 34.52 ** −2.21 16.16 −9.01 (16.45) (18.41) (11.04) (6.72) Ln pop. −0.45 ** −0.39 * −0.34 ** −0.27 ** −0.30 * −0.15 −0.27 −0.24 −0.26 * −0.30 ** −0.25 * −0.30 ** (0.21) (0.22) (0.14) (0.13) (0.16) (0.14) (0.22) (0.19) (0.15) (0.12) (0.15) (0.14) Ln area 0.15 0.27 0.07 0.17 0.09 0.16 0.09 0.21 * 0.29 0.12 0.15 0.08 (0.16) (0.19) (0.10) (0.11) (0.11) (0.10) (0.14) (0.12) (0.19) (0.13) (0.11) (0.10) Dist. equator 0.05 *** 0.06 *** 0.05 *** 0.06 *** 0.06 *** 0.07 *** 0.06 *** 0.07 *** 0.03 * 0.04 *** 0.04 *** 0.05 *** (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.02) (0.02) (0.02) (0.01) (0.01) (0.01) Constant 6.08 *** 5.14 *** 6.65 *** 5.73 *** 5.63 *** 4.93 *** 5.39 *** 4.55 *** 4.86 *** 5.79 *** 5.46 *** 5.85 *** (1.24) (1.25) (0.64) (0.62) (0.71) (0.95) (0.81) (1.06) (1.10) (0.84) (0.52) (0.45) Observations 52 50 52 50 52 50 52 50 52 50 52 50 Colonial/geo controls No Yes No Yes No Yes No Yes No Yes No Yes K-P F-stat 0.953 0.847 6.171 4.468 1.814 2.123 0.225 3.337 3.738 1.695 9.923 4.850 SW F-stat for Trade 1.905 1.751 9.866 8.120 7.467 7.938 1.088 13.35 19.40 11.32 24.52 10.92 SW F-stat for Mig. 8.659 15.56 5.727 7.168 2.139 3.164 0.604 2.382 9.958 3.580 20.74 7.414 SY 10% max IV size 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 SY 25% max IV size 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 Intra-African Africa with no ind. Africa with ind. LP LP NLP NLP LP LP NLP NLP LP LP NLP NLP (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 8.28 13.97 0.86 4.86 4.44 3.50 5.15 6.68 ** 2.08 ** 3.42 ** 3.05 *** 4.51 *** (10.26) (13.00) (2.85) (3.90) (2.87) (3.40) (4.60) (2.92) (0.89) (1.38) (0.93) (1.43) Immig. 0.10 3.40 4.94 6.14 5.36 12.33 ** 6.80 6.21 (4.69) (4.56) (5.20) (4.03) (6.00) (5.22) (16.22) (10.34) Emig. 34.52 ** −2.21 16.16 −9.01 (16.45) (18.41) (11.04) (6.72) Ln pop. −0.45 ** −0.39 * −0.34 ** −0.27 ** −0.30 * −0.15 −0.27 −0.24 −0.26 * −0.30 ** −0.25 * −0.30 ** (0.21) (0.22) (0.14) (0.13) (0.16) (0.14) (0.22) (0.19) (0.15) (0.12) (0.15) (0.14) Ln area 0.15 0.27 0.07 0.17 0.09 0.16 0.09 0.21 * 0.29 0.12 0.15 0.08 (0.16) (0.19) (0.10) (0.11) (0.11) (0.10) (0.14) (0.12) (0.19) (0.13) (0.11) (0.10) Dist. equator 0.05 *** 0.06 *** 0.05 *** 0.06 *** 0.06 *** 0.07 *** 0.06 *** 0.07 *** 0.03 * 0.04 *** 0.04 *** 0.05 *** (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.02) (0.02) (0.02) (0.01) (0.01) (0.01) Constant 6.08 *** 5.14 *** 6.65 *** 5.73 *** 5.63 *** 4.93 *** 5.39 *** 4.55 *** 4.86 *** 5.79 *** 5.46 *** 5.85 *** (1.24) (1.25) (0.64) (0.62) (0.71) (0.95) (0.81) (1.06) (1.10) (0.84) (0.52) (0.45) Observations 52 50 52 50 52 50 52 50 52 50 52 50 Colonial/geo controls No Yes No Yes No Yes No Yes No Yes No Yes K-P F-stat 0.953 0.847 6.171 4.468 1.814 2.123 0.225 3.337 3.738 1.695 9.923 4.850 SW F-stat for Trade 1.905 1.751 9.866 8.120 7.467 7.938 1.088 13.35 19.40 11.32 24.52 10.92 SW F-stat for Mig. 8.659 15.56 5.727 7.168 2.139 3.164 0.604 2.382 9.958 3.580 20.74 7.414 SY 10% max IV size 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 7.03 SY 25% max IV size 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 3.63 Notes: The dependent variable is the log of income per capita. LP (NLP) stands for linear predicted trade and migration based on the OLS (non-linear Poisson, PPML) gravity estimates. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large At this stage, it is important to check the relevance of gravity-based instruments since the lack of significance of trade openness may come from a problem of weak instruments. To this end, we implement (i) the Kleibergen and Paap (2006) rk Wald F-stat test (KP test) which tests for the null hypothesis of jointly weak instruments, and (ii) the Sanderson and Windmeijer (2015) F-stat test (SW test) of weak identification for each endogenous regressor separately.14 Doing so allows us to evaluate whether each individual endogenous regressor is well identified separately, by partialling-out the influence of the other endogenous regressors. For the world sample (columns (1–4) of Table 5), the null of (jointly and individual) weak identifications is rejected at conventional level of significance, except for column (1). Particularly, for the most relevant specification relying on non-linear gravity-based instruments and using all control variables, the KP test statistic for jointly weak identification is 5.96, which is above the Stock and Yogo’s (2005) critical value at 25% max IV size (3.63). For this specification, for each endogenous regressor, we reject the null of individual weak identification. The SW test statistic for individual weak identification is 14.60 for openness to trade and 8.63 for openness to immigration; both values being above all Stock and Yogo’s critical values. On the whole, these results indicate that the weak instrument issue is not a severe concern in our estimations. Let us now contrast the previous results with those obtained on the African sample. Columns (5–8) in Table 5 report the impact of African openness with the world. These results differ from those obtained for the world sample, and highlight the relevance of isolating this subgroup of countries. In particular, focusing on non-linear estimation (columns (3) and (4) and (7) and (8)) two main conclusions can be drawn from our findings. First, comparing columns (3) and (7) on the one hand and (4) and (8)—which include all control variables—on the other hand, highlights that the significance of trade is identical between world openness and African openness. Second, contrasting columns (4) and (8) shows that immigration is significant in the full sample while is not the case in the African sample. On the whole, focusing on our variables of interest, the major difference between the full and the African samples concerns the role of immigration, while the impact of trade remains similar. Finally, looking specifically at the African sample, based on both linear and non-linear predicted openness (columns (5) and (7)), the coefficient of trade is significantly positive when we control for country size and distance to equator. However, when we consider all the other controls, there is no evidence of positive and significant impact of trade, and immigration has a positive significant effect only in the linear case. For all specifications, we are not able to reject the null of weak instruments. 4.2.2 Identifying partner-varying impact of openness The fact that results for Africa differ from those obtained at a world level—especially using migration openness—justifies the need to deepen the analysis of the potential growth-enhancing impact of openness in Africa. Indeed, it is reasonable to think that the difficulty to find strong evidence of the influence of openness at the aggregate level in Africa comes from the characteristics of African openness. As aforementioned, trade with advanced economies may have more income-enhancing impact in Africa than trade with other countries like China and African neighbours. With regards to migration, the impact of Africa’s openness to migration with developed countries (net emigration) may differ from that of intra-African migration (immigration and emigration within African countries). To test these conjectures, we estimate the impact of openness on income in Africa depending on the type of partners; the gravity-predicted openness being derived from the estimation of Equation (6). Regarding migration, as a large part of African migration consists in intra-African migration and emigration to developed countries, we consider three cases: (i) inter-African immigration (columns (1–4)) which refers to immigration from African countries, (ii) immigration from developing countries, including African countries (columns (5–8)), and (iii) emigration to industrialised countries (columns (9–12)).15 The corresponding estimation results are reported in Table 6. The results in columns (1–4) show that there is no evidence of growth-enhancing impact of intra-African openness. Neither intra-continental trade nor immigration from African countries significantly influence growth, based on both linear and non-linear gravity-based instruments. In the non-linear case, the hypothesis of (jointly and individual) weak identification of endogenous regressors cannot be retained. The results reported in columns (5–8) also indicate that there is no strong evidence of income-enhancing impact of openness with low-income countries (including African economies). When we only control for country size (population and area) and for distance to equator (columns (5) and (7)), neither trade nor immigration from non-industrialised countries has a significant impact and we cannot reject the weak identification of endogenous regressors. Including other controls in columns (6) and (8), there is a significant positive impact of trade in the linear case (column (6)) and a significant positive effect of immigration in the non-linear case (column (8)); in both cases, only trade is not weakly identified. Turning to openness with advanced economies (trade with industrialised countries and emigration to industrialised economies), the results are reported in columns (9–12) of Table 6. They show overwhelming evidence of a growth-enhancing impact of trade with industrialised countries. When we only control for country size and distance to equator, we found a positive significant role of both trade and emigration on income per capita in the linear case (column (9)) and a positive significant effect of only trade in the non-linear case (column (10)); in both cases, trade and emigration are not weakly identified. In columns (10) and (12), when we include other control variables, there is only a positive significant impact of trade openness with advanced countries, and both endogenous regressors are not weakly identified, except for emigration in column (10). To sum up, in analysing the impact of openness in Africa depending on partners, we found strong evidence that trade with industrialised economies promotes economic development in African countries, while we do not establish a strong impact of openness to migration contrary to Ortega and Peri (2014). In addition to emphasising the interest of accounting for the type of partner countries, these findings corroborate the theoretical intuition that African countries may benefit more from trading with advanced economies which are more technologically innovative (Yanikkaya, 2003). The underlying idea is that if growth is driven by technological progress, trade allows African countries to benefit from the advances in R&D activities of their trading partners. It is worth mentioning that the type of the partner country is linked to the composition of trade flows. Indeed, as shown by Figures 1 and 2, (i) exports of African countries are mainly composed by primary commodities, while they mostly import manufactured goods, and (ii) Africa’s trade is essentially realised with developed countries which principally export manufactured goods. Such goods, which are imported by African countries, are typically more technology and skill intensive than those from developing countries (including Africa).16 In this sense, the development stage of the trading partner and the composition of trade flows are strongly related. As a result, trade with advanced countries which mainly concerns manufactured goods leads to technology transfers that are benefit for Africa, while it is not the case for trade with developing countries that essentially focuses on primary commodities. Besides, our findings also reflect the aforementioned ambiguous relationship between growth and emigration. In other words, we found a compensation between the adverse impact of emigration (through brain drain) and its positive effect (through remittances, human capital of returning migrants, knowledge transfer, improving institutions).17 Furthermore, the absence of significant impact of immigration from African countries may reflect some lack of complementarity in African labour force. 4.3 Sensitivity analyses For the sake of completeness and as robustness checks, we now provide some sensitivity analyses. First, we investigate the robustness of our results to the African countries included in our sample. To this end, we focus on Sub-Saharan Africa (SSA) by excluding North African countries from the sample because of their geographical proximity to Europe.18 As a second robustness check, we test the sensitivity of our results to the choice of the considered year in line with Irwin and Terviö (2002). In addition to our baseline year 2000, two over years are considered, namely 2005 and 2010. Third, we go further than a usual sensitivity analysis and address a hot-debated topic, namely the trade links between China and Africa. This relationship is of particular interest since trade between China and Africa is growing since the early 2000s while it was insignificant in the 1980s and 1990s. China has become the largest trading partner of many African economies including countries where trade was formerly largely oriented toward Europe or America because of colonial or historical ties, such as Angola, Benin, Ethiopia, Gambia, Ghana, Mauritania, Sudan, Togo and Tanzania. Specifically, China that accounted for only 3% of total African trade in 2000 rose to 12% in 2010 and continues to increase its market share on the continent, while the share of trade with traditional partners of Africa as France and the United States is declining. It is thus worthy of interest to investigate whether trade with this ‘new partner’ is growth-enhancing for Africa. It should be mentioned that we focus on the trade side, since emigration of Africans to China is almost zero. Thus, in terms of openness between China and Africa, only the trade channel is relevant to consider. The results of the sensitivity analyses are displayed in Table 7.19 They confirm the evidence of income-enhancing impact of trade with advanced countries. Indeed, in line with our previous findings, Table 7 shows a significant positive impact of trade with industrialised countries for the subsample of SSA economies and whatever the period (year) of analysis. Besides and interestingly, there is no evidence of a strong positive impact of trade with China. Table 7: Sensitivity Analysis 2000, SSA only 2000 2005 2010 African No ind. ind. China African No ind. ind. China African No ind. ind. China (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 6.29 8.38 * 3.74 *** 42.51 1.61 2.40 3.85 *** 19.38 2.40 2.74 4.48 *** 15.83 ** (7.03) (4.77) (1.44) (30.21) (2.59) (1.55) (1.23) (12.11) (2.82) (1.75) (1.61) (8.00) Immig. 5.15 15.55 * 4.51 1.52 4.42 −9.16 (4.13) (9.24) (4.94) (10.88) (5.72) (23.07) Emig. −5.43 −4.00 −4.73 (5.59) (7.05) (7.88) Ln pop. −0.42 −0.36 −0.31 ** −0.04 −0.25 ** −0.26 * −0.26 * −0.07 −0.25 ** −0.29 −0.22 * 0.09 (0.26) (0.26) (0.15) (0.26) (0.12) (0.15) (0.14) (0.20) (0.11) (0.19) (0.13) (0.19) Ln area 0.19 0.24 0.08 −0.02 0.16 0.21 * 0.10 0.02 0.17 0.22 * 0.10 −0.02 (0.17) (0.18) (0.10) (0.16) (0.11) (0.11) (0.10) (0.13) (0.10) (0.13) (0.10) (0.14) Dist. equator 0.03 0.06 0.04 ** 0.07 *** 0.05 *** 0.06 *** 0.04 *** 0.08 *** 0.05 *** 0.04 * 0.03 ** 0.07 *** (0.02) (0.03) (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.01) (0.01) Constant 6.12 *** 4.32 *** 5.99 *** 5.30 *** 6.07 *** 5.47 *** 5.90 *** 5.56 *** 6.19 *** 5.85 *** 6.06 *** 5.47 *** (0.78) (1.56) (0.56) (1.04) (0.61) (0.79) (0.45) (0.90) (0.63) (0.98) (0.45) (0.91) Observations 44 44 44 50 50 50 50 50 50 50 50 50 Colonial/geo controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes K-P F-stat 2.897 1.063 1.856 1.061 6.071 0.872 7.396 2.354 6.007 0.723 5.954 5.498 SW F-stat for Trade 5.730 3.189 4.209 1.061 6.976 4.358 17.35 2.354 7.019 2.789 12.08 5.498 SW F-stat for Mig. 5.885 1.918 11.27 5.089 1.759 8.124 6.033 1.710 10.54 SY 10% max IV size 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 SY 25% max IV size 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 2000, SSA only 2000 2005 2010 African No ind. ind. China African No ind. ind. China African No ind. ind. China (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 6.29 8.38 * 3.74 *** 42.51 1.61 2.40 3.85 *** 19.38 2.40 2.74 4.48 *** 15.83 ** (7.03) (4.77) (1.44) (30.21) (2.59) (1.55) (1.23) (12.11) (2.82) (1.75) (1.61) (8.00) Immig. 5.15 15.55 * 4.51 1.52 4.42 −9.16 (4.13) (9.24) (4.94) (10.88) (5.72) (23.07) Emig. −5.43 −4.00 −4.73 (5.59) (7.05) (7.88) Ln pop. −0.42 −0.36 −0.31 ** −0.04 −0.25 ** −0.26 * −0.26 * −0.07 −0.25 ** −0.29 −0.22 * 0.09 (0.26) (0.26) (0.15) (0.26) (0.12) (0.15) (0.14) (0.20) (0.11) (0.19) (0.13) (0.19) Ln area 0.19 0.24 0.08 −0.02 0.16 0.21 * 0.10 0.02 0.17 0.22 * 0.10 −0.02 (0.17) (0.18) (0.10) (0.16) (0.11) (0.11) (0.10) (0.13) (0.10) (0.13) (0.10) (0.14) Dist. equator 0.03 0.06 0.04 ** 0.07 *** 0.05 *** 0.06 *** 0.04 *** 0.08 *** 0.05 *** 0.04 * 0.03 ** 0.07 *** (0.02) (0.03) (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.01) (0.01) Constant 6.12 *** 4.32 *** 5.99 *** 5.30 *** 6.07 *** 5.47 *** 5.90 *** 5.56 *** 6.19 *** 5.85 *** 6.06 *** 5.47 *** (0.78) (1.56) (0.56) (1.04) (0.61) (0.79) (0.45) (0.90) (0.63) (0.98) (0.45) (0.91) Observations 44 44 44 50 50 50 50 50 50 50 50 50 Colonial/geo controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes K-P F-stat 2.897 1.063 1.856 1.061 6.071 0.872 7.396 2.354 6.007 0.723 5.954 5.498 SW F-stat for Trade 5.730 3.189 4.209 1.061 6.976 4.358 17.35 2.354 7.019 2.789 12.08 5.498 SW F-stat for Mig. 5.885 1.918 11.27 5.089 1.759 8.124 6.033 1.710 10.54 SY 10% max IV size 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 SY 25% max IV size 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 Notes: The dependent variable is the log of income per capita. Predicted trade and migration are obtained from non-linear Poisson gravity estimates. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 7: Sensitivity Analysis 2000, SSA only 2000 2005 2010 African No ind. ind. China African No ind. ind. China African No ind. ind. China (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 6.29 8.38 * 3.74 *** 42.51 1.61 2.40 3.85 *** 19.38 2.40 2.74 4.48 *** 15.83 ** (7.03) (4.77) (1.44) (30.21) (2.59) (1.55) (1.23) (12.11) (2.82) (1.75) (1.61) (8.00) Immig. 5.15 15.55 * 4.51 1.52 4.42 −9.16 (4.13) (9.24) (4.94) (10.88) (5.72) (23.07) Emig. −5.43 −4.00 −4.73 (5.59) (7.05) (7.88) Ln pop. −0.42 −0.36 −0.31 ** −0.04 −0.25 ** −0.26 * −0.26 * −0.07 −0.25 ** −0.29 −0.22 * 0.09 (0.26) (0.26) (0.15) (0.26) (0.12) (0.15) (0.14) (0.20) (0.11) (0.19) (0.13) (0.19) Ln area 0.19 0.24 0.08 −0.02 0.16 0.21 * 0.10 0.02 0.17 0.22 * 0.10 −0.02 (0.17) (0.18) (0.10) (0.16) (0.11) (0.11) (0.10) (0.13) (0.10) (0.13) (0.10) (0.14) Dist. equator 0.03 0.06 0.04 ** 0.07 *** 0.05 *** 0.06 *** 0.04 *** 0.08 *** 0.05 *** 0.04 * 0.03 ** 0.07 *** (0.02) (0.03) (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.01) (0.01) Constant 6.12 *** 4.32 *** 5.99 *** 5.30 *** 6.07 *** 5.47 *** 5.90 *** 5.56 *** 6.19 *** 5.85 *** 6.06 *** 5.47 *** (0.78) (1.56) (0.56) (1.04) (0.61) (0.79) (0.45) (0.90) (0.63) (0.98) (0.45) (0.91) Observations 44 44 44 50 50 50 50 50 50 50 50 50 Colonial/geo controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes K-P F-stat 2.897 1.063 1.856 1.061 6.071 0.872 7.396 2.354 6.007 0.723 5.954 5.498 SW F-stat for Trade 5.730 3.189 4.209 1.061 6.976 4.358 17.35 2.354 7.019 2.789 12.08 5.498 SW F-stat for Mig. 5.885 1.918 11.27 5.089 1.759 8.124 6.033 1.710 10.54 SY 10% max IV size 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 SY 25% max IV size 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 2000, SSA only 2000 2005 2010 African No ind. ind. China African No ind. ind. China African No ind. ind. China (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Trade 6.29 8.38 * 3.74 *** 42.51 1.61 2.40 3.85 *** 19.38 2.40 2.74 4.48 *** 15.83 ** (7.03) (4.77) (1.44) (30.21) (2.59) (1.55) (1.23) (12.11) (2.82) (1.75) (1.61) (8.00) Immig. 5.15 15.55 * 4.51 1.52 4.42 −9.16 (4.13) (9.24) (4.94) (10.88) (5.72) (23.07) Emig. −5.43 −4.00 −4.73 (5.59) (7.05) (7.88) Ln pop. −0.42 −0.36 −0.31 ** −0.04 −0.25 ** −0.26 * −0.26 * −0.07 −0.25 ** −0.29 −0.22 * 0.09 (0.26) (0.26) (0.15) (0.26) (0.12) (0.15) (0.14) (0.20) (0.11) (0.19) (0.13) (0.19) Ln area 0.19 0.24 0.08 −0.02 0.16 0.21 * 0.10 0.02 0.17 0.22 * 0.10 −0.02 (0.17) (0.18) (0.10) (0.16) (0.11) (0.11) (0.10) (0.13) (0.10) (0.13) (0.10) (0.14) Dist. equator 0.03 0.06 0.04 ** 0.07 *** 0.05 *** 0.06 *** 0.04 *** 0.08 *** 0.05 *** 0.04 * 0.03 ** 0.07 *** (0.02) (0.03) (0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.01) (0.01) Constant 6.12 *** 4.32 *** 5.99 *** 5.30 *** 6.07 *** 5.47 *** 5.90 *** 5.56 *** 6.19 *** 5.85 *** 6.06 *** 5.47 *** (0.78) (1.56) (0.56) (1.04) (0.61) (0.79) (0.45) (0.90) (0.63) (0.98) (0.45) (0.91) Observations 44 44 44 50 50 50 50 50 50 50 50 50 Colonial/geo controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes K-P F-stat 2.897 1.063 1.856 1.061 6.071 0.872 7.396 2.354 6.007 0.723 5.954 5.498 SW F-stat for Trade 5.730 3.189 4.209 1.061 6.976 4.358 17.35 2.354 7.019 2.789 12.08 5.498 SW F-stat for Mig. 5.885 1.918 11.27 5.089 1.759 8.124 6.033 1.710 10.54 SY 10% max IV size 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 7.030 7.030 7.030 16.38 SY 25% max IV size 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 3.630 3.630 3.630 5.530 Notes: The dependent variable is the log of income per capita. Predicted trade and migration are obtained from non-linear Poisson gravity estimates. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large 5. Transmission channels 5.1 The channel through which trade with developed countries affects income in Africa While our previous results show that only trade with developed countries positively affects income in Africa, they do not provide information on the channel through which this positive effect operates. This section tackles this issue by providing some insights regarding those transmission channels. 5.1.1 Export and import sides Trade with developed countries may enhance growth in Africa through (i) the exploitation of economies of scale and comparative advantages which are reflected in exports, or (ii) technology transfers captured by imports. To disentangle these two channels, we estimate our regressions by considering two measures of trade openness, namely exports to GDP ratio and imports to GDP ratio (Yanikkaya, 2003).20 The results are reported in Table 8 (columns (1–4)) and show that trade with developed countries promotes growth in Africa through both exports and imports, with a slightly more pronounced impact for imports. Therefore, our findings support the hypothesis that African countries that are more open to trade with developed countries experience higher growth by importing technologically advanced goods, but also by accessing to a larger market allowing to exploit economies of scale and comparative advantages. Our results are in line with Yanikkaya (2003) regarding trade with the U.S., reflecting that imports and exports are complementary. Overall, it should be noted that while this decomposition of trade is useful in attempting to examine the dominant channel, it is often preferable to aggregate exports and imports in particular because they are complementary. Table 8: The Channels of Growth-enhancing Impact of Trade with Developed Countries (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) lnY/L lnY/L lnY/L lnY/L lnY/L α1−αlnK/Y lnH/L lnA lnY/L α1−αlnK/Y lnH/L lnA Trade 2.28 *** 0.01 0.78 ** 2.04 *** 3.54 ** −0.37 0.58 4.00 *** (0.73) (0.30) (0.36) (0.72) (1.52) (0.48) (0.63) (1.54) Exports 6.18 *** 7.11 *** (1.69) (1.96) Imports 7.94 *** 9.38 ** (2.80) (4.54) Ln pop. −0.19 −0.28 * −0.33 * −0.32 * −0.22 * −0.05 −0.00 −0.16 −0.26 ** −0.08 * −0.05 −0.16 (0.16) (0.15) (0.17) (0.17) (0.12) (0.04) (0.06) (0.12) (0.13) (0.05) (0.04) (0.15) Ln area −0.07 0.01 0.16 0.29 * 0.03 −0.01 −0.03 0.04 0.07 −0.00 0.02 0.07 (0.11) (0.11) (0.13) (0.15) (0.08) (0.03) (0.05) (0.08) (0.09) (0.03) (0.03) (0.11) Dist. equator 0.05 *** 0.04 *** 0.04 *** 0.04 *** 0.05 *** −0.00 0.01 ** 0.05 *** 0.05 *** −0.00 0.00 0.05 *** (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.01) (0.01) (0.00) (0.00) (0.01) Constant 6.51 *** 6.27 *** 5.54 *** 4.59 *** −3.92 *** −0.02 −0.90 *** −3.43 *** −4.60 *** 0.26 −0.99 *** −4.42 *** (0.34) (0.35) (0.60) (0.96) (0.35) (0.11) (0.12) (0.37) (0.47) (0.17) (0.18) (0.49) Observations 52 50 52 50 46 46 49 45 45 45 47 44 Colonial/geo controls No Yes No Yes No No No No Yes Yes Yes Yes K-P F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SW F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SY 10% max IV size 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 SY 25% max IV size 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) lnY/L lnY/L lnY/L lnY/L lnY/L α1−αlnK/Y lnH/L lnA lnY/L α1−αlnK/Y lnH/L lnA Trade 2.28 *** 0.01 0.78 ** 2.04 *** 3.54 ** −0.37 0.58 4.00 *** (0.73) (0.30) (0.36) (0.72) (1.52) (0.48) (0.63) (1.54) Exports 6.18 *** 7.11 *** (1.69) (1.96) Imports 7.94 *** 9.38 ** (2.80) (4.54) Ln pop. −0.19 −0.28 * −0.33 * −0.32 * −0.22 * −0.05 −0.00 −0.16 −0.26 ** −0.08 * −0.05 −0.16 (0.16) (0.15) (0.17) (0.17) (0.12) (0.04) (0.06) (0.12) (0.13) (0.05) (0.04) (0.15) Ln area −0.07 0.01 0.16 0.29 * 0.03 −0.01 −0.03 0.04 0.07 −0.00 0.02 0.07 (0.11) (0.11) (0.13) (0.15) (0.08) (0.03) (0.05) (0.08) (0.09) (0.03) (0.03) (0.11) Dist. equator 0.05 *** 0.04 *** 0.04 *** 0.04 *** 0.05 *** −0.00 0.01 ** 0.05 *** 0.05 *** −0.00 0.00 0.05 *** (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.01) (0.01) (0.00) (0.00) (0.01) Constant 6.51 *** 6.27 *** 5.54 *** 4.59 *** −3.92 *** −0.02 −0.90 *** −3.43 *** −4.60 *** 0.26 −0.99 *** −4.42 *** (0.34) (0.35) (0.60) (0.96) (0.35) (0.11) (0.12) (0.37) (0.47) (0.17) (0.18) (0.49) Observations 52 50 52 50 46 46 49 45 45 45 47 44 Colonial/geo controls No Yes No Yes No No No No Yes Yes Yes Yes K-P F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SW F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SY 10% max IV size 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 SY 25% max IV size 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 Notes: In columns (5–12), the explained variables are normalised by the value of the US. The predicted values of trade, exports and imports are those based on the non-linear estimation. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 8: The Channels of Growth-enhancing Impact of Trade with Developed Countries (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) lnY/L lnY/L lnY/L lnY/L lnY/L α1−αlnK/Y lnH/L lnA lnY/L α1−αlnK/Y lnH/L lnA Trade 2.28 *** 0.01 0.78 ** 2.04 *** 3.54 ** −0.37 0.58 4.00 *** (0.73) (0.30) (0.36) (0.72) (1.52) (0.48) (0.63) (1.54) Exports 6.18 *** 7.11 *** (1.69) (1.96) Imports 7.94 *** 9.38 ** (2.80) (4.54) Ln pop. −0.19 −0.28 * −0.33 * −0.32 * −0.22 * −0.05 −0.00 −0.16 −0.26 ** −0.08 * −0.05 −0.16 (0.16) (0.15) (0.17) (0.17) (0.12) (0.04) (0.06) (0.12) (0.13) (0.05) (0.04) (0.15) Ln area −0.07 0.01 0.16 0.29 * 0.03 −0.01 −0.03 0.04 0.07 −0.00 0.02 0.07 (0.11) (0.11) (0.13) (0.15) (0.08) (0.03) (0.05) (0.08) (0.09) (0.03) (0.03) (0.11) Dist. equator 0.05 *** 0.04 *** 0.04 *** 0.04 *** 0.05 *** −0.00 0.01 ** 0.05 *** 0.05 *** −0.00 0.00 0.05 *** (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.01) (0.01) (0.00) (0.00) (0.01) Constant 6.51 *** 6.27 *** 5.54 *** 4.59 *** −3.92 *** −0.02 −0.90 *** −3.43 *** −4.60 *** 0.26 −0.99 *** −4.42 *** (0.34) (0.35) (0.60) (0.96) (0.35) (0.11) (0.12) (0.37) (0.47) (0.17) (0.18) (0.49) Observations 52 50 52 50 46 46 49 45 45 45 47 44 Colonial/geo controls No Yes No Yes No No No No Yes Yes Yes Yes K-P F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SW F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SY 10% max IV size 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 SY 25% max IV size 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) lnY/L lnY/L lnY/L lnY/L lnY/L α1−αlnK/Y lnH/L lnA lnY/L α1−αlnK/Y lnH/L lnA Trade 2.28 *** 0.01 0.78 ** 2.04 *** 3.54 ** −0.37 0.58 4.00 *** (0.73) (0.30) (0.36) (0.72) (1.52) (0.48) (0.63) (1.54) Exports 6.18 *** 7.11 *** (1.69) (1.96) Imports 7.94 *** 9.38 ** (2.80) (4.54) Ln pop. −0.19 −0.28 * −0.33 * −0.32 * −0.22 * −0.05 −0.00 −0.16 −0.26 ** −0.08 * −0.05 −0.16 (0.16) (0.15) (0.17) (0.17) (0.12) (0.04) (0.06) (0.12) (0.13) (0.05) (0.04) (0.15) Ln area −0.07 0.01 0.16 0.29 * 0.03 −0.01 −0.03 0.04 0.07 −0.00 0.02 0.07 (0.11) (0.11) (0.13) (0.15) (0.08) (0.03) (0.05) (0.08) (0.09) (0.03) (0.03) (0.11) Dist. equator 0.05 *** 0.04 *** 0.04 *** 0.04 *** 0.05 *** −0.00 0.01 ** 0.05 *** 0.05 *** −0.00 0.00 0.05 *** (0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.01) (0.01) (0.00) (0.00) (0.01) Constant 6.51 *** 6.27 *** 5.54 *** 4.59 *** −3.92 *** −0.02 −0.90 *** −3.43 *** −4.60 *** 0.26 −0.99 *** −4.42 *** (0.34) (0.35) (0.60) (0.96) (0.35) (0.11) (0.12) (0.37) (0.47) (0.17) (0.18) (0.49) Observations 52 50 52 50 46 46 49 45 45 45 47 44 Colonial/geo controls No Yes No Yes No No No No Yes Yes Yes Yes K-P F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SW F-stat 18.25 9.448 21.43 5.851 33.62 33.62 36.76 33.37 9.373 9.373 15.35 9.246 SY 10% max IV size 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 16.38 SY 25% max IV size 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 5.530 Notes: In columns (5–12), the explained variables are normalised by the value of the US. The predicted values of trade, exports and imports are those based on the non-linear estimation. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 10% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large 5.1.2 Output decomposition To reinforce the technology transfer hypothesis, we propose a decomposition of income based on a simple Cobb-Douglas production function in the same spirit as and Jones (1999), Frankel . Specifically, we consider the following function: Yi=Kiα(HiAi)1−α, (9) where Yi stands for output in country i, Ki is the stock of physical capital, Hi denotes the human capital, Ai is a productivity term, and α represents the labour share in income. Rewriting this production function in terms of output per worker and using the logarithmic transformation, we get: lnyi=α(1−α)lnKiYi+lnhi+ln(Ai), (10) where yi=Yi/Li denotes output per worker, and hi=Hi/Li is human capital per worker. Data on stock of physical capital, workers and output are taken from the Penn World Tables. Regarding the level of productivity, it is calculated from the production function assuming that α=1/3 in line with standard neoclassical approach.21 Moreover, following Hall and Jones (1999), we consider human capital as a function of returns to schooling ( Hi=eϕ(Si)Li) as estimated in a Mincerian wage regression. ϕ(S) is a function reflecting the efficiency of a unit of labour with S years of schooling. Data on the number of years of schooling are taken from UNDP HDR statistics that rely on several sources: UNESCO Institute for Statistics, Barro and Lee (2013), ICF Macro Demographic and Health Surveys and UNICEF’s Multple Indicator Cluster. Each of the components on the right-hand side of Equation (10) contributes to the improvement of income per worker. The results of the 2SLS estimation of Equation (10) considering successively each term (respectively the log of income per worker, the log of the capital-output ratio, the log of human capital per worker, and the log of productivity) as the dependent variable are reported in Table 8. In columns (5–8), our results are not controlled by geographical and historical (or colonial) variables, while it is the case in columns (9–12). In line with our previous findings, an increase in trade between Africa and the developed economies significantly contributes to improve income per worker in Africa regardless of the specification. Considering the model without geographical and colonial controls, we found that trade between Africa and the industrialised economies positively affects income per worker in Africa mainly through an improvement in human capital per worker and higher productivity. Specifically, our estimates indicate that a rise of one percentage point (pp) in trade increases the contribution of the intensity of physical capital to income of 3/4 pp whereas it increases the contribution of productivity to income about two pp. These elasticities are close to those of Frankel and Romer (1999) with a more heterogeneous sample. When the complete specification is estimated (columns (9–12)), only productivity appears as the main transmission channel of the positive effect of trade on the standards of living, with an elasticity twice as high as before. Indeed, controlling for geographic and colonial regressors, we found that a one pp increase in trade between Africa and industrialised countries leads to an increase of the contribution of productivity to income per worker of four pp. This relatively high elasticity reflects to some extent the low initial level of productivity in African countries. Furthermore, in all configurations in which trade has a significant influence on income per worker or its components, the distance from equator significantly explains the differences in levels of income per worker between African countries. Moreover, our findings show that trade has not significantly helped to improve the contribution of capital intensity per worker to income. To sum up, if international trade with developed countries increases income in Africa, productivity is clearly the main transmission channel. Figure A1 in Appendix A shows the contents of Africa’s imports in skills and technology intensity. Imports from the developed OECD countries are clearly more equipped in skills and technology intensive, contributing to the improvement of productivity in Africa. Indeed, if the content in low skill of Africa’s imports is proportionally the same regardless of the partner, imports from the developed OECD countries are approximately twice as equipped with middle-skill and high-skill than those from Africa or other developing countries. These facts clearly support our empirical results. The latter also found several anchorage points in the literature even if the specific case of Africa has not been studied before. Indeed, the decisive role of trade openness in improving total factor productivity has been empirically emphasised by Edwards (1998). Miller and Upadhyay (2000) confirmed this result by showing that over and above the positive impact of trade openness on total factor productivity, outward-oriented countries also experience an improvement in total factor productivity. 5.2 Remittances and institutions While we do not find any evidence of growth-enhancing impact of openness to migration in Africa, it is possible that a potential benefice of emigration to developed countries is compensated by the adverse impact of brain drain. According to the literature on international migration, the positive effects of emigration operate through remittances (Giuliano and Ruiz-Arranz, 2009) and by improving political institutions (Spilimbergo, 2009; Docquier et al., 2016). We investigate here the relevance of these two channels by focusing on openness with developed countries, since (i) there is evidence of growth-enhancing effect of trade only with advanced economies, and (ii) most part of remittances received in Africa originate from developed countries. To examine whether remittances from African emigrants in developed countries improve economic growth, we regress output per capita (in log) on total remittances (in % of GDP),22 considering the same controls as above. The corresponding estimation results for the 2000 reference year are reported in column (1) of Table 9. As shown, we confirm the growth-enhancing impact of trade with developed countries, while remittances do not have a significant effect. The non-significance of remittances reflects the ambiguity about the link between remittances and growth. Indeed, on the one hand, remittances are perceived to overcome credit constraints and imperfections in financial markets, leading to higher rates of capital accumulation. On the other hand, if remittances are pure transfers aiming to sustain consumption of family members left behind, they may be detrimental for economic growth due to a moral hazard problem (Chami et al., 2003). Moreover, using remittances for final consumption may cause a Dutch-disease effect (Amuedo-Dorantes and Pozo, 2004). Our findings confirm those of Ahamada and Coulibaly (2013) showing that remittances do not promote growth in sub-Sahara African countries because they are not generally used for investment purposes. Our results are also in line with Bettin and Zazzaro (2012) who propose a new measure of efficiency of the domestic banking system and show that remittances promote growth only in countries whose banks are efficient, which is not the case in Africa. Our findings corroborates those of Konte (2017) as well who aims at understanding the difficulty to find an enhancing effect of remittances on growth despite (i) the important amount of those transfers in several developing countries, and (ii) the literature which has evidenced their positive impact on poverty and inequality. Relying on a recent finite-mixture-of-regressions approach, the author shows that the impact of remittances differs depending on the growth regime to which the considered country belongs. Table 9: Remittances and Institutions (1) (2) (3) (4) lnY/L Polity 2 Political rights Civil liberties Trade 4.20 ** −0.75 −0.42 −0.57 (1.99) (0.57) (0.62) (0.63) Emigration 2.21 4.39 * 3.56 * (2.23) (2.32) (2.09) Remittances 0.01 (0.26) Ln pop. −0.29 * −0.04 −0.01 −0.01 (0.17) (0.03) (0.03) (0.02) Ln area 0.15 (0.19) Dist. equator 0.04 −0.01 ** −0.00 −0.00 (0.03) (0.00) (0.00) (0.00) Constant 5.41 ** 0.76 *** 0.40 ** 0.44 *** (2.28) (0.16) (0.17) (0.16) Observations 48 49 50 50 Colonial/geo controls Yes Yes Yes Yes K-P F-stat 3.646 5.220 5.274 5.274 SW F-stat for Trade 5.501 10.80 10.74 10.74 SW F-stat for Emig. 8.698 8.398 8.398 SW F-stat for Rem. 6.827 SY 10% max IV size 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 (1) (2) (3) (4) lnY/L Polity 2 Political rights Civil liberties Trade 4.20 ** −0.75 −0.42 −0.57 (1.99) (0.57) (0.62) (0.63) Emigration 2.21 4.39 * 3.56 * (2.23) (2.32) (2.09) Remittances 0.01 (0.26) Ln pop. −0.29 * −0.04 −0.01 −0.01 (0.17) (0.03) (0.03) (0.02) Ln area 0.15 (0.19) Dist. equator 0.04 −0.01 ** −0.00 −0.00 (0.03) (0.00) (0.00) (0.00) Constant 5.41 ** 0.76 *** 0.40 ** 0.44 *** (2.28) (0.16) (0.17) (0.16) Observations 48 49 50 50 Colonial/geo controls Yes Yes Yes Yes K-P F-stat 3.646 5.220 5.274 5.274 SW F-stat for Trade 5.501 10.80 10.74 10.74 SW F-stat for Emig. 8.698 8.398 8.398 SW F-stat for Rem. 6.827 SY 10% max IV size 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 Notes: The predicted trade and emigration values are those based on the non-linear estimation. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 25% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Table 9: Remittances and Institutions (1) (2) (3) (4) lnY/L Polity 2 Political rights Civil liberties Trade 4.20 ** −0.75 −0.42 −0.57 (1.99) (0.57) (0.62) (0.63) Emigration 2.21 4.39 * 3.56 * (2.23) (2.32) (2.09) Remittances 0.01 (0.26) Ln pop. −0.29 * −0.04 −0.01 −0.01 (0.17) (0.03) (0.03) (0.02) Ln area 0.15 (0.19) Dist. equator 0.04 −0.01 ** −0.00 −0.00 (0.03) (0.00) (0.00) (0.00) Constant 5.41 ** 0.76 *** 0.40 ** 0.44 *** (2.28) (0.16) (0.17) (0.16) Observations 48 49 50 50 Colonial/geo controls Yes Yes Yes Yes K-P F-stat 3.646 5.220 5.274 5.274 SW F-stat for Trade 5.501 10.80 10.74 10.74 SW F-stat for Emig. 8.698 8.398 8.398 SW F-stat for Rem. 6.827 SY 10% max IV size 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 (1) (2) (3) (4) lnY/L Polity 2 Political rights Civil liberties Trade 4.20 ** −0.75 −0.42 −0.57 (1.99) (0.57) (0.62) (0.63) Emigration 2.21 4.39 * 3.56 * (2.23) (2.32) (2.09) Remittances 0.01 (0.26) Ln pop. −0.29 * −0.04 −0.01 −0.01 (0.17) (0.03) (0.03) (0.02) Ln area 0.15 (0.19) Dist. equator 0.04 −0.01 ** −0.00 −0.00 (0.03) (0.00) (0.00) (0.00) Constant 5.41 ** 0.76 *** 0.40 ** 0.44 *** (2.28) (0.16) (0.17) (0.16) Observations 48 49 50 50 Colonial/geo controls Yes Yes Yes Yes K-P F-stat 3.646 5.220 5.274 5.274 SW F-stat for Trade 5.501 10.80 10.74 10.74 SW F-stat for Emig. 8.698 8.398 8.398 SW F-stat for Rem. 6.827 SY 10% max IV size 7.030 7.030 7.030 7.030 SY 25% max IV size 3.630 3.630 3.630 3.630 Notes: The predicted trade and emigration values are those based on the non-linear estimation. Colonial controls are dummy variables for former French colony, former English colony and the share of population of European descent in 1900, and geographic control is a landlocked dummy. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. K-P F-stat is the Kleibergen and Paap (2006) rk Wald F-stat test of jointly weak identification. SW F-stat is the S F-stat test of weak identification for each endogenous regressor separately. In the case of a single endogenous regressor, the SW F-stat is identical to the K-P F-stat. SY 10% max IV size and SY 25% max IV size are the Stock and Yogo (2005) critical values under the i.i.d. assumption. View Large Finally, we investigate whether emigration to developed countries helps to promote institutional quality as notably highlighted by Spilimbergo (2009) and Docquier et al. (2016) who put forward the role of transfer of norms. To this end, we regress institutional quality on openness to trade with developed countries and emigration to advanced countries, using the same controls as above which are also perceived as influencing political institutions (Acemoglu et al., 2001). To proxy institutional quality, we rely on (i) the Polity 2 (P2) index drawn from the Polity IV database, and (ii) Political Rights (PR) and Civil Liberties (CL) indexes of Freedom House. P2 is a composite index based on the several concepts of democracy including (i) the presence of institutions and procedures through which citizens can express effective preferences about alternative policies and leaders, (ii) the existence of institutionalised constraints on the exercise of power by the executive, (iii) the guarantee of civil liberties to all citizens in their daily lives and in acts of political participation. P2 ranges from −10 (complete autocracy) to +10 (complete democracy). The PR index is based on a checklist of 10 political rights’ questions that are classified into three subcategories: electoral processes (3 questions), political pluralism and participation (4 questions), and functioning of the government (3 questions). The CL index is based on 15 civil liberties’ questions that are classified into four subcategories: freedom of expression and belief (4 questions), associational and organisational rights (3 questions), rule of law (4 questions), and personal autonomy and individual rights (4 questions). Both PR and CL checklists are translated into a rating ranging from 1 (most free) to 7 (least free). To make our results fully comparable across different measures of institutions—as in Acemoglu et al. (2008)—we normalise each measure on a scale ranging from 0 (complete autocratic/least free) to 1 (complete democratic/most free). The results for the year 2000 reported in Table 9 (columns (2–4)) show a positive influence of emigration to developed countries on political institutions that is significant only at the 10% level for the PR and CL indexes of Freedom House. This quite weak significance can reflect the fact that emigration, by lowering incentives, may delay democratic reforms (Docquier et al., 2016). 6. Conclusion This paper contributes to the debate on the growth-enhancing impact of openness by focusing on Africa. Examining the nature of the relationship both in a South-South and North-South perspective, we jointly consider openness to trade and to migration as vehicles of globalisation; both types of openness being instrumented by gravity-based predictors. Without distinguishing between partners, we show that international openness (both to trade and to migration) has a mitigated impact on per capita income in Africa; the positive effect of trade and migration being sensitive to the inclusion of control variables and the method used (linear or non-linear) to predict the geographic component of openness. In contrast, discriminating according to the partner allows us to obtain interesting and robust findings. Specifically, we establish that trade between Africa and the industrialised countries has a clear and significant positive impact on the standards of living in Africa, and show that such fostering effect operates through an improvement in total factor productivity. This growth-enhancing effect is consistent with the new trade theory suggesting that countries benefit from the advances in technological progress of their trading partners. Conversely, South-South openness fails to improve income in Africa regardless of the configuration (intra-African openness or openness between Africa and other developing countries). Turning to openness to migration, we do not find evidence of any growth-enhancing effect in Africa, whatever the partner country. South-South trade fails to improve the standards of living in Africa, probably because of the similarity of traded products and production techniques, which leave little room for learning between countries and technology transfer. Similarly, homogeneity of skills and institutions is likely to explain the failure of intra-African immigration to improve per capita income. Moreover, the lack of evidence of clear-cut positive effect of emigration from Africa to developed countries may result from two opposite effects at play in the African context: the negative effect of the ‘brain drain’ and the positive impact of the transfer of knowledge. On the whole, while our findings do not attribute a key role to openness to migration, they show that trade increases income in Africa, provided that the trade partner is an advanced country. In other words, our results underline that the characteristics of the trading partner—i.e., whether it is a developed or developing country—strongly matters in the trade-growth relationship. These findings highlight the importance of trade flows and trade-promoting policies between Africa and industrialised economies to foster African growth in a North-South perspective. Acknowledgements We would like to thank Christophe Destais, Sébastien Jean and participants at the EconomiX-CNRS and BETA-CNRS seminars for their remarks. We are also grateful to an anonymous referee and the Editor for very useful suggestions that helped us improve the paper. Footnotes 1 For a survey, see Edwards (1995) and Rodrik (1995) among others. 2 More precisely, openness to trade and openness to migration are jointly introduced in the income equation, being instrumented by the same geographical factors. 3 Recall that this framework is based on the gravity model of trade in which countries’ geographic characteristics are used to obtain instrumental variables estimates of trade effect on income. 4 See among others Edwards (1998) and Miller and Upadhyay (2000). See also our analysis in Section 5. 5 These transmission channels will be investigated in Section 5. 6 It would be interesting to reevaluate this effect in a few years (especially for China), when more—recent—observations will be available to better capture a potential medium to long-term growth-enhancing impact. 7 See the influential papers of Hall and Jones (1999), Acemoglu et al. (2001) and Rodrik et al. (2004). 8 We follow standard practice in measuring trade as imports plus exports divided by GDP (trade intensity ratio). An alternative would have been to examine import and export shares separately but, as shown by Frankel and Romer (1999) among others, the geographic components of the two shares are—not surprisingly—almost identical. Thus, the effects of imports and exports cannot be separated in the sense that the trade intensity ratio does not allow to distinguish countries having similar GDP and characteristics, but different trade-flows composition. However, in combining total trade (imports plus exports) and domestic economy, this ratio provides information regarding country’s involvement in trade—i.e., how important is trade for the considered country—which is the key information for our purpose (for a detailed discussion on trade indicators, see Ng, 2002). Nevertheless, for the sake of completeness and in order to disentangle the two channels through which trade with developed countries may promote growth in Africa—economies of scale and comparative advantages in the one hand, and transfer of technologies in the other hand—we will also consider the two ratios (exports over GDP and imports over GDP) separately (see Section 5). 9 See e.g. Caballero et al. (2008) and Mendoza et al. (2009). 10 The 52 countries are: Angola, Benin, Botswana, Burkina Faso, Burundi, Cabo Verde, Cameroon, Central African Republic, Chad, Comoros, Congo Dem. Rep., Congo Rep., Cote d’Ivoire, Djibouti, Egypt, Equatorial Guinea, Eritrea, Ethiopia, Gabon, Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Libya, Madagascar, Malawi, Mali, Mauritania, Mauritius, Morocco, Mozambique, Namibia, Niger, Nigeria, Rwanda, Sao Tome and Principe, Senegal, Seychelles, Sierra Leone, Somalia, South Africa, South Sudan, Sudan, Swaziland, Tanzania, Togo, Tunisia, Uganda, Zambia, and Zimbabwe. 11 Trade share for each country is calculated as the sum of its observed bilateral trade divided by GDP. Compared to the use of aggregate data, doing so allows us to specifically identity the partner country and, consequently, to account for potential different effects of openness depending on the type of the partner. 12 Frankel and Romer (1999) also emphasise the problem of accuracy in the estimation of this coefficient. 13 To save space, Figure 4 considers only gravity-predicted openness based on non-linear (PPML) estimation. 14 Note that this test constitutes a modification and improvement of the procedure described by Angrist and Pischke (2009). 15 For the sake of completeness and to disentangle immigration from and emigration to different types of countries separately, we have also regressed per capita income of African country i on emigration from this country i to other African countries and developing countries. The results—available upon request to the authors—corroborate those presented in Table 6 as we found that such type of emigration does not robustly impact growth. The case of immigration from advanced countries has not been considered, as such type of migration is very limited and, therefore, negligible. 16 This will be investigated further in Section 5. 17 See Section 5. 18 The six following countries are excluded from the African sample: Algeria, Djibouti, Egypt, Libya, Morocco and Tunisia. 19 To save space, we do not report the corresponding gravity estimates but they are available upon request to the authors. 20 The gravity-predicted values of exports and imports are considered as instruments in the corresponding regressions (see Table A1 in Appendix A). 21 The empirical value of this parameter we get for African countries is 0.314, which is very close to the reference value. 22 Data on bilateral remittances are not available. Remittances are taken from WDI and are defined as the sum of personal transfers and compensation of employees. The predicted values of trade with developed countries and emigration to developed countries are used as instruments. References Acemoglu D. , Johnson S. , Robinson J. ( 2001 ) ‘ The Colonial Origins of Comparative Development: An Empirical Investigation ’, American Economic Review , 91 : 1369 – 1401 . Google Scholar CrossRef Search ADS Acemoglu D. , Johnson S. , Robinson J. A. , Yared P. ( 2008 ) ‘ Income and Democracy ’, American Economic Review , 98 : 808 – 42 . Google Scholar CrossRef Search ADS Ahamada I. , Coulibaly D. ( 2013 ) ‘ Remittances and Growth in Sub-saharan African Countries: Evidence from a Panel Causality Test ’, Journal of International Development , 25 : 310 – 24 . Google Scholar CrossRef Search ADS Alesina A. , Spolaore E. , Wacziarg R. ( 2000 ) ‘ Economic Integration and Political Disintegration ’, American Economic Review , 90 : 1276 – 96 . Google Scholar CrossRef Search ADS Amuedo-Dorantes C. , Pozo S. ( 2004 ) ‘ Workers’ Remittances and the Real Exchange Rate: A Paradox of Gifts ’, World Development , 32 : 1407 – 17 . Google Scholar CrossRef Search ADS Angrist J. D. , Pischke J.-S. ( 2009 ) Mostly Harmless Econometrics: An Empiricist’s Companion . Princeton : Princeton University Press . Barro R. , Lee J. W. ( 2013 ) ‘ A New Data Set of Educational Attainment Dataset in the World, 1950–2010 ’, Journal of Development Economics , 104 : 184 – 198 . Google Scholar CrossRef Search ADS Beine M. , Docquier F. , Rapoport H. ( 2008 ) ‘ Brain Drain and Human Capital Formation in Developing Countries: Winners and Losers ’, Economic Journal , 118 : 631 – 52 . Google Scholar CrossRef Search ADS Bettin G. , Zazzaro A. ( 2012 ) ‘ Remittances and Financial Development: Substitutes or Complements in Economic Growth? ’, Bulletin of Economic Research , 64 ( 4 ): 509 – 36 . Google Scholar CrossRef Search ADS Caballero R. J. , Fahri E. , Gourinchas P. O. ( 2008 ) ‘ An Equilibrium Model of Global Imbalances and Low Interest Rates ’, American Economic Review , 98 ( 1 ): 358 – 93 . Google Scholar CrossRef Search ADS Chami R. , Fullenkampn C. , Jahjah S. 2003 . Are immigrant remittance flows a source of capital for development? IMF Working Paper 189. Docquier F. , Lodigiani E. , Rapoport H. , Schif M. ( 2016 ) ‘ Emigration and Democracy ’, Journal of Development Economics , 120 : 209 – 23 . forthcoming. Google Scholar CrossRef Search ADS Docquier F. , Özden Ç. , Peri G. , 2010 . The wage effects of immigration and emigration. Working Paper 16646. National Bureau of Economic Research. Dollar D. , Kraay A. ( 2003 ) ‘ Institutions, Trade, and Growth ’, Journal of Monetary Economics , 50 : 133 – 62 . Google Scholar CrossRef Search ADS Easterly W. , Levine R. ( 2001 ) ‘ It’s Not Factor Accumulation: Stylized Facts and Growth Models ’, World Bank Economic Review , 15 ( 2 ): 177 – 219 . Google Scholar CrossRef Search ADS Edwards S. ( 1995 ) ‘Trade policy, exchange rates, and growth’, in Dornbusch R. , Edwards S. (eds) , Reform, Recovery, and Growth: Latin America and the Middle East . Chicago : University of Chicago Press , pp. 13 – 49 . Edwards S. ( 1998 ) ‘ Openness, Productivity and Growth: What Do We Really Know? ’, Economic Journal , 108 : 383 – 98 . Google Scholar CrossRef Search ADS Frankel J. A. , Romer D. ( 1999 ) ‘ Does Trade Cause Growth? ’, American Economic Review , 89 : 379 – 99 . Google Scholar CrossRef Search ADS Frankel J. A. , Rose A. ( 2002 ) ‘ An Estimate of the Effect of Common Currencies on Trade and Income ’, Quarterly Journal of Economics , 117 : 437 – 66 . Google Scholar CrossRef Search ADS Freund C. , Bolaky B. ( 2008 ) ‘ Trade, Regulations, and Income ’, Journal of Development Economics , 87 : 309 – 21 . Google Scholar CrossRef Search ADS Gallup J. L. , Sachs J. D. , Mellinger A. D. ( 1999 ) ‘ Geography and Economic Development ’, International Regional Science Review , 22 : 179 – 232 . Google Scholar CrossRef Search ADS Giuliano P. , Ruiz-Arranz M. ( 2009 ) ‘ Remittances, Financial Development and Growth ’, Journal of Development Economics , 90 : 144 – 52 . Google Scholar CrossRef Search ADS Grossman G. , Helpman E. ( 1991 a) Innovation and Growth in the Global Economy . Cambridge, MA : MIT Press . Grossman G. , Helpman E. ( 1991 b) ‘ Trade, Knowledge Spillovers, and Growth ’, European Economic Review , XXXV : 517 – 26 . Google Scholar CrossRef Search ADS Hall R. E. , Jones C. I. ( 1999 ) ‘ Why Do Some Countries Produce so much more Output Per Worker than Others? ’, Quarterly Journal of Economics , 114 : 83 – 116 . Google Scholar CrossRef Search ADS He Y. ( 2013 ) ‘ Does China’s Trade Expansion Help African Development?—An Empirical Estimation ’, China Economic Review , 26 : 28 – 38 . Google Scholar CrossRef Search ADS Head K. , Mayer T. , Ries J. ( 2009 ) ‘ How Remote is the Offshoring Threat? ’, European Economic Review , 53 : 429 – 44 . Google Scholar CrossRef Search ADS Head K. , Mayer T. , Ries J. ( 2010 ) ‘ The Erosion of Colonial Trade Linkages After Independence ’, Journal of International Economics , 81 : 1 – 14 . Google Scholar CrossRef Search ADS Helpman E. , Krugman P. ( 1985 ) Market Structure and Foreign Trade . Cambridge, MA : MIT Press . Irwin D. A. , Terviö M. ( 2002 ) ‘ Does Trade Raise Income?: Evidence from the Twentieth Century ’, Journal of International Economics , 58 : 1 – 18 . Google Scholar CrossRef Search ADS Kaplinsky R. ( 2013 ) ‘ What Contribution Can China Make to Inclusive Growth in Sub-Saharan Africa? ’, Development and Change , 44 : 1295 – 1316 . Google Scholar CrossRef Search ADS Kleibergen F. , Paap R. ( 2006 ) ‘ Generalized Reduced Rank Tests Using the Singular Value Decomposition ’, Journal of Econometrics , 133 : 97 – 126 . Google Scholar CrossRef Search ADS Konte M. ( 2017 ) ‘ Do Remittances Not Promote Growth? A Finite Mixture-of-Regressions Approach ’, Empirical Economics , 1 – 36 . https://doi.org/10.1007/s00181-016-1224-z. La Porta R. , Lopez-de-Silanes F. , Shleifer A. ( 2008 ) ‘ The Economic Consequences of Legal Origins ’, Journal of Economic Literature , 46 : 285 – 332 . Google Scholar CrossRef Search ADS La Porta R. , Lopez-de-Silanes F. , Shleifer A. , Vishny R. ( 1999 ) ‘ The Quality of Government ’, Journal of Law, Economics, & Organization , 15 : 222 – 79 . Google Scholar CrossRef Search ADS Lyons M. , Brown A. ( 2010 ) ‘ Perception of Boom, Bust and the China-Africa Trade in Lomé and Bamako ’, World Development , 38 : 771 – 82 . Google Scholar CrossRef Search ADS Mendoza E. , Quadrini V. , Ríos-Rull J.-V. ( 2009 ) ‘ Financial Integration, Financial Development, and Global Imbalances ’, Journal of Political Economy , 117 ( 3 ): 371 – 416 . Google Scholar CrossRef Search ADS Miller S. M. , Upadhyay M. P. ( 2000 ) ‘ The Effects of Openness, Trade Orientation, and Human Capital on Total Factor Productivity ’, Journal of Development Economics , 50 : 133 – 62 . Ng F. ( 2002 ) ‘Trade indicators and indices’, in Hoekman B. , Mattoo A. , English P. (eds) , Development, Trade and the WTO, A Handbook . Washington, DC : The International Bank for Reconstruction and Development and World Bank , pp. 585 – 88 . Noguer M. , Siscart M. ( 2005 ) ‘ Trade Raises Income: A Precise and Robust Result ’, Journal of International Economics , 65 : 447 – 60 . Google Scholar CrossRef Search ADS Ortega F. , Peri G. ( 2014 ) ‘ Openness and Income: The Roles of Trade and Migration ’, Journal of International Economics , 92 : 213 – 51 . Google Scholar CrossRef Search ADS Rodriguez F. , Rodrik D. ( 2000 ) ‘ Trade Policy and Economic Growth: A Skeptic’s Guide to the Cross-national Evidence ’, NBER Macroeconomics Annual , 15 : 261 – 338 . Google Scholar CrossRef Search ADS Rodrik D. ( 1995 ) ‘Trade and industrial policy reform’, in Behrman J. , Srinivasan T. N. (eds) , Handbook of Development Economics , Vol. 3B . Amsterdam : North-Holland , pp. 2925 – 82 . Rodrik D. , Subramanian A. , Trebbi F. ( 2004 ) ‘ Institutions Rule: The Primacy of Institutions Over Geography and Integration in Economic Development ’, Journal of Economic Growth , 9 : 131 – 65 . Google Scholar CrossRef Search ADS Rose A. ( 2000 ) ‘ One Money, One Market? The Effect on Common Currencies on International Trade ’, Economic Policy , XXX : 7 – 45 . Rose A. ( 2001 ) ‘ Currency Unions and Trade: The Effect is Large ’, Economic Policy , XXXI : 433 – 62 . Sachs J. D. , Warner A. M. ( 1997 ) ‘ Fundamental Sources of Long-run Growth ’, American Economic Review , 87 ( 2 ): 184 – 8 . Sanderson E. , Windmeijer F. ( 2015 ) ‘ A Weak Instrument F-test in Linear IV Models with Multiple Endogenous Variables ’, Journal of Econometrics , 190 ( 2 ): 212 – 21 . Google Scholar CrossRef Search ADS Silva J. M. C. S. , Tenreyro S. ( 2006 ) ‘ The Log of Gravity ’, Review of Economics and Statistics , 88 : 641 – 58 . Google Scholar CrossRef Search ADS Silva J. M. C. S. , Tenreyro S. ( 2010 ) ‘ On the Existence of the Maximum Likelihood Estimates in Poisson Regression ’, Economic Letters , 107 : 310 – 2 . Google Scholar CrossRef Search ADS Spilimbergo A. ( 2009 ) ‘ Foreign Students and Democracy ’, American Economic Review , 99 : 528 – 43 . Google Scholar CrossRef Search ADS Stark O. , Helmenstein C. , Prskawetz A. ( 1997 ) ‘ A Brain Gain with a Brain Drain ’, Economics Letters , 55 : 227 – 34 . Google Scholar CrossRef Search ADS Stock J. H. , Yogo M. ( 2005 ) ‘Testing for weak instruments in linear IV Regression’, in Stock J. H. , Andrews D. W. K. (eds) , Identification and Inference for Econometric Models: Essays in Honor of Thomas J. Rothenberg . Cambridge : Cambridge University Press , pp. 80 – 108 . Chapter 5. Google Scholar CrossRef Search ADS Yanikkaya H. ( 2003 ) ‘ Trade Openness and Economic Growth: A Cross-country Empirical Investigation ’, Journal of Development Economics , 72 : 57 – 89 . Google Scholar CrossRef Search ADS Figure A1: View largeDownload slide Composition of Africa’s Imports. Notes: Data are extracted from UNCTAD and correspond to mean values over the 1995–2014 period. Figure A1: View largeDownload slide Composition of Africa’s Imports. Notes: Data are extracted from UNCTAD and correspond to mean values over the 1995–2014 period. Table A1: Gravity Regression, African Openness with Industrialised Countries Exports Imports OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.90 *** −0.78 *** −0.69 *** −0.77 *** (0.18) (0.15) (0.15) (0.22) Ln pop. origin −0.05 −0.34 ** −0.20 *** −0.20 *** (0.06) (0.14) (0.07) (0.06) Ln pop. dest. 1.43 *** 1.11 *** 1.65 *** 1.12 *** (0.05) (0.10) (0.06) (0.08) Ln area origin −0.03 0.21 *** −0.01 −0.00 (0.04) (0.07) (0.05) (0.04) Ln area dest. −0.34 *** −0.18 −0.43 *** −0.40 *** (0.04) (0.13) (0.05) (0.06) Sum landlocked −1.09 *** −0.82 *** −1.00 *** −0.89 *** (0.12) (0.23) (0.14) (0.11) Common language 0.78 *** 0.74 ** 1.03 *** 0.25 (0.24) (0.38) (0.40) (0.34) Common off. lang. 0.08 −0.12 −0.23 0.38 (0.22) (0.31) (0.39) (0.32) Colonial ties 2.46 *** 1.37 *** 2.77 *** 1.46 *** (0.27) (0.37) (0.25) (0.21) Hegemony −2.08 *** −2.57 *** −3.05 *** −2.80 *** (0.34) (0.40) (0.41) (0.38) Time zone diff. −0.11 *** 0.05 * −0.12 *** 0.02 (0.03) (0.03) (0.04) (0.03) Constant −2.59 ** −2.30 −3.93 ** −0.22 (1.17) (1.93) (1.54) (1.45) Observations 1,892 2,300 1,936 2,300 R-squared 0.47 0.17 0.39 0.44 Exports Imports OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.90 *** −0.78 *** −0.69 *** −0.77 *** (0.18) (0.15) (0.15) (0.22) Ln pop. origin −0.05 −0.34 ** −0.20 *** −0.20 *** (0.06) (0.14) (0.07) (0.06) Ln pop. dest. 1.43 *** 1.11 *** 1.65 *** 1.12 *** (0.05) (0.10) (0.06) (0.08) Ln area origin −0.03 0.21 *** −0.01 −0.00 (0.04) (0.07) (0.05) (0.04) Ln area dest. −0.34 *** −0.18 −0.43 *** −0.40 *** (0.04) (0.13) (0.05) (0.06) Sum landlocked −1.09 *** −0.82 *** −1.00 *** −0.89 *** (0.12) (0.23) (0.14) (0.11) Common language 0.78 *** 0.74 ** 1.03 *** 0.25 (0.24) (0.38) (0.40) (0.34) Common off. lang. 0.08 −0.12 −0.23 0.38 (0.22) (0.31) (0.39) (0.32) Colonial ties 2.46 *** 1.37 *** 2.77 *** 1.46 *** (0.27) (0.37) (0.25) (0.21) Hegemony −2.08 *** −2.57 *** −3.05 *** −2.80 *** (0.34) (0.40) (0.41) (0.38) Time zone diff. −0.11 *** 0.05 * −0.12 *** 0.02 (0.03) (0.03) (0.04) (0.03) Constant −2.59 ** −2.30 −3.93 ** −0.22 (1.17) (1.93) (1.54) (1.45) Observations 1,892 2,300 1,936 2,300 R-squared 0.47 0.17 0.39 0.44 Notes: The dependent variables ‘Exports’ and ‘Imports’ stand, respectively, for the sum of bilateral exports divided by GDP and the sum of bilateral imports divided by GDP. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large Table A1: Gravity Regression, African Openness with Industrialised Countries Exports Imports OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.90 *** −0.78 *** −0.69 *** −0.77 *** (0.18) (0.15) (0.15) (0.22) Ln pop. origin −0.05 −0.34 ** −0.20 *** −0.20 *** (0.06) (0.14) (0.07) (0.06) Ln pop. dest. 1.43 *** 1.11 *** 1.65 *** 1.12 *** (0.05) (0.10) (0.06) (0.08) Ln area origin −0.03 0.21 *** −0.01 −0.00 (0.04) (0.07) (0.05) (0.04) Ln area dest. −0.34 *** −0.18 −0.43 *** −0.40 *** (0.04) (0.13) (0.05) (0.06) Sum landlocked −1.09 *** −0.82 *** −1.00 *** −0.89 *** (0.12) (0.23) (0.14) (0.11) Common language 0.78 *** 0.74 ** 1.03 *** 0.25 (0.24) (0.38) (0.40) (0.34) Common off. lang. 0.08 −0.12 −0.23 0.38 (0.22) (0.31) (0.39) (0.32) Colonial ties 2.46 *** 1.37 *** 2.77 *** 1.46 *** (0.27) (0.37) (0.25) (0.21) Hegemony −2.08 *** −2.57 *** −3.05 *** −2.80 *** (0.34) (0.40) (0.41) (0.38) Time zone diff. −0.11 *** 0.05 * −0.12 *** 0.02 (0.03) (0.03) (0.04) (0.03) Constant −2.59 ** −2.30 −3.93 ** −0.22 (1.17) (1.93) (1.54) (1.45) Observations 1,892 2,300 1,936 2,300 R-squared 0.47 0.17 0.39 0.44 Exports Imports OLS PPML OLS PPML (1) (2) (3) (4) Ln distance −0.90 *** −0.78 *** −0.69 *** −0.77 *** (0.18) (0.15) (0.15) (0.22) Ln pop. origin −0.05 −0.34 ** −0.20 *** −0.20 *** (0.06) (0.14) (0.07) (0.06) Ln pop. dest. 1.43 *** 1.11 *** 1.65 *** 1.12 *** (0.05) (0.10) (0.06) (0.08) Ln area origin −0.03 0.21 *** −0.01 −0.00 (0.04) (0.07) (0.05) (0.04) Ln area dest. −0.34 *** −0.18 −0.43 *** −0.40 *** (0.04) (0.13) (0.05) (0.06) Sum landlocked −1.09 *** −0.82 *** −1.00 *** −0.89 *** (0.12) (0.23) (0.14) (0.11) Common language 0.78 *** 0.74 ** 1.03 *** 0.25 (0.24) (0.38) (0.40) (0.34) Common off. lang. 0.08 −0.12 −0.23 0.38 (0.22) (0.31) (0.39) (0.32) Colonial ties 2.46 *** 1.37 *** 2.77 *** 1.46 *** (0.27) (0.37) (0.25) (0.21) Hegemony −2.08 *** −2.57 *** −3.05 *** −2.80 *** (0.34) (0.40) (0.41) (0.38) Time zone diff. −0.11 *** 0.05 * −0.12 *** 0.02 (0.03) (0.03) (0.04) (0.03) Constant −2.59 ** −2.30 −3.93 ** −0.22 (1.17) (1.93) (1.54) (1.45) Observations 1,892 2,300 1,936 2,300 R-squared 0.47 0.17 0.39 0.44 Notes: The dependent variables ‘Exports’ and ‘Imports’ stand, respectively, for the sum of bilateral exports divided by GDP and the sum of bilateral imports divided by GDP. Heteroskedasticity-robust standard errors are in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% confidence level, respectively. View Large © The Author(s) 2018. Published by Oxford University Press on behalf of the Centre for the Study of African Economies, all rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Journal of African EconomiesOxford University Press

Published: Aug 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off