A Case Study of Three Country Groups
MSc Economics Track International Economics and Globalization
Tim Martens 10393218
Department of Economics Universiteit van Amsterdam Supervisor: Drs. Naomi Leefmans Second reader: Dr. K. Mavromatis February 2014
1.Introduction 1
2.Literaturereview 6
2.1 Theoretical literature on export diversification 7 2.2 Empirical literature on export diversification 9 2.2.1 Relationship between export diversification and productivity 9 2.2.2 Influence of export concentration on income, terms of trade and investments 10 2.2.3 Relationship between export diversification and economic growth 14 2.2.4 Determinants of export diversification 17
3.Datadescription 20
4.Methodology 22
4.1Determiningthecountrygroups 22
4.1.1Resource-richcountries 22
4.1.2 Oil-exporting countries and non-oil-exporting countries 23 4.1.3Resource-poorlandlockedcountries 23
4.1.4Resource-poorcoastalcountries 23
4.2Measurementofdiversification 24
4.3Herfindahl-HirschmanIndex(HHI) 26
4.4Hall-TidemanIndex(HTI) 27
4.5EntropyMeasure(ENT) 28
4.6CompositeIndex(CI) 29
4.7Numberofactiveexportlines 30
5.Empiricalanalysis 32
5.1Diversificationmeasures 32
5.2 The outcome of the weighted average normalized Herfindahl-Hirschman Index 32 5.3 The outcome of the weighted average normalized Entropy Measure 35 5.4 The outcome of the weighted average Hall-Tideman Index 38
5.6 The outcome of the weighted average number of active export lines 42
6.Conclusion 46
SSA Sub-Saharan Africa
IMF International Monetary Fund
OECD Organisation for Economic Cooperation and Development
GDP gross Domestic product
RRC resource-rich countries
OEC oil-exporting countries
NOEC non-oil-exporting resource-rich countries
RPCC resource-poor coastal countries
RPLC resource-poor landlocked countries
1. Introduction
The last years have seen substantial progress for the Sub-Saharan African (SSA) countries. The GDP growth in the SSA countries was 4.6% in 2012, while investments have increased from 15.9% of GDP in 2000 to over 22% of GDP in 2012 (World Bank, 2013a). Despite the fact that the medium-term growth of the Sub-Saharan Africa region has been strong, the debate over the quality of this development continues, especially in reference to the composition of the export basket. The export basket usually consists of a small range of export products, a fact that has consequences for the countries’ long-term growth prospects. In 2013 Sub-Saharan Africa is a region that exhibits a high dependency on primary products such as copper and iron. The export industry in various SSA countries is still based on a small number of primary products, which is in contrast to many other developing countries that have experienced a shift from primary commodities to manufacturing (Collier, 2002). For this reason export diversification has been a focal point in the discussion on how to improve the economic performance of the SSA countries.
Economic theory provides two theories that explain the specialization of countries in a small range of export products. David Ricardo’s theory of comparative advantage states that its ability to export allows a country to specialize in a sector that exhibits a comparative advantage and hence promotes growth. A country that produces a certain good with the lowest opportunity costs can trade the surplus. As a consequence, countries have an incentive to specialize in goods with a comparative advantage (Ricardo, 1817). An alternative theory, developed in 1985 by Elhanan Helpman and Paul Krugman, focuses on the dynamic effects of international trade. Accumulation of knowledge and technology improves productivity and economic performance. The theory states that the increase in efficiency encourages specialization (Helpman & Krugman, 1985).
Specialization in only few products can however have severe side effects on developing economies. In 1950 the economists Raúl Prebisch and Hans Singer developed the theory that developing countries that have a highly concentrated export basket with a focus on primary products may be affected negatively by income changes. Primary products generally exhibit low income elasticities of demand, while manufacturing products exhibit higher income elasticities of demand. A worldwide increase in income results in
developing countries that focus on primary products having to pay more for their imports (mainly non-primary products) and receiving relatively less for their exports (mainly primary products). This results in deteriorating terms of trade and a worsening economic outlook (Prebisch, 1950; Singer, 1950). A second severe side effect of specialization is connected with the volatility of export revenues. Due to a highly concentrated export basket, many developing countries are considerably affected by price fluctuations of their export goods. This volatility is caused by demand or supply shocks and may have negative consequences for the country’s economic performance. Especially the frequent non-existence of a capital market in developing countries amplifies the negative effects. In contrast to e.g. Mexico, which invests heavily in derivatives in order to offset oil-price fluctuations, most developing countries have no access to these instruments (NZZ, 2013). For this reason, the diversification of the export basket is seen as an approach to damp these negative effects and to start economic development by many international organizations and economists (Agosin, 2009).
The current debate however lacks an understanding of the different economic and geographic conditions of the various regions in SSA and their specific requirements with regard to the policy actions targeting their level of diversification. The area comprising SSA is very large and consists of 50 countries. Thus a detailed analysis has to take into account the geographic conditions of the individual countries in order to create robust results concerning the diversification capacities of the SSA countries.
Earlier research has found that the SSA countries have developed different levels of export diversification over the course of time. Bonaglia and Fukasaku (2003) investigated the diversification of export baskets and markets in low income countries for the period from 1966 to 2000. They examined the share of manufacturing in the export basket in order to measure the diversification in export products and the weight of the OECD imports in order to measure export diversification in export markets. Another study by Songwe and Winkler (2012) for the period from 1995 to 2008 uses the Herfindahl-Hirschman Index, which is based on the shares of the individual export products, to measure the export diversification of 30 SSA countries. Both studies have shown that the individual SSA countries have developed very different levels of export diversification concerning their
export baskets and export destinations.
Next to the diversification experience of the SSA countries, earlier research has found various determinants that influence the level of export diversification. Two geographic factors, in particular, appear to be very relevant for the export composition of developing countries: Natural resource endowment and access to the sea (Collier & O’Connell, 2007; Bebczuck & Berrettoni, 2006). Four groups of countries arise when the two factors are applied to analyses of African economies: rich landlocked countries, resource-rich coastal countries, resource-scarce landlocked countries and resource-scarce coastal countries. The resource-rich coastal countries and the resource-rich landlocked countries can be merged to one country group: resource-rich countries. This step is possible because it does not appear to be a major drawback if a resource-rich country is landlocked. In addition, the growth performance of resource-rich coastal countries does not differ significantly from resource-rich landlocked countries on a global scale.
In contrast, the performance of the resource-scarce coastal group differs significantly from that of the resource-scarce landlocked group. While the resource-scarce coastal countries have exhibited the best economic growth, the resource-scarce landlocked countries have shown the worst level of economic growth worldwide (Collier, 2006). This situation illustrates the need for three different strategies for the three different country groups in reference to export diversification and economic growth in SSA. To separate the resource-rich from the resource-scarce countries, this thesis applies the definition of the IMF, according to which resource-rich countries have on average either natural resource revenues or natural resource exports of at least 20% of total fiscal revenues and exports, respectively (IMF, 2012).
An analysis of the diversification process in SSA requires a definition of export diversification. Export diversification is the change in the composition of a country’s existing export product mix or export destination (Samen, 2010). This change covers horizontal diversification (diversification within the same sector) and vertical diversification (diversification across different sectors). The research in this study focuses on the description of the different levels of export diversification in the export product mix between the groups of resource-rich countries, coastal resource-poor countries and
landlocked resource-poor countries. Four different measures are used to determine the exact level of export diversification and identify the relevant parts of the diversification process (e.g. diversification in the smallest and the largest sector). Three of these measures are based on the individual export shares and provide results that can easily be compared to the results of other measures. An additional measure based on the number of active export lines is implemented to offset price effects.
This thesis aims at answering the following research question: Does the extent of export diversification in Sub-Saharan Africa from 1993 to 2010 differ between the groups of resource-rich, coastal resource-poor and landlocked resource-poor countries? The objective of this thesis is to analyze the development of export diversification in those groups of countries and to contribute to the understanding of the determinants of export diversification in SSA. The thesis should identify a clear diversification trend between the country groups that can be explained by two geographic determinants (natural resources, sea access). In contrast to earlier research, this thesis focuses on the influence of resource-endowment and geographic location factors while it offsets the impact of other influencing factors due to the assumption that the grouping leads to an equal distribution of other factors. The result is a detailed picture of the quantitative impact, in terms of diversification measures, of the two determinants on the export basket in SSA. The thesis contributes to the discussion about instruments to increase the diversity in the export basket.
The thesis is structured as follows: Chapter 2 provides a literature review of the relationship between export diversification and economic performance. It describes the current research with reference to the determinants of export diversification and explores the channels through which export diversification enhances growth. Furthermore, the chapter presents evidence for the positive effect of exports on economic growth and its influence on various additional economic factors.
Chapter 3 examines the database used in the empirical analysis. It describes the quality issues that exist in reference to the export data of SSA countries.
Chapter 4 explores the methodology used to measure export diversification and the characteristics of this analysis.
Chapter 5 analyzes the level of export diversification in the resource-rich countries, the resource-scarce landlocked countries and the resource-scarce coastal countries. A second part of the analysis covers the investigation of resource-rich countries that have a significant stock of oil and resource-rich countries that have no significant stock of oil.
Chapter 6 summarizes the results and elaborates their consequences. Finally, it discusses whether the findings are useful to provide policy recommendations for SSA.
2. Literature review
In the past, a large amount of research has dealt with export diversification. The literature on the different facets of export diversification can be divided into two groups. The theoretical literature explains the theory that supports the call for export diversification in SSA. The empirical literature describes earlier research that has dealt with the influence of export diversification on economic growth and the determinants of export diversification.
First it is important to stress the economic importance of exports in general and the influence of exports on growth. In the past, several papers have found a positive influence of export growth on economic growth.
A paper written by John Mbaku in 1989 investigated the export and growth data of 37 African countries over a period from 1970 to 1981. Based on this data, the author found a positive influence of exports on economic growth. This effect was even stronger for middle-income countries than for low-income countries. A paper by Onafowora and Owoye (1998) researched the influence of export growth on economic growth in SSA and confirmed earlier results for 10 out of 12 countries over a period from 1963 to 1993. The economist Victor Ukpolo showed that non-fuel primary exports contributed to economic growth for low-income African countries from 1969 to 1988 (Ukpolo, 1994). Dodaro (1991) used a regression analysis of 41 developing countries over the period from 1973 to 1985 to find a positive influence of manufacturing exports on GDP growth.
These results show that exports are indeed an important factor for economic development and growth. Hence meaningful policy recommendations require a thorough understanding of all processes related to exports.
Chapter 2.1 gives an overview of the theoretical background of export diversification. The empirical literature supporting the theoretical part is presented in chapter 2.2. Within chapter 2.2, section 2.2.4 describes the current state of research regarding the determinants of export diversification.
2.1 Theoretical literature on export diversification
This section of the thesis describes the theories which explain the diversification process and helps understand the positive effects of export diversification on the economy. The presented paper by Agosin (2009) does not cover all facets of export diversification and its influence but it gives an overview over the main effects. Additional influence, like terms of trade volatility, will be covered in the empirical part. In his paper “Export diversification and growth in emerging economies”, Agosin (2009) gives a detailed overview of the argumentation behind the influence of export diversification on economic growth. He presents two effects of export diversification that support growth, the Portfolio Effect and the Dynamic Effects. The Portfolio Effect (figure 1) supports growth by decreasing the volatility of export earnings.
Figure 1 – Portfolio Effect
The idea behind the Portfolio Effect is that a low level of export diversification (i.e. high export concentration) leads to highly volatile export earnings, which leads to a higher variance of GDP growth. Many developing countries are not able to smooth consumption because they have only limited access to the capital markets. In times of worldwide contraction, the consequences are a reduction of capacities and deskilling of the national labour force. The following decrease of human capital in the end affects the economic development negatively. Furthermore, countries with an export basket that is highly focused on one or only few products tend to have a more volatile real exchange rate than countries with a broader export basket. Real exchange rate volatility generally discourages investments in the tradable goods sector.
The Dynamic Effects of export diversification support growth through learning and knowledge spillovers, as shown in figure 2.
Figure 2 – Dynamic Effects
Source: Figure 2 constructed by author, based on Agosin (2009).
growth because a comparative advantage in one sector is likely to exist in a related sector. The author used an ordinary least squares estimation to estimate the effects of export diversification on income. He found that countries with an export profile close to high-income countries converge faster towards high high-income levels than countries with an export portfolio that is close to low-income countries. In addition, Agosin’s research supported the hypothesis that exports diversification is likely to start the growth process.
In summary it can be stated that Agosin (2009) described two major effects that explain the theoretical background of the positive influence of export diversification on economic growth. The Portfolio Effect decreases volatility of export earnings, which leads to an increase in economic growth due to economic stability. The Dynamic Effects support growth due to knowledge spillovers and accumulation of human capital.
2.2 Empirical literature on export diversification
This section of the thesis reviews the empirical literature that supports the theoretical considerations presented in section 2.1. It focuses on explaining the relationship between export diversification and productivity, the influence of export concentration on domestic income, on the terms of trade, and on investments. Furthermore, it also provides an overview of the relationship between export diversification and economic growth and the determinants of export diversification.
2.2.1 Relationship between export diversification and productivity
The positive influence of exports on productivity levels receives a lot of attention as explanation for the positive relationship between exports and growth from the scientific community. The idea behind this relationship is that an increase in the exporting sector leads to an increased level of competition, which leads to an increase in productivity. In addition, many exporting sectors are a potential source of spillover effects and learning effects for other sectors of the economy. Thus an extension of the export base towards a more diverse export basket proves to be a useful example for the application of this relationship. Bigsten et al. (2004), using firm data from Cameroon, Ghana, Kenya and
Zimbabwe over the period from 1991/92 to 1993/94, found a positive influence of exports on productivity. Their result was confirmed in 2005 by Johannes Van Biesebroeck, who used firm data from Burundi, Cameroon, Cote d’Ivoire, Ethiopia, Ghana, Kenya, Tanzania, Zambia, and Zimbabwe over the period from 1992 to 1996. The economist De Loecker (2007) researched the Slovenian manufacturing sector over the period from 1994 to 2000. He found that the productivity of firms increased after the firms had started to export. Furthermore, the difference in productivity between exporting companies and non-exporting companies has increased over time. Another key finding in this paper is the influence of the trading partner on the productivity development. Trade with high-income countries led to a higher productivity increase than trade with low-income countries, which was possibly caused by the stronger learning effects stemming from the contact with higher developed trading partners. This is closely related to the topic of export diversification in reference to a diversification of export destinations. A diversified set of export destinations can increase the productivity in SSA and is therefore relevant for policy makers. Blalock and Gertler (2004) confirmed these results as they found an increase in productivity for the Indonesian manufacturing sector over the period from 1990 to 1996, a fact they attributed to the learning effects.
All of the mentioned papers agree that exports increase the productivity level due to learning effects, which is important since it represents an opportunity to improve economic growth with the aid of export diversification. The empirical findings also support the Dynamic Effects theory developed by Agosin (2009). Additionally, they also show that export to higher developed countries has a stronger positive effect than trade with lower developed countries due to better learning opportunities.
2.2.2 Influence of export concentration on income, terms of trade and investments
This section presents various channels through which export concentration (i.e. the opposite of export diversification) influences the volatility of terms of trade, the volatility of income and the rate of investments, all of which influence economic growth. Four papers will be elaborated in this context. The first paper by Busch (2011) investigates the
influence of export concentration on the volatility of terms of trade. Bacchetta, Jansen, Piermartini and Amurgo-Pacheco (2007) researched the impact of this volatility of terms of trade on domestic income and were able to prove that it influences growth. A third paper by Dawe (1996) explores the influence of export instability caused by export concentration on the rate of investments. The fourth paper, Aizenmann and Marion (1999) described the impact of exchange rate volatility on private investments, which is relevant since exchange rate volatility is the result of a concentrated export basket. These papers and their findings are elaborated below.
The economist Christian Busch (2011) explored the causal relationship between export concentration and volatility of export growth rates and terms of trade. The author used a two-stage least squares regression to identify the causal relationship, while using a generalized entropy measure, similar to the measure used in this thesis, to calculate the level of concentration. The dependent variables in this paper are the terms of trade, the export growth and the exchange rate. The sample period covers export data from 1980 to 2000. The terms of trade are the ratio of export prices to import prices. A highly concentrated export basket leads to a situation where the terms of trade are very vulnerable to individual price changes since they have a much higher impact on the overall price level then in the case of a diversified export basket. To avoid possible endogeneity issues the author used geographic characteristics as instruments for measuring export concentration. The analysis confirmed the hypothesis that export concentration has a positive influence on volatility in export growth rates and in terms of trade which in the end influences economic growth negatively, as the paper by Bacchetta, Jansen, Piermartini and Amurgo-Pacheco (2007) demonstrates. Furthermore the characteristic of being landlocked appears to increase the level of export concentration. This finding underlines the need for different models for different country groups in SSA and motivates the research question of this thesis.
Bacchetta, Jansen, Piermartini and Amurgo-Pacheco (2007) investigated the influence of the trading partners on the domestic income. They were able to show that volatility in the terms of trade affects the volatility of income in developing countries. Busch (2011) proved that a concentrated export basket affects the volatility of the terms of trade positively. His
paper is especially interesting since it points out the influence of this volatility on terms of trade. Instead of using the Herfindahl-Hirschman Index, Busch made use of a measure that takes the level of volatility of trading partners. He used panel data to run a regression analysis designed to verify the effect of export diversification of the trading partners on the economic performance. The outcome is that the export diversification of trading partners and markets reduces income fluctuations and therefore has a positive influence on the economic performance. Furthermore, Busch demonstrated that import demand shocks are an important determinant of income volatility. Figure 3 shows that the group of least developed countries (LDC) exhibits the highest level of output volatility and hence is also most likely to be influenced by this volatility. Therefore the topic of export diversification is of special interest for the SSA countries.
Figure 3 – Output volatility by country group
Dawe (1996) researched the direct effect of export instability on investments and growth He used a regression analysis based on the sample period from 1965 to 1985 with the rate of investment and the growth rate as dependent variables. Dawe found that export instability leads to lower rate of investments and hence hampers economic growth. An interesting result is that an unstable income in one sector influences the demand in other sectors. This process affects the prices in the other sectors, and the result is a nontransparent situation where the real price is hard to extract. This missing transparency influences the rate of investments negatively. Export instability thus can be regarded as a negative influence on economic growth, a finding that supports the Portfolio Effect theory by Agosin (2009). Another important channel through which a concentrated export basket can influence private investments in a country is the volatility of real exchange rates. Given a concentrated export basket, price changes may lead to either a strong increase or decrease in demand for a certain currency. This could influence the exchange rate and lead to an increase in the volatility of real exchange rates, a relation that is also supported by Agosin (2009). Aizenmann and Marion (1999) investigated the relation between the volatility of government consumption, the volatility of nominal money growth, the volatility of real exchange rates and private investment. They found a negative correlation between volatility of real exchange rates and private investments. This outcome, which is based on a panel of 43 developing countries over the period from 1970 to 1992, shows that high volatility of real exchange rates could act as a disincentive to investment. Even more important is the fact that existing capital-market imperfections may increase this effect in developing countries.
All in all we can conclude that various variables will be affected by a concentrated export basket and a limited number of trading partners. Terms of trade volatility influences income negatively and hence the diversification of trading partners may improve the economic performance. Dawe (1996) showed that export instability, which is the result of a concentrated export basket, leads to a lower rate of investments which hampers growth. Finally the paper by Aizenmann and Marion (1999) demonstrated the negative relation between exchange rate volatility, which is the result of a concentrated export basket, and private investments.
2.2.3 Relationship between export diversification and economic growth
This section describes a selection of papers that provides an overview of the current state of research on the relationship between export diversification and growth.
Naudé and Rossouw (2008) researched the extent of export diversification in South Africa for the period from 1962 to 2000 and its relationship to GDP per capita. The authors used a computable general equilibrium (CGE) model to find the impact of greater export diversification versus greater export concentration in the South African export basket. The CGE approach showed that export diversification led to a higher GDP per capita and employment. A Granger causality test showed that greater export diversity Granger-causes an increase in GDP per capita. A crucial finding is that up to a certain level, an increase in income is accompanied by an increase in diversification. When said level is reached, the degree of diversification decreases again. Lederman and Klinger (2006) confirmed this U-shaped relationship by using a sample of 73 countries (of all levels of development) over the period from 1994 to 2002. Due to the fact that the authors used the normalized Herfindahl-Hirschman Index, which attaches a value of one to maximally concentrated export baskets and a value of zero to maximally diversified export baskets, the relationship is not an inverse U-shape relationship but a regular U-shape relationship. A high level of diversification (a value close to zero) is up to a certain degree accompanied by a high level of income.
Carrère, Strauss-Kahn and Cadot (2010) as well confirmed that poor countries on average exhibited a higher concentration in the export basket, while an increase in income usually meant a more diversified export basket. They used a sample of 62 middle income countries and 32 low income countries and covered the period from 1988 to 2006. It remains unclear whether there was a causal relation for the outcome. Contrary to earlier studies, they were not able to identify a U-shaped relationship. They also found a positive correlation between liberalization and diversification. This may be interpreted as evidence that a framework that supports trade may also support export diversification. A critical remark concerning their study is that the authors failed to research a causal relationship between income and diversification.
Another paper by Bertinelli, Salins and Strobl (2006) evaluated the export baskets of SSA countries and other developing countries using the UN Comtrade database. Covering the period from 1962 to 2002, the authors investigated the trade-off between the expected earnings and the level of diversification in the export basket on the basis of modern portfolio theory. The existence of welfare gains stemming from the move towards a more optimal portfolio, in this case a more diversified export basket, is remarkable and important for policy actions. The extent of these gains depends on the degree of risk aversion, i.e. on how the country weights the loss due to fluctuating export returns relative to efficiency losses due to diversification. Given the Epstein and Zin (1989) utility function, it is possible to compare the welfare gains of different portfolios: All countries would exhibit a significant welfare gain if they moved towards a more diversified export basket.
Amin Gutiérrez de Piñeres and Ferrantino (1997) researched the relationship between export diversification, GDP growth and various macroeconomic variables in Chile from 1962 to 1991. Two measures were applied to identify the level of export diversification, namely the change in the export composition in one year and the Herfindahl-Hirschman Index. The study found a positive long-run relationship between the economic growth and the export diversification in Chile. The negative short-run relationship between structural changes in the export composition and the GDP growth indicates the existence of short-run adjustment-costs. This was confirmed by a follow-up paper by Akbar and Naqvi (2000) which researched the relationship between export diversification and growth of exports in Pakistan from 1979 to 1998. The authors used a similar methodology to the paper by Amin Gutiérrez de Piñeres and Ferrantino. A regression analysis showed that a diversified export basket leads to growth in exports and GDP. The results confirm as the other papers did for Chile, that export diversification enhanced the economic growth in Pakistan.
An interesting paper by Herzer and Nowak-Lehmann Danzinger (2006) describes the long-run relationship between export diversification and growth. The research is grounded on time series data from 1962 to 2000 from Chile. The results confirm the hypothesis that export diversification is related to economic growth by learning externalities. A remarkable finding is that an orientation of additional sectors towards
exporting has a larger influence on growth than an increase of the industrial export share in total exports. This development can be explained with the expansion of resource-based industries in Chile that exhibit low or medium levels of technology, which have only limited potential for spillover effects, in contrast to diversification and intensification of industrial exports in high technology industries, which potentially generates stronger growth effects.
The research of Al-Marhubi (2000) is based on a regression analysis using a cross-country sample of 91 countries for the period from 1961 to 1988. The empirical analysis consists of two parts designed to capture the direct and the indirect effect of export diversification: The growth equation and the investment equation. The growth equation captures the direct effect of diversification, while the investment equation captures the indirect effect via investments. The analysis shows that greater export diversification is associated with faster economic growth. The author did not find a significant effect of export diversification on investments for the whole sample. For the sample of just the developing countries, the author found that export diversification influences growth directly and indirectly via investments.
A study by Saviotti and Frenken (2008) covered the effect of export diversification on developed countries over the period from 1964 to 2003. The investigation of 20 OECD countries found that diversification has a positive effect on economic growth. A regression analysis based on the entropy measure is used to identify this relationship. In addition, the countries that experienced a fast development also experienced a strong rise in diversification, which again confirmed earlier research. To achieve detailed results, the authors distinguished between diversification within sectors and diversification between sectors. Diversification within sectors describes new products that are related to existing products in the same sector. In contrast, diversification between sectors describes new products that are related to other sectors. The result is that diversification within sectors enhances growth immediately, while diversification between sectors enhances growth with a delay. The delay can be explained with costs that emerge before the profits of diversification can be realized. This is in line with the research done by Amin Gutiérrez de Piñeres and Ferrantino (1997).
Lederman and Maloney (2007) researched the growth data of 66 countries over the period from 1980 to 1999. The authors used the Herfindahl Index and the share of natural resources in exports in order to measure export concentration. They found that resource abundance supports growth, while a concentrated export basket hampers growth. This finding is particularly interesting since it shows it is not the resource stock but the export concentration that is related to this resource stock that hampers growth. In other words, the negative influence of export concentration is wrongly attributed to the existence of a significant resource stock. It is to clarify if this relationship holds for the SSA countries or if resource endowments drive export concentration.
In sum, the results of these papers show a positive relationship between export diversification and economic growth. This not only supports the theoretical groundwork of Agosin (2009), but also represents the motivation for this research.
2.2.4 Determinants of export diversification
The following part describes the current state of research regarding the determinants of export diversification. With the research question in mind, this section provides additional insight into the export diversification determinants and supports the research setup.
The economists Bebczuck and Berrettoni (2006) found a positive relationship between the share of fuel exports in total exports and the concentration of exports. The authors used the outcome of the Herfindahl Index of 56 countries to find this relationship. A possible explanation is the existence of the Dutch disease, which is an unhealthy appreciation of the local currency. In addition, high returns due to the export of resources may lead to an inefficient allocation of capital and workers. This can influence the long-term growth prospects negatively, while it may decrease the incentive to diversify the export basket. As additional determinants, the export to GDP ratio, the per capita GDP and other macroeconomic variables are cited. The study is especially interesting since the results regarding the effect of resource abundance on export diversification are contrary to the results of Lederman and Maloney (2007). Furthermore, it underlines the importance of this research to identify the influence of a significant resource stock on export diversification.
Another paper by Kamuganga (2012) found a negative relationship between the infrastructure related trade frictions (export cost, time to export, procedures to export) and the level of export diversification. The author used the export statistics of African countries for the period from 1995 to 2009. In addition, the author found other variables that influence export diversification, namely intra-Africa regional cooperation, product and market experience as well as macroeconomic developments (exchange rate volatility, financial underdevelopment, inappropriate foreign direct investments).
An extensive study by Agosin, Alvarez and Bravo-Ortega (2011) investigated various determinants of export diversification using a data set of 161 countries over the period from 1962 to 2000. The study also measured concentration with the Herfindahl-Hirschman Index and explored the determinants with a regression analysis. To receive unbiased results, the authors used a Generalized Method of Moments, which uses the lags of the explanatory variables as instruments. A remarkable finding is that trade openness supports export concentration instead of export diversification. The level of financial development does not play a significant role for export diversification, and neither do the real exchange rate volatility and overvaluation of the currency. Human capital accumulation (measured in years of schooling) supports the diversification process, while the economic distance influences the export diversification negatively by raising the trade costs.
Papageorgiou and Spatafora (2012) investigated the development of Tanzania, Bangladesh, Vietnam, and Malaysia. For the analysis of the SSA countries, the development of Tanzania is of special interest because it experienced several market-oriented reforms. Those reforms supported growth in mining, manufacturing, construction, and services, while at the same time the share of production and exports of traditional agricultural products saw a decrease. Data shows that Tanzania has experienced a diversification of exports since the 1990s. The main line of this case study is that diversification can be the result of reforms or policy measures and hence can be influenced. Reforms may target macroeconomic (e.g. exchange rates) or microeconomic factors (e.g. infrastructure or investment climate) in order to increase the diversification of export baskets.
Keeping these findings in mind, this thesis tries to determine the quantitative influence in terms of different diversification measure of the trade frictions (distance to the sea) and the share of resource exports in total exports.
This chapter presented the theoretical groundwork necessary for further research on the positive influence of export diversification on economic growth as described by Agosin (2009). The positive effects are attributed to less export volatility and hence more stability and learning effects as well as spillovers. Closely related to the topic of export diversification is the relationship between exports and productivity, especially since the increase of export sectors is a way to increase the productivity and hence economic growth. The economists Bigsten et al. (2004) and De Loecker (2007) attribute this influence to learning effects and spillovers, which also supports Agosin’s (2009) theoretical groundwork. Furthermore the chapter presented the influence of export concentration on income, terms of trade and investments. It described the negative influence of terms of trade volatility, which is the result of a concentrated export basket, on income. This leads to the conclusion that the diversification of the export basket can increase the domestic income, which is crucial for the motivation of this research. Additional research by Dawe (1996) as well as Aizenmann and Marion (1999) showed that export instability and exchange rate volatility resulting from a concentrated export basket, lead to a lower rate of investments which hampers growth. The fourth part of the chapter took a look at the consequences of these relationships and investigated the relationship between export diversification and growth. The result here is that export diversification is connected to an increase in economic growth and income. Finally, various determinants that influence the level of export diversification were subject to investigation. As a result, we can record the fact that infrastructure-related trade frictions and the share of fuel exports in total exports support the decision to pool the countries in three country groups.
3. Data description
This chapter describes the database used for the analysis of the level of export diversification in SSA. Furthermore, it describes the issues that exist in reference to the low quality of the export data of the SSA countries. Finally, the chapter presents a set of countries that cannot be considered for this study due to data limitation. My research uses the UN Comtrade database. It is based on the Standard International Trade Classification (SITC) which attaches a number to every export product. This procedure allows a detailed analysis of the export basket. Although the SITC has been revised multiple times over the last decades, this analysis applies the Revision 1 for two reasons. It is consistently defined for a long time period and provides the basis for a large number of economic studies and hence allows a comparison with other papers. The Revision 1 provides five different levels of aggregation ranging from ten product categories (one-digit code) to 1312 product categories (five-digit code). The availability of very detailed and disaggregated data helps obtain a detailed picture of the diversification process of SSA. This analysis makes use of the four-digit code. Its level of aggregation covers 625 products and provides an extensive analysis of all diversification efforts. Beyond that it also makes sure that the trade-off between completeness and the needed accuracy is maintained. It is not useful to apply the five-digit code due to the poor IT infrastructure of the UN Comtrade database that doesn't allow an extraction of the five-digit data. The four-digit data are a good compromise and create meaningful results. The aggregation is relevant to cover horizontal diversification which appears to be an important part of the diversification process. The economists Saviotti and Frenken (2008) showed that diversification within the same sector immediately supports economic performance. Herzer and Nowak-Lehnmann (2006) found that horizontal diversification supports economic growth through knowledge accumulation. In addition, these findings support the selection of diversification measures in Chapter 4.
The research is confronted with severe data quality issues due to the fact that many SSA countries experienced conflicts or capacity problems that impeded the reporting of correct numbers. Therefore the use of mirror data is required in order to achieve robust results. That is, instead of relying solely on low-quality export data of the SSA countries, this
study uses high-quality import data from the importing countries. These mirror data, mainly provided by the industrial countries, are more accurate than the export data provided by the SSA countries. The countries used here are the USA, Australia, Brazil, Canada, China, India, Indonesia, Malaysia, Thailand, Taiwan, Austria, Belgium, Finland, France, Germany, Spain, Greece, Ireland, Italy, The Netherlands, Portugal, Sweden, UK, Hong Kong, Japan, Republic of Korea, Singapore, Luxembourg, Norway, Switzerland and Denmark.
The trade between the SSA countries also plays an important role. Therefore the data set covers also the import data of SSA countries. The sample covers a period of 18 years (1993 to 2010). This sample provides a detailed insight into the export diversification process in SSA, allowing the analysis of structural changes in the export basket. Since data are missing for the years 2011 to 2013 this research is based on data dating up to the year 2010. In this way, accuracy can be preserved while obtaining a meaningful sample period. The year 1993 was chosen as a start year for two reasons. First, since 1990 the volume of world trade and the trade between developing countries have increased dramatically (WTO, 2011 & WTO, 2013). The building of new production capacities takes time and therefore the diversification effect of this increase should be visible with a delay of a couple of years. Second, the database required to move the sample period to 1993 in order to avoid further dropping of countries due to data incompleteness.
The result is a database consisting of 41 SSA countries at a four-digit level in SITC, Revision 1. The database exhibits the exports of SSA in nominal US dollars and includes the period from 1993 to 2010.
4. Methodology
This chapter explores the country groups and the four measures applied to determine the level of diversification. It then describes the composite index that combines the export share based indices and the normalization process necessary to compare their results.
4.1 Determining the country groups
The countries in this study are classified according to their resources and geographic location, and form three distinct combinations: resource-rich countries, resource-poor landlocked countries and resource-poor coastal countries. In addition, the group of resource-rich countries consists ofoil-exporting countries and non-oil-exporting countries.
4.1.1Resource-rich countries
The thesis investigates export diversification on the basis of the sample countries that are subdivided into three country groups: resource-rich countries, resource-poor coastal countries and resource-poor landlocked countries: It is necessary to define the country groups precisely. This thesis adopts the definition of the IMF to classify countries as resource-rich or resource-poor. The average percentage of either natural resource revenue or exports in resource-rich countries comes to at least 20% of total fiscal revenue or exports for the years 2006 – 2010. Based on this criterion, the resource-rich countries are Angola, Cameroon, Central African Republic, Chad, Rep. of Congo, Dem. Rep. Congo, Côte d'Ivoire, Equatorial Guinea, Gabon, Ghana, Guinea, Liberia, Madagascar, Mali, Mauritania, Mozambique, Niger, Nigeria, Sierra Leone, Sudan, Tanzania, Togo, Uganda, Zambia (IMF, 2012). Next to the convenient effect of using an already classified set of countries, this classification has the advantage that it captures both exports and revenue, which is of advantage since it provides a detailed picture of the resource endowment.
4.1.2Oil-exporting countries and non-oil-exporting countries
It is necessary to assess additional factors that may influence the outcome of the empirical analysis. Therefore I distinguish between oil-exporting countries and non-oil-exporting resource-rich countries. Developed countries and SSA countries enter strategic partnerships that help diversify the export basket on the basis of the importance of oil for developed countries. Alternatively, a high income from oil export may act as a disincentive for diversification. This could influence the level of export diversification in oil-exporting countries significantly and therefore bias the analysis. Hence I investigate if oil-rich countries have diversified their export baskets stronger or weaker than other resource-rich countries. The 13oil-exporting countries are Angola, Cameroon, Chad, Rep. of Congo, Dem. Rep. Congo, Côte d'Ivoire, Equatorial Guinea, Gabon, Ghana, Madagascar, Nigeria, Sudan and Uganda. Again the thesis applies the classification of the IMF (2012), which identifies a country as oil-rich if oil accounts for at least 20 percent of export or fiscal revenue using average data for 2006 – 2010. Corresponding the 11 non-oil exporting countries are Central African Republic, Guinea, Liberia, Mali, Mauritania, Mozambique, Niger, Sierra Leone, Tanzania, Togo and Zambia.
4.1.3Resource-poor landlocked countries
Resource-poor landlocked countries have no direct access to the sea and no significant stock of resources. In contrast to earlier research, this thesis does not consider big rivers as economically appropriate alternatives to a sea access. The decision whether or not the size of a river qualified it for trade that is comparable to sea trade is too extensive for this study. According to this requirement, relevant resource-poor landlocked countries are Burkina Faso, Burundi, Ethiopia, Rwanda and Zimbabwe.
4.1.4Resource-poor coastal countries
All countries that have access to the sea and that have no significant stock of resources belong to the group of resource-poor coastal countries. Their group consists of Benin, Cape
Verde, Comoros, Djibouti, Eritrea, The Gambia, Guinea-Bissau, Kenya, Malawi, Mauritius, Senegal and Seychelles.
Due to the fact that Botswana, Lesotho, Namibia, South Africa and Swaziland have formed the Southern African Customs Union, the Comtrade database does not provide individual export data until 2000. In addition, the data for Reunion, Somalia, Western Sahara and Sao Tome and Principe is not fully available. Thus the present study does not consider those countries in favor of robust results.
4.2 Measurement of diversification
Next to the quality of data, the measurement of export diversification is a crucial part of the empirical analysis. The following part explores methods of measuring export diversification and examines the characteristics of different measures. The measures applied to determine the level of export diversification were initially developed to measure the concentration of firms. Export concentration is the opposite of export diversification.
The different concentration measures mainly differ in two characteristics that are responsible for their outcomes. The first characteristic is the weighting scheme of the concentration measure that determines how the measures react on changes in the smallest and largest export sectors. The economist Marfels (1971) illustrated the four different weighting schemes. The first weighting scheme is based on an arbitrary choice that determines if a sector receives a weight of one or zero in the calculation. One example for this weighting scheme is the Concentration Ratio. The second weighting scheme attaches a weight to every export product amounting to the share in total exports. Greater shares lead to a greater weight in the Index. The Herfindahl-Hirschman Index belongs to this category. In contrast to the second weighting scheme, the third weighting scheme attaches a weight to every export share correspondent to the ranking of this export product. The smallest export share receives the highest ranking, and the largest export share receives
the lowest ranking. The third weighting scheme attaches a value of one to the largest export share and a value of n to the smallest export share, with n being the number of export sectors. The Hall-Tideman Index exhibits this weighting scheme. The second and the third weighting schemes are indirect reverses, which influence the magnitude of the respective measure but not the general direction of the outcome. In other words, an increase in export lines will lead to an increase in all diversification measures, while the magnitude of the increase differs. The fourth weighting scheme attaches a weight to every export product, amounting to the negative value of the logarithm of the share in the export basket. One example for the fourth weighting scheme is the Entropy Measure.
The second characteristic is illustrated in a paper by Bikker and Haaf (2002): the structure of the measure, which is either discrete or cumulative. The discrete measures describe the distribution over an arbitrary set of export products. In contrast, the cumulative measures describe the distribution over the complete export basket. This feature is important with regard to the influence of small shares on the outcome. The motivation behind this analysis is to find out about the positive effects of export diversification on economic growth. If changes in the smallest export sector do not affect economic growth at all, e.g. companies in the smallest sector may be too small to create spillovers or to offset export fluctuations, a discrete measure like the Concentration Ratio is a good instrument to measure the relevant export diversification. If changes in the smallest export sector play a role, cumulative measures like the Herfindahl-Hirschman Index, the Hall-Tideman Index and the Entropy Measure are good instruments.
At this point it is important to stress why this analysis uses different measures. The motivation for this research are the positive effects of export diversification on economic growth and economic stability. The question arises whether diversification in the largest sector is more important for the positive influence on economic growth than diversification in the smallest sector. If the diversification in the smallest sector has a stronger impact on economic growth and a country shows a strong diversification in the smallest export sector, while the measure weights changes in the smallest sectors relatively small, this could lead to a wrong picture of the level of relevant export diversification compared to other countries. Since it is not the objective of this thesis to determine which
part of the export diversification is the most relevant part for the positive influence, a number of export diversification measures will be used in order to obtain an extensive picture of the level of export diversification without assumptions about the relevant part of the diversification process. The following section illustrates the Herfindahl-Hirschman Index, the Hall-Tideman Index, the Entropy Measure and the Index based on the number of active export lines. Four different ways of measuring concentration were chosen in order to get a detailed picture of the export basket composition in SSA. In addition, all three export share based indices are used to create the Composite Index because it is unclear which part of the diversification process is most relevant for the positive effects (e.g. spillovers and offsetting of fluctuations). Accordingly the result is independent of the assumption which part of export diversification is relevant for the positive effects and hence relevant for the overall research. All measures describe the export diversification, but the magnitudes of the measures differ. The birth of a new export sector has a stronger impact on the Hall-Tideman Index than on the Herfindahl-Hirschman Index, while both indices will show an increase in diversification.
The existing research done by Saviotti and Frenken (2008) as well as Herzer and Nowak-Lehnmann (2006) supports the argument that diversification in every export sector influences the economic development. Therefore the selection of measures in my thesis covers only the cumulative measures that capture export diversification in every sector. It is based on three cumulative measures that exhibit three different weighting schemes.
4.3 Herfindahl-Hirschman Index (HHI)
The first measure used to determine export diversification is the Herfindahl-Hirschman Index (HHI) constructed by Orris Clemens Herfindahl and Albert Otto Hirschman in 1964 (Hirschman, 1964). The index is defined as the sum of the squared export shares in percent. The HHI can be expressed as:
where is the number of export categories and is the export share of the individual export category in percentage. The HHI ranges from 1⁄ to one, with a HHI of 1⁄ indicating a maximally diversified export basket and a HHI of one indicating a maximally concentrated export basket. The index belongs to the second weighting scheme that attaches a weight to every export product amounting to the share in total exports. This weighting scheme makes the index less sensitive to changes in the number of export categories when the number of categories is large (Davies 1979). The Herfindahl-Hirschman Index is used to weight concentration in the largest sector stronger than in the smallest sector. The application of this scheme allows for the argumentation that the positive influence of export diversification is mainly driven by sectors that have already a certain size that enables spillover effects. This can be illustrated with the example of a company with three employees that starts as the first firm to export regional music instruments. The company may produce less spillover effects than a middle class car supplier with 5000 employees that starts as the first firm to export polyurethane (plastic) products. In order to capture this argumentation, the HHI has been selected as one measure to determine the level of export diversification in this thesis.
4.4 Hall-Tideman Index (HTI)
The second indicator that measures export diversification is the Hall-Tideman Index (HTI). The index was developed by Marshall Hall and Nicolaus Tideman in 1967 (Hall and Tideman, 1967). The index can be expressed as:
= 1 2 − 1 ; = 1,2, . . . ,
where is the number of export categories, is the export share of the individual export category in percentage and is the ranking of this share. The index ranges from zero to one. A HTI of zero indicates a maximally diversified export basket and a HTI of one indicates a maximally concentrated export basket. The HTI belongs to the third weighting
scheme that attaches a weight correspondent to the ranking of the category. The smallest export share receives the highest ranking and the largest export share receives the lowest ranking. Using this weighting scheme, the HTI takes the starting condition of the country into account. Diversification is seen as relatively easy if the export basket exhibits many sectors. More sectors increase the chances for horizontal diversification. In contrast, diversification is seen as relatively difficult if the export basket exhibits a small number of sectors (Bikker and Haaf, 2002). Especially developing countries with small existing export baskets require a measure that captures this characteristic. In contrast to an industrial country that can use spillover effects to enlarge the export base, many SSA countries have only limited access to related knowledge and capacities. Therefore the effort needed to increase the export base is higher for developing countries with a small number of export products. To capture this characteristic, the HTI was chosen.
4.5 Entropy Measure (ENT)
The third index to measure export diversification is the Entropy Measure (ENT) constructed by Claude Elwood Shannon in 1948 (Shannon, 1948). The index can be expressed as:
= − ; = 1,2, . . . ,
where is the number of export categories and is the export share of the individual export category in percentage. The ENT ranges from zero to where an ENT of indicates a maximally diversified export basket and an ENT of zero a maximally concentrated export basket. The index belongs to the fourth weighting scheme that attaches a weight to every export category amounting to the negative value of the logarithm of the export share. Because of the logarithm, the ENT weights small export sectors relatively strong and big export sectors relatively weak. Few big sectors lead to a number close to zero, which indicates a high concentration and vice versa. This measure
takes into account the initial economic situation and does not weight the biggest sectors as weak as the HTI.
Table 1 – Summary of diversification measure characteristics
Index type Range Typical features
HHI 1⁄ ≤ ≤ 1 Considers all export
products. Not sensitive to the introduction of new export products when the number of existing sectors is large.
HTI 0 < ≤ 1 Emphasis on the total
number of export products.
ENT 0 ≤ ≤ Based on logarithm.
Source: Table 1 constructed by author.
4.6 Composite Index (CI)
The next step in my analysis is to create a Composite Index (CI) that combines those three measures. The composite index is necessary to gain a comprehensive picture of the diversification process. In this thesis the CI is the average over the three measures and hence covers three different weighting schemes. Without the knowledge about the relevant part of the diversification process, the combination of those three measures proves to be a good approximation of the level of export diversification that is responsible for the positive influence on economic growth.
The HHI and the ENT do not range from zero to one. Therefore the calculation of the average requires the calculation of a Normalized Index Score (NIS) of the different measures. This thesis uses the formula:
= −−
where is the score of the index for country j and and are the left and right borders of the index (Meilak, 2008). The result is that all three measures range from zero to one where an index of zero indicates a maximally diversified export basket and an index of one indicates a maximally concentrated export basket. In the following the normalized HHI will be named NHHI and the normalized ENT will be named NENT.
4.7 Number of active export lines
The last measure that is used in this thesis is the number of active export lines. The advantage of this measure is its relative independence from price effects. A price increase in one export sector does not affect the number of active export lines immediately. In long-term perspective a price change in one export sector may lead to a shift from one export line to another due to an increased or decreased profitability. The result of this development is not clear since the shift may lead to a net increase, a net decrease or a stable number of export lines. A net increase may happen due to a shift to new peripheral industries, which is larger than the movement to existing industries. A decrease may happen due to a shift to already existing industries that is larger than the movement to new industries. A stable number of active export lines could be the result of a decrease of existing export lines and an increase of new export lines that offset each other.
The determination of the level of diversification of the country groups requires the calculation of the average over the included countries. To increase the quality of this determination, the thesis is based on two weighting schemes that construct the country group average: Population weighting and GDP weighting determine the weight of the individual country in the country group. This means that for the first weighting scheme, the population determines the weight of the country in the country group average. A country with a large population has a larger share in the group average than a country with a small population.
For the second weighting scheme, the GDP determines the weight of the country in the country group average. Hence a country with a large GDP has a larger share in the group average than a country with a small GDP. This helps capture the economic importance of a country in its specific country group. In contrast to the economic weight of the country in this region, the population weighting scheme ignores the economic performance of the countries in the country group. The result of this weighting scheme is less sensitive to changes in prices since price increases or decreases do not change the weight in the country group. Population statistics are taken from the Penn World Table 7.1 (https://pwt.sas.upenn.edu/). The GDP statistics are taken from the World Bank and are measured in current US Dollars (World Bank, 2013b).
Finally the empirical analysis also covers the variance of the diversification measures. The analysis provides insights into the volatility of export diversification over the sample period. A country group that shows a high level of variance in the export diversification may go through major structural changes.
The Herfindahl-Hirschman Index (HHI), the Hall-Tideman Index (HTI), the Entropy Measure (ENT), the Composite Index (CI) and the Index based on the number of active export lines differ in their weighting scheme which influences the levels of export diversification. The magnitude of the results is depending on the changes in the largest and the smallest sectors. On the previous pages, this chapter presented a selection of measures that concentrate on export diversification in the largest (HHI) and the smallest sectors (HTI and ENT).
5. Empirical analysis
The empirical analysis examines results of this research and tries to answer the research question: Does the extent of export diversification in Sub-Saharan Africa from 1993 to 2010 differ between the groups of resource-rich countries, resource-poor coastal countries and resource-poor landlocked countries? The analysis deals with the export diversification in those three groups and investigates the export diversification in oil-exporting countries compared to that in non-oil-exporting countries.
5.1 Diversification measures
The following figures show the development of the individual diversification measures for the three groups. The analysis takes a look at the individual indices, each of which has its own weighting scheme and hence shows different levels of export diversification for each country group. The determination of the level of diversification of the country groups requires the calculation of the average over the included countries. To increase the quality of this determination, the thesis is based on two weighting schemes that construct the country group average. The population weighting is the first approach and the GDP weighting is the second approach to determine the weight of the individual country in the country group. For the first weighting scheme the population determines the weight of the country in the country group average. For the second weighting scheme the GDP determines the weight of the country in the country group average. This helps capture the economic importance of a country in its specific country group. Contrary to the economic weight of the country in this region, the population weight ignores the economic performance of the countries in the country group. The result of this weighting scheme is less sensitive to changes in prices. Finally, the plain number of export products is analyzed in order to receive a relative price independent picture.
5.2 The outcome of the weighted average normalized Herfindahl-Hirschman Index
2010. The HHI weights changes in the biggest export sectors relatively strong compared to changes in the smallest export sectors. On the premise of a large number of exports, the index does not react strongly to changes of this number. Due to the normalization, this index ranges from zero (high diversification) to one (low diversification). Figure 4 shows the normalized HHI weighted by the population, while figure 5 shows the normalized HHI weighted by the GDP. Figure 6 shows the oil price in US Dollars per barrel.
Figure 4 - Outcome of the population weighted average NHHI
Source: Figure 4 constructed by author, based on UN Comtrade Data1.
1RRC stands for resource-rich countries. OEC stands for oil-exporting countries.
NOEC stands for non-Oil-exporting resource-rich countries. RPCC stands for resource-poor coastal countries.
RPLC stands for resource-poor landlocked countries. USA stands for United States of America.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 199319941995199619971998199920002001200220032004200520062007200820092010
Weighted (Pop) Average NHHI
RRC OEC NOEC RPCC RPLC USA
Figure 5 - Outcome of the GDP weighted average NHHI
Source: Figure 5 constructed by author, based on UN Comtrade Data.
Figure 4 shows a big gap between the resource-poor coastal countries (RPCC) and the resource-rich countries (RRC) on the one hand and the resource-poor landlocked countries (RPLC) on the other. In reference to these three country groups the RRC and the RPLC exhibit the highest level of concentration with 0.28 respectively 0.34 on average over 1993-2010, while the RPCC exhibit the lowest level of concentration with 0.16. The RPCC have shown a steady decrease in concentration since 1993 (from approximately 0.2 to 0.1). The split of RRC into oil-exporting countries (OEC) and non-oil-exporting resource-rich countries (NOEC) reveals that the oil-exporting countries are responsible for a substantial part of the export concentration. The OEC exhibit a NHHI of 0.52, while at the same time the NOEC show a NHHI of 0.27. The variance analysis of the population weighted NHHI in table 2 shows that the RPLC have the highest level of fluctuations in their index, which is in line with the argument that these countries are highly dependent on the surrounding countries. This argument is backed by the increase in concentration in the RPLC around 1998 that goes hand in hand with a decrease in concentration in the OEC and a drop in the oil price. A weak economy in the OEC due to a weak oil price could lead to decrease in
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 199319941995199619971998199920002001200220032004200520062007200820092010
Weighted (GDP) Average GDP NHHI
RRC OEC NOEC RPCC RPLC USA
demand for RPLC exports that could raise the lasting export shares in the RPLC even higher. At the same time an increasing oil-price has led to a decline in concentration after 1998. Figure 5 shows a much bigger gap between the RRC and the RPLC. The average NHHI of the RRC is 0.52, while it is 0.29 for the RPLC. The value of the RRC is much higher when weighted for GDP than for population. This explains the significantly higher level of concentration in figure 5 which has simultaneously risen together with the oil price since 1993, see figure 6. The USA exhibits the highest level of export diversification, which supports the theory that a high income is related to a high level of export diversification. This is also supported by the low variance in both weighting schemes for the USA.
Table 2 - Descriptive statistics for the Normalized HHI for various country groups
Average NHHI over 18 years
Weighted by Pop. Variance Weighted by GDP Variance RRC 0.282712329 0.000763962 0.52828243 0.008394179 OE 0.523555754 0.00152549 0.578937003 0.011313545 NOE 0.273515416 0.00136296 0.279476985 0.00288767 RPCC 0.167936308 0.001783954 0.142908422 0.00106365 RPLC 0.344436069 0.007366289 0.297422287 0.003296891 USA 0.01610279 4.82271E-06 0.01610279 4.82271E-06
Source: Table 2 constructed by author, based on UN Comtrade Data.
5.3 The outcome of the weighted average normalized Entropy Measure
After analyzing the NNHI, it is important to discuss additional measures to gain a comprehensive picture of the diversification process. The second measure used in this thesis is the normalized ENT which weights small export shares relatively strong, while weighting big export shares relatively weak. Because of the logarithm, changes in the smallest sector will not be weighted as strong as in the case of the HTI. Accordingly, changes in the biggest sector will not be weighted as weak as in the case of the HTI. Due to the normalization the NENT ranges from zero to one. A NENT of zero indicates a maximally diversified export basket and a NENT of one indicates a maximally concentrated export basket. Figure 7 shows the normalized ENT weighted by the population, while Figure 8 shows the normalized ENT weighted by the GDP.
Figure 6 – Oil Price development
Source: Figure 6 constructed by author, based on St. Louis FED Data.
Figure 7 - Outcome of the population weighted average NENT
Source: Figure 7 constructed by author, based on UN Comtrade Data.
0.000 20.000 40.000 60.000 80.000 100.000 120.000 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Oil Price in Dollars per barrel
0 0.02 0.04 0.06 0.08 0.1 0.12
Weighted (Pop) Average NENT
RRC OEC NOEC RPCC RPLC USA
