Exports have continued to increase dramatically from low income countries in the past couple of years, as a result of the increasing role played by exports. In many countries, especially those in East Asia, economic growth led by exports has successfully elevated millions from the poverty trap through employment creation. Yamagata (2006)’s study on Cambodia, concluded by confirming that  the  new  export  orientated  garments  industry  has substantially increased employment and reduce poverty. A similar story is told by Kabeer and Mahmud (2004) in their study in Bangladesh and supported by Razafimahefa (2005) while researching on the role of export-oriented manufacturing in reducing poverty in Madagascar. Many other studies (Dizaji and Badri, 2014; Sousa et al., 2012)   have arrived at the same conclusion supporting the notion that export oriented manufacturing can increase employment. In addition, some studies (Hesse, 2008; Agosin, 2007; Lederman and Maloney, 2007) have shown that export expansion may also contribute indirectly to economic growth by providing the much needed foreign exchange to pay for strategic exports.

Recently, in the discussion of export – led growth strategies, focused have been shifted to the growing role of the composition and diversification of the export bases of developing economies. The changing composition and diversification of exports may have important implication on employment, poverty reduction and overall economic development, especially as a majority of low income countries are now involved in divergence, while others are still in the course. This certainly means that there is great disproportion in experiences involving diversify-cation. In addition to the focus by low income countries, many international institutions among which include the World Bank, the United Nations and the OECD, have also joined their voices to outline the potential benefits of export diversification. Furthermore, a number of studies including Lederman and Maloney (2007), Herzer and Nowak-Lehmann (2006) and Ghosh and Ostry (1994) have also noted a number of benefits accruing to economies with diversified export bases. These benefits among others include a lower terms of trade volatility and increased macroeconomic stability. In other studies outlining additional benefits of diversification, Agosin (2007) and Hesse (2008) concluded that economies which diversified their export bases also witnessed higher income growth rates, while Lederman and Maloney (2007) showed that export concentration correlates negatively with economic growth. Export diversification has also been found to contribute to export growth especially in low income nations, for instance, Brenton and Newfarmer (2007) found that export diversification accounted for 57% of the total export growth in their study of some African nations.

From the forgoing discussion, it is therefore not surprising if several countries especially those of developing and emerging countries are striving to broaden their export bases. The interesting issue in this new perspective is that Foreign Direct Investment (FDI) which may not only contribute directly to economic benefits, but which can have great implications on export diversification or broadening has been overlooked. In the past decades, many low income countries have experienced large inflow in FDI and are putting in place competitive social and political conditions to  continuously pull FDI. According to Newfarmer et al. (2009), the strategies adopted by some countries, notably Kenya, Botswana, Cambodia for diversification and promotion of exports directly cite the key role that FDI can play in boosting and developing new industries engaged in exportation. In Kenya for instance, the reform on export broadening focuses on a diversion from primary production to the improvement in the quantity and quality of manufactured goods (International Trade Centre, 2001). Vietnam has also transformed its economy from a low-end agricultural exporter to a successful middle-range manufacturing exporter in less than two decades, while Tanzania and Uganda have also undergone significant export diversification in the last 20 years.

In terms of indirect effect, Banga (2003) argues that FDI can foster the growth of the exportation base through the spillover effects on the level of export of the home grown firms who are involved in the traditional export sector. This spillover effect generated from the fact that the export of foreign firms in this sector lowers the fixed cost of introducing the sector’s product in the foreign market. In addition to lowering the fixed cost, the home-firms may also learn and gain experience from the foreign firms and become knowledgeable about the existence of new markets and openings in the foreign markets. All these put together will promote the export diversification base of the home economy.

It is therefore evident from the above that FDI can potentially play a great role in influencing export diversi-fication. Countries that have credit their diversification as being driven by FDI flows include Costa Rica, Mauritius and Chile. Costa Rica and Mauritius partly attribute their diversification in electronics industry on this. Some specific case studies of instances where FDI helped develop new export industries have been documented in many countries including India (Banga, 2003) and Bangladesh (Rhee, 1990). While large bodies of literature have examined the drivers of export diversification; the importance of export diversification and the benefits of FDI; only a few have explored the links between FDI and export diversification, especially empirically. Thus, it is imperative to investigate the effect of FDI on export diversification. This paper is therefore aimed at investigating the effect of FDI on export diversification on the economies in the CEMAC Region using a fractionalized logit model estimation technique.

The rest of the paper is divided into four sections. The section on literature review makes a review of existing literature on export diversification with a focus on key factors influencing the growth of export diversification. In the section for empirical model specification and estimation, the empirical estimation methodology adopted in this study is explained, while the section for presentation and discussion of results presents the stylized facts existing between FDI and export diversification and also presents the results obtained from the empirical analyses. The last section draws conclusion with focus on the policy implications of the results.

The literature on export diversification is quite vast and focus on issues dealing with measurement of diversification (Cottet et al., 2012; Hausmann and Klinger, 2006; Rodrik, 2006), pattern of export diversification (Amurgo-Pacheco and Pierola, 2008; Hummels and Klenow, 2005), the effect of export diversification on growth (Klinger and Lederman, 2006; Hesse, 2008; Cadot Strauss-Kahn and Carrere, 2011) and the determinants of export diversification (Agosin et al., 2011; Changbiao, 2009; Jayaweera, 2009; Munemo, 2007). Other groups of studies have shown that export diversification can potentially help stabilize the export earnings of countries in the long run. Among these studies include those of Ghosh and Ostry (1994) and Bleaney and Greenaway (2001).

Recently, some studies (Hesse, 2008; Cadot et al., 2011) have shown that diversifying the exportation base can be directly related to improvement in per capita income. According to these studies, an increase in per capita income will result to an increase in export diversification, though this may only be up to a given level of income. For instance, in the study of Hesse (2008), a system Generalized Methods of Moment estimation technique was employed to investigate the effect of export concentration on economic growth in a sample of 99 countries. The study included a squared term for the Herfindahl-Hirschman Index of concentration to investigate the existence of non-linearity between growth and diversification. The results obtained showed some evidence of a non-linear relationship existing between concentration of export and per capita income growth, albeit the coefficient on the squared term was statistically insignificant. However, in an earlier study by Cadot, Strauss-Kahn and Carrere (2011), they found out that the relationship between per capital income and diversification was U-shaped supporting the view that countries tend to diversify their export bases as they grew from low income to middle income nations, but then concentrated their exports after reaching a high income level. The results obtained supported those obtained by Imbs and Wacziarg (2003) that also suggested the existence of a U-shaped between concentration and per capita income.

Al-Marhubi (2000) in a conventional cross sectional country growth regression using different measures of export diversification/concentration to link to the standard growth equation, found out that export diversification promotes economic growth. These findings were robust and consistent to the different models specified. In a similar cross-sectional study carried out by Agosin (2007), the results showed that the effect of export diversification on per capita income growth  was  stronger only when a country’s exports grew faster than the per capita income growth. Other studies, notably Herzer and Nowak-Lehmann (2006) and Lederman and Maloney (2007) also found evidence that supports the fact that export diversification has a positive influence on economic growth.

A new strand of literature considers the role played by the pattern of export diversification, that is, it investigates whether export growth is predominantly motivated by growth at the extensive or intensive margin. In terms of extensive margin growth, countries export a wider set of products to existing or new geographical markets whereas under intensive margin growth, an increase of existing products to current markets occurs. Hummels and Klenow (2005) in their study using data on shipments by 126 exporting countries to 59 importing countries in 5,000 product categories; showed that the extensive margin accounts for around 60 percent of the greater exports of larger economies. This result was supported by those of Pham and Martin (2007) using a cross-sectional analysis who found out that about 70 percent of the growth in exports in their sample of 120 exporters and 76 importers was explained by extensive margin growth. The results obtained by Hummels and Klenow (2005) and by Pham and Martin (2007) were in contrast with the results of the time series analyses employed by Brenton and Newfarmer (2007) and Amiti and Freund (2006). Both results showed low rates of extensive margin growth over time suggesting that exporting larger quantities of existing products matter more than exporting a wider set of products. The results also point to the fact that exporting existing products to new geographical markets carries a higher weight in explaining export growth than discovery of new products. The conflicting results on the role of the pattern of diversification may be as a result of the different methodology and set of countries used in the different analyses.

In terms of determinants, Munemo (2007) analyzed the effect of foreign aid on export diversification using panel data from 69 developing countries employing the instrumental variables (IV) estimation technique. The findings of the study indicated that foreign aid had a negative effect on export diversification. On his part, Changbiao (2009) used a panel data to examine the determinants of exports in Chinese electronics industry for the period 1999 to 2002. The study showed that FDI was a significant and positive determinant of export growth in China. Similarly, by considering FDI as one of the determinants of export diversification, Jayaweera (2009) built an econometric model of instrumental variables to estimate the relationship between FDI and export diversification for a panel of 29 low income nations for the period 1990-2006. The findings showed that FDI had a positive impact on export diversification.

Using a panel data-set for 60 countries from 1985-2004, Parteka and Tamberi (2011) assessed the role played  by   country   specific  factors  in  determining  the exports diversification process by applying different synthetic indices of specialization. Their findings revealed that countries located far from the economic core of the world and those for which barriers to trade are large tend to have less diversified manufacturing exports. In a similar study, Agosin et al. (2011) analyzed the determinants of export diversification around the world from 1962 to 2000, employing the Herfindahl et al. indices as measures of export diversification. Their data were analyzed using the Generalized Method of Moments (GMM) and the findings showed that human capital accumulation contributed positively to diversify exports while real exchange rate volatility encouraged export concentration.

Empirical model specification and estimation

Within a panel data setting in which a cross sectional unit has relatively smaller time periods, the empirical model to be estimated in this study is generally stated as:

Where; EXDIV=Export Diversification index is measured using the World Bank’s measure of the Herfindahl-Hirschman Concentration Index. The index is a measure of the degree of market concentration. It has been normalized to obtain values ranking from 0 to 1. An index value that is close to 1 indicates a very concentrated market (maximum concentration). On the contrary, values closer to 0 reflect a more equal distribution of market shares among exporters or importers. In other words, when the value of the Herfindahl-Hirschman Index approaches one, the country has a greater reliance on a limited group of exports, while a value closer to zero represents a higher degree of export diversification.

Thus, it can be considered as an indicator of the exporter’s vulnerability to trade shocks and since it is measured over time, a fall in the index may be an indication of diversification in the exporter’s trade profile.

This index is computed using the formula: 

Where:

HHI= Herfindahl-Hirschman Index (proxy for diversification)

 = value of exports of product i 

 and n = number of products

The choice of this measure of diversification is contingent on the fact that it helps determine whether the majority of a country’s export earnings comes from a small range of export products, (which is more an indication of export concentration) or the source of export earnings are more evenly spread across a given range of export goods (an indication of export diversification).

 is a vector of regression coefficients to be estimated, X_it is a matrix of regressors which are; Foreign Direct Investment Net Inflow (LFDI), trade openness (LOPEN), rents from natural resource endowment (LRENT); Real Gross Domestic Product per capita (LGDPPC), official exchange rate (LOER) and manufacturing value added (LMVA). All these variables were logged and a detailed definition of the variables is described in Table A1 in the Appendix.

The linear structure of equation (1) is thus specified as;

In estimation of the above model, though the Ordinary Least Squares (OLS) estimator, the Instrumental Variable estimator, the panel data estimator or even the GMM of a linear model may yield consistent estimates even by ignoring the bounded nature of the dependent variable, they however, do not guarantee that their fitted values lie within the unit interval nor that their partial effect estimates for the regressors’ extreme values are good (Nam, 2014). Thus, an additional innovation about this study hinges on the fact that the estimation technique used is the fractionalized logit model which counters the weakness of the former techniques of estimation by considering the fractional nature of the concentration index. We adopt the fractional logit model after Papke and  Wooldridge   (1996)   which   is    a   quasi-likelihood method that does not assume any distribution but only requires the conditional mean to be correctly specified for consistent parameter estimates with an identical likelihood function similar to that of a Bernoulli distribution. This was implemented in Stata 12.0 through the Generalized Linear Model (GLM), using the logit link function and the Gaussian distribution family to take care of the fractional nature of the dependent variable (concentration index) which ranges between [0,1].

The data used in this study are panel data collected for a  period   of   19  years   (1995 - 2013)   for    six   countries (Cameroon, Central Africa Republic (CAR), Chad, Gabon, Equatorial Guinea and Congo). Most of the data were obtained from the World Development Indicators databank (Table A1 on definition of the variables). The concentration index was obtained from the United Nations Conference on Trade and Development Statistics (UNCTADstat), which produces a comprehensive data of more than 150 indicators and statistical time series essential for the analysis of International trade, economic trends, foreign direct investment, external financial resources, population and labor force, commodities, information economy, creative economy and maritime transport. From the data obtained a summary statistics of the variables used in the estimation is presented in Table A2 in the Appendix by country. The summary table shows the mean of the variables used in the model, while the pairwise correlation results are found in Table A3.

From the table, the average on trade diversification is 0.65 for the CEMAC zone, while in terms of individual countries the statistics shows that the most diversified economies in the CEMAC zone are Cameroon (0.39), Central Africa Republic (0.45), Gabon (0.74), while the most concentrated are Chad (0.80), Equatorial Guinea (0.76) and Congo (0.75). In terms of FDI as a percentage to GDP, the average in the sub region is 7.9 percent, while Equatorial Guinea, Chad and Gabon received a lion share of the FDI in the region with 23.14, 12.81 and 7.38%, respectively. The average GDP per capita in the region stands at $3202.64, with Equatorial Guinea, Gabon and Congo topping the list with $9069.82, $6627.09 and $1714.96, respectively, while CAR has the least per capita income ($365.29) followed by Chad ($542.46) and Cameroon ($896.23). Looking at trade interaction with the rest of the world measured using trade openness, the average for the region stands at 109.53, while the most open economies in the Sub Region are Equatorial Guinea (271.71), Congo (136.11) and Gabon (92.82), Chad (74.18), Cameroon (42.78) and CAR (39.59). Concerning total rents obtained from the stock of natural resource endowment, the summary statistics shows an average of 31.67 as a percentage to GDP for the region, while Congo tops the list with 63.14%, followed by Gabon (46.22%) and Equatorial Guinea (33.08%)

The summary fails to give a clue of the key relationship that may exist between FDI and the export diversification for the six countries in the CEMAC zone. A  preliminary analysis using a pair wise correlation for the variables is presented in Table A3 in the Appendix, while a scatter plot to better understand the association between these two variables is presented in Figure A1 in the Appendix, for each country. The scattered plot for FDI as a percentage to GDP and the concentration index measured by the Herfindahl Index, for the various economies clearly shows that except for Congo, a negative relationship exists between the concentration index and FDI as  a  percentage  to  GDP.  This  result  is supported by the pairwise correlation matrix presented in Table A4 between FDI and the Herfindahl concentration index for the various economies. The implication from these results is very clear indicating that an increase in FDI as a percentage to GDP is associated with a decrease in the concentration index. In other words an increase FDI as a percentage to GDP promotes diversification in the export bundles. It is also evident from Figure A1 and Table A4, that this relationship is not the same for all the economies. This is supported by the empirical results presented in Table 1. 

The standard errors of the results presented in Table 1 are robust to control for potential heteroscedasticity that may result from misspecification. The results presented in column one were implemented controlling for the panel nature of the data composed of six countries (clusters) over a period of 19 years, while the rest of the results were estimated for each country. Worth mentioning that Equatorial Guinea and Gabon had only 9 and 13 observations, respectively due to missing values for key variables included in the estimation, as observed in the table below.

The results presented in column one clearly showed that a majority of the variables used in the analysis, albeit for real GDP per capita are statistically significant. The results indicate that foreign direct investment, openness, real GDP per capita and value added in manufacturing have a negative effect on concentration, while the official exchange rate and rents from natural resource endowment were found to have a positive effect on export concentration. The negative relation between concentration and the respective variables implies that an increase in the value of these variables will result to de-specialization or diversification, while the positive relationship indicates a movement towards more specialization or concentration.

Specifically, the results show that on the one hand, an increase in foreign direct investment, trade openness and value added in manufacturing will increase the likelihood of a country to diversify. The effects of these variables are statistically significant at various levels as shown in Table 1. The results from column one for the entire CEMAC region show that an increase in Foreign Direct Investment (% GDP), trade openness, GDP per capita and value added in manufacturing by one percent will result to an increase in the likelihood to diversify the exportation bundle in the CEMAC region by 0.0843, 0.2712, 0.0805 and 0.4058 percent, respectively. On the other hand, a percent increase in the official exchange rate and the rent from natural resource endowment will increase the likelihood to concentrate (i.e. likelihood to diversify is reduced) by 0.6450 and 0.1151 percent, respectively, everything being equal.

Other results presented in column two to seven for the individual economies are in support of the results presented in column one for the entire sub region, especially   when   the   signs   of   the    coefficients   are considered. Exceptions that are quite visible are observed with Chad and Equatorial Guinea where real GDP per capita has a positive and statistically significant effect on the concentration index, implying that an increase in per capita income in these two economies will result to an increase in the likelihood to concentrate (i.e. de-specialization).  In terms of country specificity, the results obtained for Congo are quite different from those of the other countries in this region. For instance, the effect of FDI, GDP per capita income and value added in manufacturing are positive contrary to the results obtained for the entire region, Cameroon, CAR and Gabon. The positive effect observed for per capita income is however consistent with the results obtained for Equatorial Guinea and Chad.

The results obtained in this study put together showed that on average an increase in foreign direct investment as a percentage to GDP will encourage export diversification in the CEMAC region. In same light, trade openness, value added in manufacturing, increase in real GDP per capita also promotes export diversification in the region. The results also point to the fact that an increase in official exchange rate and the rents from natural resource endowment will result to export concentration. In this sub region, it is quite evident from the analysis that the effect of foreign direct investment differs across the countries in this region with the greatest influence observed in Cameroon with a FDI coefficient of 0.2763.

The  results   of   this  study  are  consistent  with  some previous findings. Specifically, the negative effect of per capita income on export diversification in the CEMAC sub region is contrary to the results obtained by Kamuganga (2012) and Cabral and Veiga (2010) who found a positive and statistical significant effect of GDP per capita on export diversification. However, the result is consistent to those of Elhiraika and Mbate (2014) and Parteka and Tanberi (2011) who found a negative and statistically significant effect of per capita on export diversification. The effect of FDI is consistent with those of Bebczuk and Berrettoni (2006) though they found FDI to have a negative and statistical insignificant effect on export diversification, as opposed to this study which shows that on average FDI has a negative and statistical significant effect on export concentration thereby promoting diversification in this region.

The effect of the manufacturing sector is consistent with those of Bebczuk and Berrettoni (2006), though our result is statistically significant. The result on the effect of trade openness is contrary to the results obtained by many of the previous studies (Omgba, 2013; Kamuganga, 2012 and Agosin et al., 2011) who all found that trade openness had a positive and significant effect on export concentration, meaning that openness discourages diversification as opposed to the results in this study which showed that trade openness promotes diversify-cation. However, Alaya (2012) in their study of 12 Middle East and North Africa countries found results consistent with  those   in   this   study,   which   showed   that  trade openness promotes export diversification. The result on the impact of rents from natural resources is consistent with Lederman and Maloney (2007)’s argument that there is a tendency to export concentration in resource-abundant countries. This result is however not consistent among the economies of the sub region. A negative and statistical significant effect was obtained for Equatorial Guinea and Gabon, while an insignificant effect was found for CAR and Chad. The result seems to point to the fact that natural resource based economies can still foster export diversification as witnessed by economies such as Uganda, Chile, Malaysia and Thailand (Newfarmer et al., 2009) 

This study aimed at investigating the export-diversifying effect of FDI as a percentage of GDP on the economies of the CEMAC sub region – notably, Cameroon, Central Africa Republic, Chad, Congo, Equatorial Guinea and Gabon. Export diversification was measured using the Herfindahl-Hirschman Index which ranges between [0, 1] with 1 showing total concentration. The data were analyzed descriptively using pairwise correlation and scattered diagrams to examine the relationship existing between export diversification and FDI in the economies in this sub region. The empirical analyses were estimated using the fractionalized logit technique. The descriptive analyses revealed that there exists great disparity in the level of diversification and FDI. The average statistics showed that Cameroon is the most diversified with an average Herfindahl Index of 0.39, followed by Central Africa Republic (0.45), Gabon (0.74) Congo (0.75), Equatorial Guinea (0.76), while Chad is the most concentrated with an index of 0.8. In terms of FDI, Equatorial Guinea has the lion share in the region, followed by Congo and Chad.

The results of the empirical analysis revealed that overall diversification for the CEMAC sub region is significantly determined by foreign direct investment, trade openness, total rents from natural resource endowment, value added in manufacturing and Official Exchange Rate while GDP per capita was an insignificant determinant of export diversification. The results showed that an increase in FDI (% GDP) promotes diversification

in the composition of the export baskets, that is, it encourages de-specialization thereby promoting a more heterogeneous export structure while an increase in natural resource based rents tend to promote specialization (that is a homogenous export structure), which fosters the exportation of goods belonging to a limited productive economic sectors or products. In the region, ceteris paribus an increase in FDI by 10 percent for instance will result to a reduction in export concentration by about 0.843% (that is, improvement in the likelihood to diversify), while an increase in rents from natural resource endowment  by   10  percent  may  be  associated  with  a 2.71% decrease in the likelihood to diversify, that is, increase in the likelihood to concentrate the export bundle. Other results showed that an increase in trade openness and an improvement in the contribution of the manufacturing sector are instrumental in fostering export diversification in the CEMAC region as a whole.

The results obtained for the individual economies were not homogenous. It was found for instance that the magnitude and statistically significance of the effect of FDI was greatest in Cameroon which is the most diversified country in the region, though receiving one of the lowest FDI as percentage to GDP. This is as opposed to Congo and Equatorial Guinea that received the greatest share of FDI (12.81% and 23.14%, respectively) in this region, but are very concentrated (0.75 and 0.76 for Congo and Equatorial Guinea). Other key differences in the results that probably need further investigation for a larger set of countries with abundant natural resources showed that export diversification can still be fostered for abundant rich natural resource economies. This effect was statistically significant for Equatorial Guinea and Gabon, which are economies receiving large share of rents (33.078 and 46.217%, respectively) from natural resources.

The above results therefore have key policy implications, suggesting that one effective way for economies in the CEMAC sub region to promote export diversification is to focus regulatory reform efforts aimed at facilitating foreign direct investment and promoting trade openness. The results also point to the need for the manufacturing sector to be encouraged. Although, the results in this study shows that on average the level of export diversification is significantly lower in natural resource-rich countries, there is some evidence that diversification can still be enhanced in natural resource based dependent economies. This could be done by implementing a broad collection of targeted strategies, such as the creation of a well-adapted export promotion arrangement, reducing cost associated with trade such as import and export duties. There is also need to foster investment and trade freedom in the exploitation and transformation of the natural resources. This will in tend increase the value added in manufacturing which will go a long way to enhance economic growth and trade diversification. 

The authors have not declared any conflict of interests.