• Aucun résultat trouvé

Aggregate demand and structural transformation in Africa

N/A
N/A
Protected

Academic year: 2022

Partager "Aggregate demand and structural transformation in Africa"

Copied!
39
0
0

Texte intégral

(1)

Occasional Paper Series

Aggregate demand and structural transformation in Africa

August 2018

(2)

i

Acknowledgements

The authors of the present study are staff members in the Macroeconomic Policy Division of the Economic Commission for Africa: Economic Affairs Officer, Nadia S. Ouedraogo;

Economic Affairs Officer, Hopestone Kayiska Chavula; and Chief of Section, Economic Affairs, Khaled Hussein.They wish to express their sincere gratitude for the guidance and suggestions provided by the Director of the Macroeconomic Policy Division, Adam Elhiraika.

The authors also wish to acknowledge the support provided through comments and suggestions by colleague and Economic Affairs Officer, Allan Mukungu, whose input contributed to the improved quality of the study. The authors assume responsibility for errors or omissions. The views expressed herein do not necessarily represent those of the Economic Commission for Africa.

(3)

ii

Contents

Overview ... iii

I. Introduction ... 1

II. Africa’s growth and structural transformation ... 3

A. Economic growth patterns ... 3

B. Africa’s demographic dynamics ... 4

C. Rapid urbanization and the rise of the consuming class ... 5

III. Demand and structural transformation: theory and evidence... 6

IV. Methodology and data ... 10

A. Data and variable definitions ... 11

B. Key control variables ... 12

V. Model estimation and empirical results ... 14

VI. Conclusion and policy implications ... 22

References ... 28

Annex I. Variables, definitions and sources of data Annex II. Complementary tests

Annex III. Robustness test Annex IV. Regression results Table

Regression results for the structural transformation equation Tables in annexes

Annex II.

Table AII.1 Inclusion of time dummies?

Table AII.2 Testing for heteroscedasticity Table AII.3 Testing for serial correlation

Table AII.4 Hausman test, Random of fixed effects?

Annex III.

Table AIII. Robustness test Annex IV.

Table AIV. Regression results for the productivity equation Figures

I. Africa’s share of manufacturing in gross domestic product is (a) falling;

and (b) real production is increasing II. Productivity growth rate

(4)

iii

Overview

Notwithstanding the extensive work done on structural transformation, empirically identifying the key economic forces that shape structural transformation remains an open question. One issue is the role played by aggregate demand. Specifically, while it is known theoretically that households’ consumption, public expenditures, human capital, capital deepening and trade, among others, have an impact on resource allocation, their impact on structural transformation has not been empirically established, especially in Africa. Resolving this issue has important implications in terms of how policies and technological change affects structural transformation. In contrast to the structural transformation process emphasized in the conventional narrative, this study empirically assesses the impact of aggregate demand by identifying key economic forces that have had an impact on Africa’s structural transformation.

The study employs the Generalized Method of Moments IV (IV-GMM), using unbalanced panel data of 54 African countries over the period 1960–2014.

The results reveal that household and government consumption are found to have the largest impact on the manufacturing sector, compared with other demand-related variables. The approach has been to emphasize the need for economic policies that stimulate private consumption and investment among African countries, such as those that could increase availability of credit facilities to the private sector. Measures to enhance private sector investment, create jobs and raise people’s incomes should be encouraged.

(5)

I. Introduction

The term “structural changes” in the economy is most frequently used to explain the transformation in the composition of production, employment, demand and trade, which appears alongside the development of a country (Doyle, 1997). Both in developing and industrialized countries, constant changes occur in the composition of input and output, driven by technology and a desire to achieve greater competitive advantage. There is a strong relationship between the economic structure and the increase in productivity of a country.

Economists, mostly, recognize this relationship and clearly emphasize the necessity of structural changes to enhance economic growth. The principal structural changes emphasized in development literature involves increases in the rates of accumulation (Lewis, 1954; Rostow, 1960); shifts in the sectoral composition of economic activity (industrialization) focusing initially on the allocation of employment (Fisher, 1939; Clark, 1940); and later onproduction and factor use in general (Kuznets,1966; Chenery, 1960), changes in the location of economic activities (urbanization), and other aspects of changes in economic structure, including demographic transitionand income distributionamong others (Syrquin, 1988a).

The interrelated processes of structural changes, which follow or are followed by economic development, are referred to as “structural transformation”. The essence of structural transformation is the accumulation of physical and human capital, and shifts in the composition of demand, trade, production and employment (Chenery, 1986). Structural transformation is therefore considered to be an economic phenomenon, while the socioeconomic processes relating to this are identified as peripheral. Beside such narrow interpretation of structural transformation, there also exists broader interpretations that consider the structural changes that also occur in social institutions as necessary for modern economic growth (Kuznets, 1955;

North, 1981).

The interrelationship between structural changes and economic growth has drawn more attention during the past decade or so in which structural transformation has been considered a fundamental driver of economic development (Herrendorf, Rogerson and Valentinyi, 2012a;

Duarte and Restuccia, 2010). In particularthe movement of labour out of a less productive semi- subsistence agriculture into the more productive sectors of manufacturing and services.

Structural transformation is necessary in both urban and rural areas to sustain increases in overall productivity and living standards, and therefore drive poverty reduction. In other words, countries that pull themselves out of poverty exhibit positive structural change.1

Structural transformation features are also in the narrative on Africa’s recent growth.

Many African countries have developed strategies for achieving middle-income status in the coming decades. To realize this target requires profound changes in the production structures of African economies and as labour and other resources move from traditional to modern economic activities, overall productivity and income rises (Lewis, 1954; Chenery 1960;

Kuznets, 1966; Chenery, Robinson, and Syrquin, 1986; Chenery and Syrquin, 1975). There is, however, little evidence that the recent growth in Africa has been associated with the structural transformation of its economies, given that levels of industrialization remain low and poverty levels remain high (McMillan and Rodrik, 2011; Page, 2012; Beegle, and others, 2016).

1 The converse, however, is not true: all countries with structural change do not also achieve poverty reduction. Structural change into protected or subsidized sectors comes at the expense of other activities and is therefore not associated with sustained growth out of poverty, for the population as a whole.

Structural change can effectively reduce poverty only when people move from lower into higher productivity activities.

(6)

2 Notwithstanding the progress made during the past decade, current policies have not been effective enough in accelerating industrialization and job creation in productive sectors. New approaches are therefore necessary to accelerate structural transformation in the face of Africa’s unique demographic and spatial dynamics. These demand-side factors (involving demographic changes, urban and economic transitions, exports, and investment and consumption expenditures, among others) could lead to a growing domestic market that would accelerate structural transformation (African Development Bank, 2015).

The evolution of aggregate demand is an important driver of structural change. Structural change is shaped by differential growth rates of demand, which together with differential growth rates of output, are a cause of differential rates of productivity growth. Demand growth and productivity growth are thus mutually sustaining (Metcalfe, Foster and Ramlogan, 2006;

Saviotti and Pyka, 2012). In addition, the increase in productivity makes it possible for a rapid increase in unit wages, without reducing profit margins. If accompanied by a rise in employment, this trend can lead to a substantial increase in investment and consumption that will lead to a rapid increase in internal demand. After some years, an improvement in productivity leads also to an increase in external competitiveness and exports and in attractiveness of foreign direct investment (FDI), so that the internal demand-led growth gradually could become an export-led growth (Valli and Saccone, 2009).

Little attention, however, has been given to the effect of aggregate demand on structural transformation. Aggregate demand generally plays a small role in modern growth theory and in structural transformation. Structural change, as featured in the endogenous growth literature, is predominantly a supply-side phenomenon of the expansion of productivity and inputs, while the demand-side effects have somewhat been neglected in structural transformation analyses (Witt, 2001; Metcalfe, Foster and Ramlogan, 2006; Dietrich, 2009).

The present paper provides a holistic and empirical approach to how structural transformation specifically can be stimulated in Africa, given the significant changes in the forces influencing structural transformation and the multiple possibilities of how transformation might unfold. While theoretically, the relative empirical importance of these demand factors for structural transformation has not been established, although it has been acknowledged that household consumption, public expenditure, human capital, urbanization and trade have, in general, affected structural transformation. Resolving this issue has important implications for policy influence with regard to structural transformation.

The objective of the present paper is therefore to empirically assess the effects of aggregate demand and identify the key economic forces that shape Africa’s structural transformation. To do so, an econometric panel approach has been adopted to analyse the effect that the components of aggregate demand had on Africa’s transformation processes for the period 1960–2014. The analysis will combine the analysis of demand and some supply-side factors in explaining Africa’s structural transformation.

The paper is structured as follows: section I presents a general introduction; section II contains information on the growth and structural transformation in Africa; section III presents the theoretical and empirical literature linking aggregate demand to structural transformation;

section IV provides a summary of the methodology used; section V presents the estimation method and results; and section VI provides conclusions and discussions on policy implications.

(7)

3

II. Africa’s growth and structural transformation

A. Economic growth patterns

African economies have experienced high growth rates in recent years, with an annual average growth rate of 5 per cent between 2000 and 2014, even though growth took a dip after 2014, due to the decline in commodity prices when average growth declined to 3.5 per cent and 1.7 per cent in 2015 and 2016, respectively. According to the Economic Commission for Africa (ECA), since 2000, the growth in the region outperforms that of the developed economies and all other developing regions, except for South and East Asia (ECA, 2017). The recent growth experience, however, was not accompanied by the desired structural transformation to enhance employment and improve standards of living among Africans (ECA, 2014).

In many countries, the share of industry in terms of gross domestic product (GDP) and employment declined. The composition of GDP by sector has remained nearly unchanged since the 1960s. Agriculture contributed about 18 per cent of GDP in the post-independence planning period, and its share of value added remained at 17 per cent in the 2000s.2 The share of industry also remained nearly unchanged, as it contributed nearly 35 per cent of total value added between 2000 and 2016, with the manufacturing sector contributing an average of 13 per cent of the value added for the period 2000–2016. Notwithstanding the share of the manufacturing value added in total GDP falling for nearly four decades, it has picked up recently, growing from 12 per cent in 2007 to 15 per cent in 2016, with manufacturing production growing by nearly 78 per cent for that period (see figure I). The service sector has been the strongest driver of growth in all periods in Africa. It has consistently exceeded other sectors, rising from 47 per cent of value added in 2000 to 55 per cent in 2016.

Figure I

Africa’s share of manufacturing in gross domestic product is (a) falling and (b) real production is increasing

(a) (b)

Source: World Bank, 2017.

2 Economic Commission for Africa (ECA) calculations are based on data from the World Bank, 2017.

0 50 100 150 200 250

1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016

Manufacturing, value added (constant 2010 US$), billions

0 5 10 15 20

1980 1984 1988 1992 1996 2000 2004 2008 2012 2016

Manufacturing, value added (percentage of GDP)

(8)

4 Even with the major challenges African countries faced some have recorded noticeable success in their manufacturing sector performance. This was achieved by using various measures and policies, including the creation of export processing zones (such as those in Mauritius and the United Republic of Tanzania); consolidation of the competitive position of the national industry in the global value chain through the emergence of new industries that specialized in high value-added products (such as those in Morocco); or the development of labour-intensive and relatively low-technology manufacturing activities with strong backward linkages to the agricultural sector (such as those in Ethiopia).

There is evidence in the literature that during long periods, a sectoral shift in developed economies has been accompanied, and to some extent, been caused by increases in labour productivity in the areas where the structure of domestic production has shifted. This led to an increase in share of manufacturing in GDP, which itself led to a further increase in labour productivity. Africa’s GDP accelerated to an average annual rate of 5.4 per cent during the period 2000–2010, but declined to 3.3 per cent during the period 2010–2015. During the two periods, productivity growth also declined from 2.3 per cent (2000–2010) to 0.8 per cent (2010–2015) (McKinsey Global Institute, 2016). At the sectoral level, the average productivity growth for manufacturing, services and agricultural sectors declined from 8.3 per cent, 7.8 per cent and 10.5 per cent, respectively (2000–2010), to 3.6 per cent. 2.9 per cent and 4.8 per cent, (2010–2014). It is important, however, to note that manufacturing is the only sector in which productivity increased from 1.9 per cent in 2013 to 3.3 per cent in 2014, while productivity declined significantly in the services and agricultural sectors. This, to some extent, could be supporting the narrative that Africa’s structural transformation is characterized by the movement of labour from higher to lower productivity activities, thereby having a negative effect on both productivity and growth (McMillan and Rodrik, 2011).

B. Africa’s demographic dynamics

Demographic dynamics involving a country’s population structure, urbanization, fertility and mortality rates have profound implications for a country’s structural transformation and growth. Africa is the continent with the fastest-growing population and the fastest and largest urbanization rate in the world; and the bulk of African countries have a very young population.

These dynamics have profound implications for African countries, as it implies a large proportion of young adults in the working-age population (over 40 per cent), a rapidly growing school age population, and high rates of workforce growth that are often associated with high levels of unemployment and political instability or backlash (Cincotta, 2010).

If the African countries manage their demographic transition wisely, a window of opportunity will open up (for example, through the demographic dividend) for faster economic growth, human development (Ross, 2004), and structural transformation. Indeed, demographic factors affect a country’s structural transformation through their effect on aggregate demand (and thereby national income and employment or prices, or both) by having an effect on consumption, net private investment and/or government expenditures on goods and services (Coale, 1960).

Over recent decades, Africa’s population has grown at an average of 2.6 per cent per year in the period 1990–2015, more than twice the world average (ECA and UNFPA, 2016).

Even though the overall rate of population growth is slowing down, Africa’s population will continue to grow as a result of the momentum built up (ECA, 2017), signifying long-term

(9)

5 growth fundamentals underlying Africa’s future growth. Africa’s population is set to continue to rise to almost 1.7 billion in 2030 and close to 2.5 billion in 2050. This will lead Africa’s share of the world population to increase from currently around 16 per cent to almost 20 per cent in 2030 and above 25 per cent in 2050 (medium scenario projections of the United Nations). The issues relating to population not only relates to its size, but also the interlinkages with components of the population, such as age, structural composition, density, distribution and their characteristics, which also have implications on a country’s demand and economic growth.

Africa’s demographic trends reveal an unprecedented growth of the young population and labour force. As the delayed effect of a mortality rate falling faster than the fertility rate, the number of young people is growing rapidly in Africa. Young people formed 19.4 per cent of the total African population in 2015, and is projected to increase by 42 per cent by 2030.

Africa’s population of young people is expected to continue to grow throughout the remainder of the twenty-first century, more than doubling from current levels by 2055 (UNFPA, 2016).

In addition, the active working age population (25–64 years) is growing more rapidly than any other age group, forming 36.5 per cent of the total population in 2015, and is projected to reach 38.1 per cent and 43.7 per cent of the total African population in 2025 and 2050, respectively. This young population and the growing labour force is a highly valuable asset for the ageing world, with 60 per cent of the world’s population living in countries with fertility rates below replacement rates, and in some countries, one-third of the workforce could retire in the coming decade. This has potential negative impacts on the countries’ economic growth prospects; however, by 2034, the expectation is that the African continent will have a larger working population than China or India, at 1.1 billion people, while its dependency ratio is declining (McKinsey Global Institute, 2016). An expanding working-age population is associated with strong rates of GDP growth and offers potential demographic dividend. Indeed, an expanding workforce has many potential economic benefits but it also poses a challenge as well, and hence the need to create sufficient jobs and equip the young population with the skills necessary to acquire employment.

C. Rapid urbanization and the rise of the consuming class

Alongside this demographic transition is the issue of rapid urbanization that many African countries are experiencing. Africa, along with parts of Asia, is the fastest and largest urbanizing region, though with the lowest proportion of urban population (ECA, 2017). In 2014, Africa was 40 per cent urbanized, and by 2050, it is projected to reach 56 per cent (United Nations Department of Economic and Social Affairs, 2014). The historic shift from a majority- rural to a majority-urban continent will occur around 2037 (McKinsey Global Institute, 2016).

In addition, the recent demographic trends in Africa are characterized not only by unprecedented rapid growth rates and extensive urbanization, but also by the rise of the middle- class African consumer.

During the past two decades, strong economic growth in Africa has been accompanied by the emergence of a sizeable middle-class population, defined as those earning between $2 and

$20 per day (AfDB, 2011). Africa’s middle class has grown from 126 million in 1980 (27 per cent of the population), to around 350 million people (34 per cent of the population) in 2010.

This represents an average growth rate of 3.1 per cent in the middle-class population from 1980 to 2010. This middle class is projected to continue to grow and reach 1.1 billion (42 per cent of the population) by 2060. The more affluent lifestyle associated with the middle class has

(10)

6 contributed to increased domestic consumption in many African countries, which grew rapidly at 5.8 per cent annually during the period 2000–2005, and 5.2 per cent during the period 2005–

2010, before declining to 3.9 per cent during the period 2010–2015, reflecting the continent’s economic slowdown (McKinsey Global Institute, 2016). The middle class is a key source for private sector growth in Africa, accounting for much of the effective demand for goods and services (AfDB, 2011).

III. Demand and structural transformation: theory and evidence

Structural transformation refers to the interrelated processes of structural change that accompany economic development. In addition, the principal changes in structure emphasized in economic literature includes increases in the rate of accumulation (investment), shifts in the sectoral composition of economic activities (industrialization) and changes in the location of economic activity (urbanization), along with other aspects of industrialization, such as demographic transition and income distribution. The accumulation of human and physical capital and shifts in the composition of demand, trade, production and employment are described as the economic core of the transformation process. Countries that manage to pull out of poverty and become richer are those that are able to diversify away from agriculture and other traditional products. As labour and other resources move from agriculture into modern economic activities, overall productivity rises, and incomes expand (McMillan, Rodrik and Verduzco-Gallo, 2014).

Structural transformation literature builds on the long-standing empirical observation that the composition of economies differs systematically by stage of development (Herrendorf, Rogerson and Valentinyi, 2014). In its initial form, composition was mainly understood as the relative size of the broad sectors – agriculture, industry and services. The process entails a declining share of agriculture in total output and employment, the rise of a modern industrial and service economy; rapid urbanization as people migrate from rural to urban areas; and a demographic transition from high birth and death rates to low birth and death rates.

The literature proposed specific models of development that explained structural transformation as an endogenous process in response to factor accumulation, increasing wealth, and sector-specific properties of demand and production functions. With a steady decline of agriculture’s share of both employment and GDP, a pattern emerged that strongly associated with income growth, urbanization, poverty reduction, and a demographic transition from high birth and death rates common in rural ones associated with better health standards. The outcome of structural transformation is an economy with well-functioning sectors and output markets that equalize the capital and labour productivity between agriculture and non-farm industrialservice and other sectors, leading to inclusive economic growth (Barret and others, 2015).

Structural change in the economy implies that some industries or sectors experience faster long-term growth, compared with others, leading to shifts of the shares of these industries or sectors in total aggregate (Kruger, 2008). The differential impact of technological innovations on the production sectors, the differences in income elasticities of domestic demand for various goods, and the changing comparative advantage in foreign trade leads to rapid changes in production structures (Kuznets, 1973). This emphasizes the two central causes of structural change of varying income elasticities of demand and the differential impact of technological progress on the sectors.

(11)

7 Structural change leads to a characteristic pattern of change among the three sectors of the economy, along with changes in the industry composition of these sectors. Supply and demand factors closely interact in shaping the process of structural change. On the supply side, production technologies allow for producing the same goods with lower unit costs leading to productivity improvements. On the demand side, factors such as prices and quality influence the quality and composition of demand. The interaction of these factors gives structural change a specific direction and influences the speed at which this process takes place. This affects the growth of aggregate output, employment and productivity.

Theoretical literature has proposed several channels as drivers of structural change. The most prominent are the differences throughout the sectors in productivity growth (Ngai and Pissarides, 2007) or in capital intensity (Melck, 2002; Acemoglu and Guerrieri, 2008) on the supply side, and income effect on the demand side.3 At present, literature has mainly emphasized supply-side sources, such as technology, innovation, human and physical capital and exchange rate policies (through the Dutch disease models). Productivity improvements resulting from innovation and technological change are considered as one of the major driving forces of structural changes. Conversely, less attention has been devoted to demand-side factors (typical of the post-Keynesian approach) that are crucial for defining the extent of the market, the division of labour and the dynamics of productivity in the economic system.

Most of the theories are based on the tenet that technological progress drives structural change, but it is frequently the demand side that is crucial for determining the industry that grows faster and the one that shrinks, which therefore shapes the direction of structural change.

Changes in the structure of production and employment results either from sectoral differences in productivity growth or from sectoral differences in income elasticities of demand. Here, the fundamental cause of structural change is the hierarchical nature of needs.

During the process of economic growth, consumers are saturated with existing products, so in order to sustain growth new goods have to be continuously introduced. This nature of demand implies that structural change takes the form of the re-allocation of resources from old to new industries (Foellmi and Zweimüller, 2005).

It is argued that the demand side determinants of productivity performance entail that the division of labour enhances workers’ skills and know-how, induces the introduction of technological innovation and changes the sectoral structure of the economy as new industries emerge leading to further specialization and productivity growth (Crespi and Pianta, 2008). It is important to note that these factors eventually form the supply-side factors that affect the structural transformation process.

Among the most uniform of changes in demand affecting industrialization and structural transformation is the decline in the share of food consumption and the rise in the share of resources allocated to investment. Literature, however, reveal that the share of consumption (food and private) declines over time, while that of non-food consumption and investment rises,

3 It should be noted that the authors are following some trends evident in the literature, for example, use of the term “manufacturing” in the present context to refer to all activity that falls outside of agriculture and services. It is more appropriate to refer to the category as “industry”, because in the case of African countries, the industry sector data include the mining sector, a category that is not included in the present study.

(12)

8 implying a shift in demand away from agricultural goods to industrial commodities and non- tradables (Syrquin, 1988b).

Consumption habits that emanate from a rise in standards of living can also initiate structural change, as consumers’ opportunities for satisfying their needs increase in tandem with their incomes. Shifts in consumption patterns will therefore have an impact on the structure of an economy due to its effect on resource allocation and consumer’s demand changes. In this regard, structural change is driven by a non-homothetic household utility function and differences in the income elasticity of demand in goods, with a lower income elasticity on agricultural goods than on non-agricultural goods. As capital accumulates and income rises throughout the development process, these differences shift demand and therefore resources and production, from goods with low demand elasticity (such as food or other necessities) to high demand elasticity goods (such as services or luxuries) fostering a relative agricultural decline in the economy (Matsuyama, 1992; Laitner, 2000; Kongsamut, Rebelo and Xie, 2001; Gollin, Parente and Rogerson, 2007; Foellmi and Zweimüller, 2008; and Gollin and Rogerson, 2014). The decline of the food consumption share is consistent with one key feature of structural change: the decline of the agriculture sector, in terms of both employment and value added, during economic development. The subsistence demand of agriculture products plays a central role to model the structural transformation.4 In addition, the service consumption will continue to rise, driven by the demand for skill intensive services, which coincides with a period of rising relative wages and quantities of high-skilled workers (Buera and Kaboski, 2012a).

Recent research has examined alternative explanations for structural transformation, including barriers to labour reallocation and adoption of modern agricultural inputs (Restuccia, Yang and Zhu, 2008); scale economies (Buera and Kaboski, 2012a); human capital (Buera and Kaboski, 2012b); education and training costs (Caselli and Coleman, 2001); tax changes (Rogerson, 2008); transportation improvement (Herrendorf, Schmitz and Teixeira, 2012b);

international trade (Uy et al. 2013); and population growth (Echevarria, 1997; Acemoglu and Guerrieri, 2008).

Aggregate demand generally plays a small role in modern growth theory, but it could have significant implications for structural transformation, mainly in developing countries.

This could be through these countries’ increasing growth, urbanization, trade, capital accumulation, through their economic policies and other factors, having implications for aggregate demand that has been one of the driving forces of these economies, in particular those in Africa (ERA, 2017).

Urbanization stemming out of industrialization and including the wake of modernization, have significant effects on structural change, although, indirectly in the long run. As economies develop, the share of agriculture in employment falls and workers migrate to cities to look for employment in the industrial and service sectors (Clark, 1940; Lewis, 1954; Kuznets, 1957).

The economic history of every country reveals a close correlation between industrialization and urbanization and many countries have attained structural transformation while experiencing urban change (Bose, 1961). Modern large-scale industries cannot develop unless there are adequate economic and social overheads and economies of scale, which will only be available in big towns and cities as they stimulate trade and commerce through their increased

4 See, for example, Echevarria (1997); Laitner (2000); Kongsamut, Rebelo, and Xie (2001); Gollin, Parente and Rogerson (2007); Restuccia, Yang and Zhu (2008); Duarte and Restuccia (2010); and Alvarez Cuadrado and Poschke (2011).

(13)

9 demand for goods and services. The growth in the number and size of towns invariably redistributes the labour force into the manufacturing and service sectors, which was initially absorbed in primary activities such as the agricultural sector.

The economic advantages of urbanization are rooted in economies of scale, also known as agglomeration economies, which arise from the proximity of economic agents and their interaction in the factor and products markets (ERA, 2017), further enhancing structural changes in the country’s economy. Urbanization provides the infrastructure necessary for industrial development and increases the efficiency of labourers by providing them with suitable employment opportunities.

Countries that have succeeded in structural transformation are urbanized and they feature a consumption and production pattern driven by productive industrial and services sectors.

Those countries have relocated resources (including labour) from low to high-productivity sectors. A key element of structural transformation involves the movement of labour out of rural activities into urban ones. Historically, as economies develop and incomes rise, the share of income that people spend on food declines, and demand for manufactured products rises. At a later stage of development, a similar process of shifting demand takes place that favours services. Changes in economic structure accompany such changes in demand and trade, with the share of employment in agriculture declining and that of industry or urban-based services rising. This ` process is generally accompanied by increasing accumulation of human and physical capital and diversification (Chenery, 1982). The shift in sector employment from agriculture to industry and services is accompanied by productivity increases. In addition, in cities, knowledge and ideas generate increasing returns to scale. Firms invest in research and development to reap the benefits of productivity, increase their market share and maximize their profits.

Capital accumulation remains one of the main drivers of aggregate demand of any economy. Investment, especially if aimed at reducing unemployment and raising people’s living standards, will facilitate shifts in labour force from less productive to more productive sectors of the economy.

It has recently been argued that investment-specific technological change could be a new and fundamental force behind structural transformation, in particular the rise in services due to the decline in agriculture and manufacturing (Garcia-Santana, Pijoan-Mas and Villacorta, 2016; Guo Hang and Yan, 2017). For a long time, the literature on structural transformation had been built on the premise of balanced growth hypothesis (with constant investment rate).

Recent studies, however, have shown that changes in the sectoral composition of investment cannot be explained by the relative price of inputs alone, but also by investment rate and investment-specific technological change, that shifts the investment demand from agriculture and manufacturing to services. This challenges the recent growth literature on structural transformation, which postulates that changes in the sectoral composition of growing economies are associated with a balanced growth path. In explaining what has led to the rise in the services sector with a reduction in the agriculture and manufacturing sector in China, Guo, Hang and Yan (2017) argue that changes in sectoral composition of investment and investment rate affects sectoral shares of total value added. Changes in sectoral composition of investment could be attributed to a technological revolution based on the development and deployment of information technology, leading to structural transformation. For China, investment-specific technological change has shifted the demand for various categories of input producing investment goods leading to profound implications for structural transformation.

(14)

10 The composition of trade and the type of specialization are determined in large part on the availability of natural resources, traditional factor proportions and by policy. In practice, however, the evolution of comparative advantage and commercial policies have combined to create an export pattern that reinforces the shift from primary to industry, which is implicit in the pattern of domestic demand (Syrquin, 1988b).

The premise of endogenous growth theory and new trade theory is that knowledge and technology intensive commodities embody higher levels of productivity and sophistication that manifests itself in their added value. It is considered that knowledge, technology transfer and horizontal and vertical spillovers between firms and industries have beneficial effects on the structural transformation of exports and building international competitiveness (Stojcic and Orlic, 2016). This leads to export diversification through an increase in the number of export sectors (hence export earnings) and knowledge spillovers arising from improved production techniques, new management practices and skills training and improved productivity with profound implications for a country’s structural transformation.

Trade allows countries to specialize in areas where they have a comparative advantage, leading to an increase in economic activity and employment in the export sector. Specializing in areas of comparative advantage, shifts resources to their most productive use, increasing the value of aggregate production and income by facilitating an efficient allocation of production across and within economies. Developing countries have integrated more into the world economy and have been more open since the early 1990s. This has facilitated technology transfer and contributed to efficiencies in production leading to structural changes and growth in many countries; however, this has not led to the expected gains in Latin America and sub- Saharan African countries, as globalization and trade openness seem to have led to negative structural change effects. Labour has moved from high-productivity to low-productivity sectors and activities (McMillan and Harttgen, 2014). Empirical work on trade liberalization has also revealed that import competition has forced manufacturing industries in developing countries to become either more efficient or those that are least productive to exit the industry, leading to reallocation of labour and other resources.

IV. Methodology and data

The methodology used in this study follows and extends the classic Lewis-type dual economy model, in which workers were moved out of subsistence or traditional agriculture into the modern manufacturing and services sectors. Apart from the difference in relative wages being a central feature in accounting for those factors, the focus of the study was on the role that demand factors might play in Africa’s growth, especially with regard to the continent’s structural transformation process. The structural transformation measure (mainly sectoral output in this regard) will be regressed on a number of plausible and mainly demand-related variables. Following the demand and supply analysis of Cornwall and Cornwall (2002) coupled with the structural transformation and demand narrative above, the following model applied is presented:

𝑦𝑖𝑡 = 𝛿𝑋𝑖𝑡+ 𝛽𝑍𝑖𝑡+ 𝜇𝑖𝑡 ...(1)

𝒚 is the proxy capturing structural transformation, 𝑿 is a vector of variables identifying the stages of development in accordance with the structural/economic transformation transition, which could be captured by GDP per capita, consumption (household and government consumption), investment (gross fixed capital formation), trade openness and government

(15)

11 policy. 𝒁 is a vector of plausible demand-related economic variables, such as urbanization, capital–labour ratio and human capital, among others. 𝜹 and 𝜷 are coefficient vectors, and 𝝁 is an error term, while 𝒊 and 𝒕 refer to country and time period, respectively.

Given the definition mentioned above of structural transformation as “the reallocation of resources from lower- to higher-productivity activities, shifting typically from agriculture to manufacturing sector and modern services and within each of these sectors from lower- to higher-productivity niches” (ECA, 2016), the value added of the agriculture, manufacturing and services sectors as proxies of structural transformation is retained in this study. The sectoral value added is also replaced with sectoral productivity growth to assess to what extent the effect that these demand factors have on productivity growth, hence the associated resource allocation in the various sectors.

A. Data and variable definitions

Rate of economic growth is used as one of the explanatory variables in the estimations to capture the stage of economic development in an economy (Cornwall and Cornwall, 2002).

Growth affects the structure of the labour force and industry and income proportions of other sectors of the economy, and because of these shifts accompanying economic growth, there will be more labour mobility, higher levels of capital goods, output and drastic changes in the composition of exports and imports, leading to significant shifts of resources between sectors.

Consumption habits, as a result of an increase in income leading to a rise in standards of living, also initiate structural change – as income rises a large proportion of income tends to be spent on goods that satisfy consumers’ needs.

One of the factors that governs structural change, especially with regard to the stages of development is government policy and in particular planning, which is regarded as the most potent weapon in the armoury of the state to manipulate the necessary structural change, especially the reallocation of production. Competition due to liberalization policies have caused many industries in developing countries to shut down and release labour to less productive activities, such as agriculture and low productivity services or the informal sector.

A related issue concerns the real exchange rate in which overvalued exchange rates, either driven by disinflationary monetary policy or increased foreign aid inflows, squeezes tradable industries as exports become expensive and imports become cheaper, while, in contrast, competitive or devalued real exchange rates could promote tradable industries. In that regard, Africa’s planning policy is grouped into three periods: 1960s–1980s; 1980s–2000s; and from the 2000s to now, to capture the import substitution period, the structural adjustment period and the rebirth of the planning period, respectively. The reason for the use of dummies is to make it possible to capture the effects of these policy measures; an indicator of trade openness in the regressors is used to capture the effects of integration and trade openness on the reallocation of resources in an economy.

In terms of urbanization, the economic history of every country reveals a close correlation between industrialization and urbanization. A proportion of the urban population has been used to capture the extent of urbanization and how it has affected the reallocation of resources among the various sectors. Urbanization has mainly been associated with structural change in industry and employment towards non-agricultural activities. In developing countries, however, market imperfections, such as labour immobility, ignorance of market conditions and rigid social structure have been acting as impediments to the proper utilization of resources (Kuchal, 1966).

(16)

12 A large body of work exists that posits technology-driven structural change. According to this theory, when the economy is poor, the economy–wide capital–labour ratio is low and the relative rental rate of capital to labour is high, implying that a sector with a larger capital share receives relatively less labour. As the economy develops, the capital–labour ratio increases and the relative rental rate of capital to labour decreases, implying an increase in the relative labour of this sector (Acemoglu and Guerrieri, 2008), thus reducing productivity levels, leading to resource reallocation in the economy. In most developing countries, labour is the abundant factor of production and therefore the capital–labour ratio is expected to relate negatively to the structural transformation process as it could have a negative effect on the efficiency of firms, which supports the factor endowment theory. In the traditional Heckscher- Ohlin-Samuelson (H-O-S) model based on perfect competition assumptions, it is argued that trade reflects the interaction between the characteristics of countries and their technology. The proposition that emerges is that countries will export goods whose production is intensive in the factors in which they are abundantly endowed. Capital intensity could also negatively relate to technical efficiency if it captures the levels of sunk costs, which may create barriers to entry or exit and therefore hamper the structural transformation process.

B. Key control variables

In addition to the variables defined above, control variables were included during the estimations to control the various economic factors. These variables also included FDI, foreign aid flows, life expectancy, natural resource endowments, quality of institutions and life expectancy.

Literature on structural transformation has contained discussions on the role of FDI and foreign aid (sum of commitments received from donors, including international organizations) with mixed evidence. A good orientation of FDI inflows into industries, especially manufacturing, might increase the productivity of the labour force, by having a positive and significant impact on the value added of industries (Takii, 2005; Boly and others, 2015).

Foreign capital can lead to an increase in the stock of physical and human capital, hence their relative productivities. It can also trigger technology accumulation in the domestic manufacturing industries, mainly if the technological gap between domestic and multinational firms is higher as is the case in Africa (Todo and others; 2006; Fauzel, 2012). Another important channel relates to the creation of forward and backward linkages with domestic firms (Amendolagine and others, 2013). By boosting the likelihood of spillovers and technological transfers between foreign and domestic firms, these linkages could be components of FDI-induced structural change and productivity growth.

Foreign aid can be viewed as economically transformative if channelled into supporting the continent’s structural change, in particular its “weakest link”, such as the manufacturing sector and the associated bottlenecks (for example, high transport costs and irregular access to electricity) (Sindzingre, 2015). Aid can act as a complement or substitute for the scarce public and private capital in low-income countries and for investment towards industrial sectors and others, including health, education and infrastructure. Literature on structural transformation gives a variety of responses to the question of why some economies have the capacity to achieve “structural change”, while others do not, such as being able to accumulate productive knowledge (Hausman, 2012) or foster innovation (Rodrik, 2013). Both Hausman and Rodrik underscore that the recurrence of market failures (regarding, for example, informational spillovers and coordination of investment policies) in developing countries make public action necessary, in particular industrial policies. The literature acknowledges the relevance of

(17)

13 institutions for the success of industrial policies, political leadership, coordination councils and accountability mechanisms.

The quality of institutions is an important factor in the design and implementation of an effective industrial policy (UNCTAD, 2016). Strong institutions facilitate such a policy and enable governments to use a wider set of industrial policy instruments. Institutions also influence distribution of power and rents in the society, affecting production structures, income levels and inequality, among other things. In the African case, the argument is that inequality and weak institutions create a system in which centralized power and informal loyalty networks often limit industrial policy incentives. This could affect the development of the private sector and lead to deepening inequalities and ethnic conflicts (Altenburg, 2013; Altenburg and Melia, 2014). While these institutional factors have, to some extent, contributed to the design and implementation of industrial policies, authors have argued that institutions evolve and strengthen with economic development and growth and, as such, it is possible to achieve structural transformation in the contexts characterized by weak institutions (Cervellati and others, 2008; Khan, 1996).

Aggregate changes in human capital (for example, education and health) can also affect structural change (Caselli and Coleman II, 2001; Afonso, 2012). The accumulation of human capital enables technology-intensive activities and new business opportunities, and endows the agents with management skills and technological knowledge (Justman and Teubal, 1991;

Afonso, 2012; Ciccone and Papaioannou, 2009). Countries with highly skilled workers tend to be more efficient in their endeavours when they incorporate advanced technologies into their activities. From a micro perspective, firms that have employees with a high level of human capital adopt complementary technologies in order to achieve maximum efficiency. The accumulation of human capital can also enhance the role of research and development in the economies by promoting the creation of new products (Caselli and Coleman, 2006; Bodman and Le, 2013), leading to structural changes in an economy.

In addition, there is a control variable for the natural resources effects. Dutch disease effects may challenge the industrialization of resource-rich countries. The economic explanation for these effects suggests that a resource windfall generates additional wealth, which raises the prices of non-tradable goods. This, in turn, leads to real exchange rate appreciation and higher wages in the non-tradable sector. The resulting reallocation of capital and labour to the non-tradable sector and to the resource-rich sector causes the manufacturing sector to contract (de-industrialization). In addition, developing countries, well-endowed with natural resources and primary products, will face a reduction in incentives to diversify towards modern manufacturing and modern service sectors (Macmillan and Rodrik, 2011). Moreover, natural resources may create disincentives for human capital investment (Gylfason, 2001; Stijns, 2005; Suslova and Volchkova, 2007). High resource dependence leads to an increased volatility that will decrease long-term investment, such as, education thus, impeding human capital accumulation.

Some authors find that a resource boom causes a decline in manufacturing exports and an expansion of the service sector (Harding and Venables, 2010). Primary sectors, such as mining, operate at a very high level of labour productivity, but with limited capacity to generate employment. Consequently, the expectation is that economies with a comparative advantage in natural resources will experience a limited contribution to structural change associated with participation in the international market. Recent developments have shown that natural resources may not necessarily represent a curse for developing countries, because with the right policy approach, commodity-based activities can benefit industrial policies and foster

(18)

14 structural transformation (Torvik, 2009; Perkins and Robbins, 2011; Kaplinsky, 2011;

UNCTAD, 2014).

V. Model estimation and empirical results

It is important to note that the choice of the methodology is conditioned on the availability of data. It would have been preferable to calculate and employ the structural transformation index, as in McMillan (2014); however, due to a lack of country-level sectoral data, this was not possible. To assess the effect that demand factors have on Africa’s structural transformation, the following three equations were estimated:

𝑉𝐴𝑚𝑎𝑛𝑢𝑓𝑎𝑐𝑖𝑡 = 𝛿(𝐺𝐷𝑃𝑝𝑝𝑐𝑜𝑛𝑖𝑡+ 𝐻𝐶𝑖𝑡 + 𝐺𝐶𝑖𝑡+ 𝐺𝐹𝐶𝐹𝑖𝑡+ 𝐺𝑜𝑣𝑃𝑜𝑙𝑖𝑐𝑦𝑖𝑡+ 𝑇𝑟𝑎𝑑𝑒) + 𝛽 (𝑈𝑟𝑏𝑖𝑡+ 𝐶𝑎𝑝𝐿𝑎𝑏𝑅𝑎𝑡𝑖𝑜𝑖𝑡+ 𝐸𝑑𝑢𝑐𝑖𝑡) + νi + 𝜇𝑡 + εit...(2) 𝑉𝐴𝑆𝑒𝑟𝑣𝑖𝑖𝑡 = 𝛿(𝐺𝐷𝑃𝑝𝑝𝑐𝑜𝑛𝑖𝑡+ 𝐻𝐶𝑖𝑡+ 𝐺𝐶𝑖𝑡+ 𝐺𝐹𝐶𝐹𝑖𝑡+ 𝐺𝑜𝑣𝑃𝑜𝑙𝑖𝑐𝑦𝑖𝑡+ 𝑇𝑟𝑎𝑑𝑒) + 𝛽 (𝑈𝑟𝑏𝑖𝑡+ 𝐶𝑎𝑝𝐿𝑎𝑏𝑅𝑎𝑡𝑖𝑜𝑖𝑡+ 𝐸𝑑𝑢𝑐𝑖𝑡) + νi + 𝜇𝑡+

εit...(3)

𝑉𝐴𝑎𝑔𝑟𝑖𝑐𝑢𝑙𝑖𝑡 = 𝛿(𝐺𝐷𝑃𝑝𝑝𝑐𝑜𝑛𝑖𝑡+ 𝐻𝐶𝑖𝑡 + 𝐺𝐶𝑖𝑡+ 𝐺𝐹𝐶𝐹𝑖𝑡+ 𝐺𝑜𝑣𝑃𝑜𝑙𝑖𝑐𝑦𝑖𝑡+ 𝑇𝑟𝑎𝑑𝑒) + 𝛽 (𝑈𝑟𝑏𝑖𝑡+ 𝐶𝑎𝑝𝐿𝑎𝑏𝑅𝑎𝑡𝑖𝑜𝑖𝑡+ 𝐸𝑑𝑢𝑐𝑖𝑡) + νi + 𝜇𝑡 + εit...(4)

The abbreviations used for the equations signify the following: VAmanufactit is the value added of the manufacturing sector, (as a percentage of GDP) in country i for year t; VAagricultit

is the value added of the agriculture sector (as a percentage of GDP); VAserviit is the value added of the services sectors (as a percentage of GDP); 𝐺𝐷𝑃𝑝𝑝𝑐𝑜𝑛 is the GDP per capita (purchasing power parity (PPP), constant 2011); HC is the share of household consumption expenditure (as a percentage of GDP); GC is the share of general government final consumption expenditure (as a percentage of GDP); GFCG is the share of gross fixed capital formation (as a percentage of GDP); GovPolicy is a proxy for government policy; Trade openness is the ratio of the sum of exports and imports to GDP (percentage); Urb is the rate of urbanization (urban population as a percentage of total population); CapLabratio is the capital–

labour ratio, that is, capital stock (at current PPPs) divided by the number of employed persons;

Educ is the education enrolment ratio, secondary (net percentage); and 𝛿 and β are the coefficients of interest to be estimated. Information on the variables, definitions and sources of data are provided in annex I, table AI.

The country’s fixed effects (νi) control for all time-invariant differences between countries, and year fixed effects (μt) control for factors that may affect countries’ growth and structural transformation; and εit is the error term. It is important to note that as the effect of education on structural transformation may take time to materialize, the time lag will likely differ depending on the context, the country and the type of reform. Experimenting with various time lag lengths when estimating the effects of education quality, resulted in the five-year lag on the education variable being the most appropriate to use. Similarly, in the choosing of a one- year lag on the GDP per capita variable, given that the effects of any changes will take time to materialize. In the second part of the analysis, the sectoral value added for equations (2)–(4) was replaced by sectoral productivity growth as a measure of structural transformation.

(19)

15 In carrying out the initial analysis and tests, the Wooldridge (2002) test for autocorrelation showed the presence of first order autocorrelation among some variables involved in the analysis. In addition, the Breusch and Pagan (1979) and the White Tests (for the Pooled OLS estimation), and the Modified Wald Test for group-wise heteroscedasticity (for the fixed effect regression model), revealed the presence of heteroscedasticity (see annex II, tables AII.1–AII.4 for test results) (Wald, 1940; White, 1980a; White, 1980b). Accordingly, the IV-GMM regression method that allows for correcting unobserved fixed effects, heteroscedasticity, serial correlation, endogeneity and issues of reverse causality and simultaneity bias, was deemed appropriate for the analysis. More specifically, the two-step GMM that presents some efficiency gains over the traditional estimators derived from using the optimal weighting matrix, the over-identifying restrictions of the model, and the relaxation of the independent and identical distribution (iid) assumption (Baum, Schaffer and Stillman, 2007) is used.5 The data for the analysis applies to a period of 54 years (1960–2014) covering 54 countries.6 Equation (1)7 was therefore estimated using a IV-GMM estimation method, using the panel equivalent of the Newey-West variance covariance matrix.8 Standard econometric tests for instrument suitability were also carried out (see the table on regression results below). The Wald test indicated the inclusion of the time-specific fixed effects in the model.

A. Empirical results

All the models present a good quality of adjustment reflected in a higher adjusted coefficient of determination (𝑅̅2), and the F-statistic and the relevant p-values, indicate that all the models are well specified, that is to say, the explanatory variables relate statistically and significantly to the structural transformation dependent variables.

The results of the Hansen (J-test) for over-identification also show that the chosen instruments are valid. The Hansen J statistic, tests the hypothesis that the variables are jointly exogenous. The Chi-square p-values of the Hansen test (which are all greater than 10 per cent) fail to reject the hypothesis of joint exogeneity of the instrumental variables. In addition, Cragg- Donald Wald F-statistic (Cragg and Donald, 1993) and Kleibergen and Paap (2006) Wald F- statistic at the bottom of the tables, indicate very strong instrumentation. As heteroscedasticity- robust standard errors are used in all regressions, the use of the Kleibergen-Paap statistic is more appropriate; it generalizes the Cragg-Donald statistic to the case of non-iid errors, allowing for heteroscedasticity, autocorrelation and/or cluster robust statistics Baum, Schaffer and Stillman (2007). In the special case of a single endogenous regressor considered in this section, though, the Kleibergen-Paap and Cragg-Donald Wald statistics are, respectively, simply the standard robust and non-robust first-stage F statistics.

5 Appropriate macroeconomic and country-specific indicators are derived from the literature and instrumented.

6 Some countries do not have the full set of data for the entire period (for example, the data for South Sudan since 2006). Using an IV-GMM allows to handle this unbalanced dataset.

7 𝒚𝒊𝒕= 𝜹𝑿𝒊𝒕+ 𝜷𝒁𝒊𝒕+ 𝝁𝒊𝒕 …………..(1) (also provided in the first paragraph of the present section).

8 The panel data version of the estimation command “ivreg2” with the “gmm2 cluster (iid)” options from the statistical software Stata 14.

(20)

16 According to Stock and Yogo (2005), the Cragg-Donald F statistic must exceed five if statisticians are to be confident at the 5 per cent level that the bias to the coefficient estimate on the structural transformation variable is less than 30 per cent of the GMM bias. Critical values have not been tabulated for the Kleibergen-Paap rk statistic as the specific thresholds depend on the type of violation of the iid assumption, which invariably differ widely across applications. Nevertheless, standard F statistics well below unity are unlikely to exceed even the most generously low thresholds

The table on regression results below shows the estimates for equations (2)-(4). The coefficients of the GDP per capita are positive and statistically significant in the three models.

These results are also economically (quantitatively) significant given the elasticity estimates of 1.01, 1.14 and 0.98 for manufacturing, services and agriculture sectors, respectively, derived from the IV-GMM estimation results as expected theoretically. For structural transformation to take place, one should first and foremost observe economic growth. The interesting part, regarding the present study, is how this growth has had an impact on the reallocation of resources in the three sectors.

Among the three sectors, income has the highest impact on services (explained by the growing importance of the service sector in Africa, previously mentioned, which contributes to approximately 50 per cent of GDP). The results show that an increase in GDP per capita of 1 per cent will boost the value added of the manufacturing sector by 1 per cent. This means that the relationship between economic growth and structural transformation in Africa is very strong. The result also indicates that growth has enhanced the service sector (1.1 per cent) more, in relative terms, than the manufacturing (1.0 per cent) and agricultural sectors (0.9 per cent), indicating that relatively more resources have been shifted towards the service sector.

Références

Documents relatifs

A major reason why agriculture in Africa has remained in subsistence form is that smallholders, who contribute around 80 per cent of Africa’s agricultural production, have

Marine Manke, Head, Labour Mobility and Human Development Division, International Organization for Migration; Carolyne Tumuhimbise, Migration Policy Officer and Liaison Officer to

growth from 1991 to 2010 was due to the structural transformation in the economy. Figure 7 shows how individual sectors contributed to these two effects. First, direct

The Malawi Farm Input Subsidy Programme (FISP) launched by the Malawian government in 2005, the Mahatma Ghandi National Rural Employment Guarantee Scheme (MGNREGS) introduced by the

The ancient examples were only a pretext and starting point for developing new solutions (Compare figs.. In light of these considerations, this chapter provides an overview of the

What effect could trade with, and investment and aid from, the BRICS (Brazil, Russian Federation, India, China and South Africa) have on growth, employment and

Part two on Structural Transformation and Natural Resources drew heavily from the knowledge of international experts gathered in Paris on November 23, 2012: Sambit

where by convention the † exchanges dotted with undotted indices and lowers the Isospin index.. Therefore at least two spinors have to