A The Dual Economy Model of Arthur Lewis

The Lewisian representation of development is based on an observation that seems universal among countries in the early stages of development: the coexistence of a mass of poor people sharing the income of low-productivity activities in subsistence agriculture or informal retail and handicraft occupation in rural or urban areas, on the one hand, and modern firms using more productive technologies and employing salaried workers at a higher level of earnings, on the other hand. The ‘dual economy’ model thus distinguishes between a modern or formal sector, essentially made up of firms organised according to a capitalist mode of production, and a traditional or informal sector, generally constituted by family farming or small family businesses in retail trade or handicrafts. Capital accumulation at the country level is taking place in the modern sector, which thus recruits an increasing number of workers at some given wage level. The marginal productivity of labour in the informal sector is assumed to be small, possibly zero if there is ‘surplus labour’ – meaning that the overall labour force may diminish in that sector without affecting the volume of production. Informal workers therefore receive earnings which, albeit low, are disconnected from the marginal productivity of labour. Various assumptions have been made about the level of these earnings. In Lewis, it is conveniently conceived as ‘customary income’, which remains roughly constant as long as the marginal productivity of labour in the informal sector is below that obtained in the formal sector.Footnote ³ More rigorously, competition in the labour market across the whole economy implies that the modern sector must pay a wage equal to, or somewhat above, earnings in the informal sector to attract workers, which modern firms can afford thanks to their capital-using technology. This assumption is of course the basis of the ‘dualism’ observed in many developing economies. Modern firms thus operate as if faced with what Lewis called, in his famous 1954 paper, an ‘unlimited supply of labour’. In alternative specifications, the wage is exogenously fixed in the modern sector at a level above the average income in the informal sector, which is allowed to vary with the size of the labour force operating in this sector. Various justifications have been adduced for such an apparent absence of competition on the labour market, in particular efficiency wage settingFootnote ⁴ or the imposition of a legal minimum wage in the formal part of the economy.

Within such a framework, development involves a structural transformation of the economy. Through a process known as ‘capital-widening’, capital accumulation in the modern sector leads to proportionate increases in the demand for workers, which help to gradually reduce the pool of excess labourers pumped into the informal sector. If accumulation proceeds at a rate faster than population growth, the share of employment and output originating in the informal sector falls with capital accumulation (in the modern sector). This process stops when there is no surplus labour left, or, roughly speaking, when the average productivity of labour in the informal sector has risen enough to catch up with the corresponding productivity in the formal sector. The reserve of labour becomes depleted and competition for workers forces the modern sector to raise its wage rate so as to remain able to attract additional labour force from the informal sector. It is then the case that earnings increase in both sectors, with the effect of prompting modern employers to use increasingly labour-saving and capital-using techniques, including in previously informal activities.Footnote ⁵ This process, called ‘capital-deepening’, succeeds the ‘capital-widening’ mechanism characteristic of the labour-surplus economy.

In this highly simplified description of the development process, the modern sector is the engine of growth of the economy and the structural transformation it generates is another way of describing the modernisation of the economy. Practically speaking, however, there are many ways in which this transformation process may be hindered or even halted. Let us mention a few of them. First, the accumulation rate in the modern sector may not be high enough, or population growth may be too fast, so that the volume of employment in the informal sector rises, even though its share in total employment may fall. Second, complementary production factors such as skilled labour or infrastructure may not be available or may grow too slowly. Third, the same may happen with imported inputs, such as key capital goods, if the capacity to export is insufficient or grows too slowly, unless foreign funds are available. Fourth, (unskilled) labour-saving technology imported from advanced countries may be so effective as to be technically efficient (or technically superior), so that it will be adopted by profit-maximising employers in spite of a relatively low cost of labour.Footnote ⁶ As a result, the labour absorption capacity of the modern sector is reduced.

In addition to structural transformation stalling for the preceding reasons, other complications may also arise in the simple dual model of development when amending it to accommodate different country contexts. An omitted factor in the model is land. As a matter of fact, Lewis stipulates that his theoretical framework is adapted to ‘overpopulated’ countries where land is scarce and fully used, so that labour in the agricultural subsistence sector is likely to be redundant. A priori, the situation is expected to be different in countries where land is abundant, which would still seem to be the case today south of Sahara. Yet, because of imperfect market integration caused by a deficient transport and communication infrastructure, lack of access to adequate technology or inputs, or the presence of institutional factors responsible for inefficient land allocation, farming may remain a low-income-earning activity in these countries in spite of its rich potential. Characteristically, such a situation will be reflected in the under-utilisation of both land and labour.Footnote ⁷ Structural transformation from low-productivity agriculture to higher-productivity sectors along the Lewisian mechanism would then remain a possible development strategy, but the promotion of a more efficient use of available land and a dynamic agriculture form another strategy for growth-cum-structural transformation. In the latter case, thanks to the removal of barriers to agricultural expansion, notably the physical isolation of remote areas, subsistence farming would progressively transform into a commercially integrated sector oriented towards domestic and/or foreign markets. As a matter of fact, the structural transformation would somehow be operated through informal production units being modernised or modern firms taking over traditional activities in rural areas. In the context of sparsely populated countries, because of the high per unit cost of connecting isolated areas in terms of providing not only transport and communication links but also social amenities, schools, health centres, and various agricultural support services, this second strategy may be deemed inferior to the Lewisian labour reallocation mechanism.Footnote ⁸ Yet it can certainly not be discarded a priori.

The simple model can be extended to take into account the demand side of the economy and the constraints that it may impose on growth and the structural transformation. In a closed economy, the expansion of the modern sector requires that the demand for its products expands at the same rate as it grows. If not, the relative price of modern products will fall, lowering the profitability of the whole sector. Capital will then tend to flee and the structural transformation to stall. This will also happen in an open economy if the modern sector is not able to compete with foreign produced goods in domestic and/or foreign markets. Even if effectively protected from foreign competition domestically, a point will be reached at which domestic demand will be saturated, so that any further extension of the modern sector will require operating on export markets, which will not be possible if it is not internationally competitive. To be sustainable, the structural transformation thus necessitates not only capital accumulation at some minimum pace but also competitiveness gains that will allow the modern sector to effectively compete with foreign producers in some product lines.

These product lines depend on the comparative advantage of the country. They may consist of labour-intensive manufactured goods, agro-industrial products – which require that the modern sector operates in agriculture or sources its inputs in traditional farms – or services such as tourism. What matters for the structural transformation to keep going is that no demand constraint bears on the expansion of the modern sector, even if this means a constantly evolving product mix.

This diversification of production may be uneasy because the external competitiveness of the modern sector is restricted to a narrow set of products, typically agro-industrial, or mineral and energy natural resources. The extraction of the latter employs little labour so that it does not directly contribute to structural change. It does it indirectly, however, thanks to the domestic demand that it generates through the income it brings to local agents, including the state, and the outlets it provides to the rest of the modern sector. Without export diversification, or possibly import substitution at some early stage, the pace of development of the economy is thus determined by the growth of primary commodity export net receipts, which may be lower than desired.

Despite its apparent simplicity, the Lewisian framework, as well as its extensions or variations in the development theory literature, offers a most useful guide to the understanding of a development process viewed through the lens of the structural transformation of the economy and the constraints weighing on it. This of course requires that the framework is sufficiently enlarged to fit the specific features of the country examined. As will now be seen, it is quite remarkable that the general dynamics of the model, suitably adapted to the context, has proven to be appropriate to describe the successful early development of East Asian countries such as South Korea and Taiwan between the mid-1950s and the mid-1970s, a period during which both countries would have been classified as low- or middle-income according to today’s international income scale. Differences between the two countries stem from the initial conditions, the pattern of structural transformation, and some major policy orientations associated with this pattern.

Back in the early to mid-1950s both countries had a level of real GDP per capita around US$1,400 at international 2017 prices – see Table 7.1. Taiwan had a slightly higher level of income than South Korea, and its share of GDP originating in agriculture, generally considered in the development literature as the main component of the informal sector and the reserve of surplus labour in the dual economy model, was then lower. Both on the aggregate and from a structural development point of view, Taiwan was thus slightly more advanced. In the two countries, however, the structural transformation in the following twenty years was astounding. The GDP share of agriculture went down by 25 percentage points in both South Korea and Taiwan, while the GDP share of manufacturing increased by 25 percentage points in Taiwan and 15 per cent in South Korea. It is more difficult to conduct the same calculation for employment because available data are not comparable across the two countries.Footnote ⁹ Nonetheless, it has been estimated that, helped by a slowdown of population growth, both countries reached the Lewisian turning point – at which the absolute volume of employment in agriculture starts declining – at about the same moment, in the early 1970s. By 1980, the GDP share of the industrial sector was as high as 45 per cent in Taiwan and 33 per cent in South Korea – in both cases, a high level by today’s standards. In the two countries, the dynamics of structural transformation by which low-productivity workers in agriculture and elsewhere in the economy were progressively absorbed into the growing modern sector of the economy, including manufacturing, thus appeared to work according to the Lewisian mechanisms set off by an extremely fast accumulation rate of capital in the latter sector. If the manufacturing export-driven expansion was a powerful engine of growth in the two countries, however, there were clear differences in the way it was activated and then sustained, and, as a matter of fact, in the way the Lewisian logic unfolded.

B Taiwan

In the case of Taiwan, the expansion of the manufacturing sector relied on a dynamic group of small and medium-sized enterprises (SME), a sizeable proportion of them being located in the rural sector and functioning symbiotically with an agricultural sector that was stimulated by a radical land reform and vigorous accompanying policies (see Park and Johnston, Reference Park and Johnston1995). The early emphasis on, and dynamism of, the agricultural sector in Taiwan are worth emphasising as they constitute a departure from the straight structural transformation model where overall productivity gains result from the mere sectoral shift of labour from agriculture to industry. Here, productivity gains within agriculture contributed to the aggregate gain, too. In any case, both in rural and urban areas, the SMEs provided rising employment opportunities to surplus agricultural workers, who were absorbed in labour-intensive manufacturing activities directed towards the domestic market, under tariff protection in a first stage, and then increasingly towards foreign markets.

The SMEs were incentivised to export by a set of policies which included a favourable exchange rate reform – a 60 per cent devaluation between 1958 and 1961 – and tariff relief for imported equipment and other inputs. The whole sector was highly competitive and financially autonomous in the sense that it could rely on internally generated savings and a small-scale private credit market, rather than on state bank credit (see Chu, Reference Chu1999). Quickly, a whole integrated network was formed, made up of SMEs buying and selling from each other in a highly efficient way. On the other hand, large firms, which were inherited from the era of Japanese colonisation and then nationalised, also contributed to the production and export of manufactured goods. Yet they proved less dynamic than the private sector and the expanding network of SMEs.

At the macro level, it must be reckoned that US assistance played a huge role in the early stage of Taiwan’s development, covering as much as about 25 per cent of public expenditures in the 1950s. However, as the Taipei government faced huge defence expenditures during that period, it is not clear how much foreign aid contributed to bridging the trade balance gap. What is certain, however, is that the trade balance regained an equilibrium position as early as 1963, and non-military US aid fell drastically after 1965.Footnote ¹⁰ The price stability achieved during these two decades, and later, is another achievement that needs to be stressed.

In short, the dramatic structural transformation of Taiwan from the mid-1950s to the mid-1970s relied on a powerful industrial export growth engine, itself based on the dynamism of a network of SMEs enjoying various favourable conditions: a stable macroeconomic context, undistorted prices and competitive market mechanisms, and a sound regulatory policy. The SMEs were also helped by several initial conditions: (i) an efficient agricultural sector made up of small and medium-sized farms and prior export experience in agricultural products, obtained during the colonial period; (ii) a dense transportation network, again inherited from the Japanese colonisers; (iii) a literate labour force and a population eager to achieve educational progress, so that an increasing supply of skilled labour became available when it was needed later on in the development process; and (iv) a relative advantage in gaining access to the US and Japanese markets. From the network of SMEs, progressively emerged larger companies which would take on the next stage of industrialisation with more capital- and technology-intensive lines of production and exports.

In twenty years, Taiwan was able to multiply its income per capita by three, thanks to an average growth rate close to 6 per cent. What is remarkable, moreover, is the fact that this fast growth and a drastic structural transformation of the economy could be achieved with apparently no change, and possibly a drop, in the degree of inequality of the distribution of income. In other words, all people saw their living standards grow in the same proportion, and possibly those at the bottom of the distribution more than others.

As a final observation, it should be noted that if Taiwan’s process of development followed the Lewisian pattern of a rapidly growing, self-financing labour-intensive modern sector that absorbed low-productivity workers from agriculture, it differed from it in one major respect. Instead of being confined to the passive role of a provider of cheap (excess) labour to the modern urban sector, the traditional informal sector, here equated to the rural sector, itself actively participated in labour surplus absorption, capital accumulation, and productivity growth. This was the result of an early modernisation of agriculture, both in terms of techniques and crop choices, and of a uniquely successful programme of rural industrialisation (unique, if we except Japan). In short, dualism in Taiwan was reduced rather rapidly both through productivity gains of the lagging sector and the rising employment share of industry.

Another point which deserves attention yet tends to be underplayed in the dual economy literature, is the export orientation of Taiwanese development, without which, most plausibly, the country could not have undergone the drastic structural transformation it experienced. This feature prevented domestic demand, the size of which was limited because of the low income of the population, from forming an obstacle to the exploitation of scale economies and the growth of the manufacturing sector. For this to be possible, the economy had to quickly become internationally competitive in a few lines of products.

Reflecting on the institutional features of the Taiwan of the 1950s, it makes little doubt that its dramatic transformation is first of all the result from a well-thought centrally elaborated strategy, which largely rested on decentralised private incentives and was implemented by an able bureaucracy. In short, at the heart of Taiwan’s success lay a successful combination of central planning and market mechanisms. It appears to owe much to the past history of the Chinese Nationalist leadership that settled in Taiwan after losing the war against the communists on the mainland.

A Bangladesh: Sustainability of Structural Transformation under Threat

In comparison with most developing countries today, Bangladesh may be considered as a success story. Since 2000, its income per capita has been multiplied by a little less than three and poverty has fallen by two thirds. The country has recently graduated from the low-income status in the World Bank classification. Its debt is at a manageable level, and it boasts a rather stable macroeconomic situation over the last three decades.

It is evident from Table 7.1 that, as of 2018 Bangladesh was coming rather close to Taiwan and South Korea in the 1970s: the level of GDP per capita and the sectoral structure of GDP and employment are roughly the same. It also shares with these countries several important historical, geographic, and economic features: a violent nation-building war, an egalitarian land reform, a high population density implying an acute land scarcity, and, today, a powerful labour-intensive Ready-Made Garment (RMG) export growth engine. Yet growth proceeds in Bangladesh at a slower pace than during the take-off of the Asian tigers, and the manufacturing sector is significantly smaller in relative terms.

Structural transformation in Bangladesh has also progressed at an impressive speed. If surplus or low-productivity labour is assumed to be essentially located in the agricultural sector, then Bangladesh would seem to have passed the so-called Lewisian turning point where the absolute number of workers in the agricultural sector starts falling and surplus, or low-productivity labour starts vanishing around the turn of the millennium. As a matter of fact, agricultural employment has been declining by 2 million people over the last three decades, especially during the 2000s. Accounting for demographic growth in rural areas, we can estimate that some 12 million people have left the agricultural sector in the last two decades.

Yet it would be wrong to believe that all these people went to work in the rest of the economy. To assess the structural transformation capacity of the Bangladeshi economy, the role of temporary migration must be brought to the fore. An estimated 10 million Bangladeshis were working abroad in 2019. Judging from the evolution of remittances, the total net outflow of migrant workers may have summed up to 9 million people over the last three decades. This is much more than the observed drop in agricultural employment, so that, absent migration, agricultural employment would have been increasing throughout that period. This remains true even if we assume that, say, half the migrants came from non-agricultural sectors. In short, the growth of the modern sector, driven by manufacturing, does not appear to have been fast enough to absorb the agricultural surplus labour defined as agricultural workers with the lowest productivity.

Of course, migration also contributed to economic growth via worker foreign currency remittances and induced demand effects on the domestic economy. If the overall growth of the Bangladeshi economy has been rather satisfactory over the last three decades, at close to 6 per cent a year and 4.4 per cent per capita, it has partly stemmed from the increasing flow of migrant remittances. It was estimated that remittances contributed to approximately a fourth of GDP per capita growth (see Raihan et al., Reference Raihan, Bourguignon and Salam2023). Yet this observation about the significant role of remittances as a source of national growth raises several issues. Should the sending of a sizeable portion of a country’s population to work abroad be considered as a valid development strategy or as a second-best policy aimed at compensating for the possibly temporary failure of the domestic modern sector to create enough jobs? To what extent is such migration-based development strategy sustainable in the long run? Relatedly, there is a social cost in migration, even when it is temporary, and this should be accounted for in evaluating development.

Even imputing migrants to the agricultural sector, the outflow of workers towards the rest of the economy would still amount to at least 5 million people over the last twenty years. Were all these workers, plus those resulting from the growth of the non-agricultural labour force, absorbed by the modern sector, the RMG export sector in particular? Or did they go to work in informal non-agricultural activities with a labour status and an income level little different from those prevailing in the informal agricultural sector? The answer is provided by the following estimate: 87 per cent of the labour force was informal in 2010 and this share has apparently not changed much since then.Footnote ¹¹ These are high proportions, which might suggest that the structural transformation was less pronounced than it would appear on the basis of a simple agricultural/non-agricultural dichotomy.

A large part of the urban informal sector may be considered as complementary to the modern sector of the economy. However, its mode of operation follows a different pattern with self-employment or micro-firms as the dominant type of production organisation. Somehow, that part of the urban informal sector may be thought as a subsector of the modern part of the economy, whose informality is mostly motivated by the possibility it offers of evading taxes and labour regulation. It is difficult to say which portion of the informal urban sector must be thus assimilated to the modern sector. There is, indeed, some ambiguity about how to measure structural transformation, or, equivalently, about how the type of jobs and the levels of earnings within the non-agricultural sector should be accounted for. In any case, this does not lessen the transformative importance of the huge shift of labour away from agriculture that took place in Bangladesh during the last decades.

As in South Korea and Taiwan at the time of their take-off, the main growth engine behind the structural transformation in Bangladesh’s economy over the last three decades has been the labour-intensive manufacturing export sector, mostly ready-made garments (RMG). It grew at an annual rate of 11 per cent since 1990 and created a little more than 2 two million jobs, close to 10 per cent of the whole increment in the labour force. Bangladesh is now the second global RMG exporter after China. Directly or indirectly, through backward and forward linkages as well as foreign currency receipts, the RMG sector contributed in a major way to the growth of GDP and living standards. Its overall contribution to GDP growth has been estimated to be as high as 40 per cent (by Raihan et al., Reference Raihan, Bourguignon and Salam2023). Its transformative impact, most notably on and through female employment, has also been substantial. Yet, if it had not been for outmigration, this would not have been enough to absorb the surplus labour present in the agricultural sector. To the extent that there is uncertainty about future migration opportunities (India’s present political regime is hostile to Muslim migrants, which hurts Bangladeshi migration), sustaining the structural transformation at its current pace, may thus prove difficult.

There is also some uncertainty about the future development of the RMG sector. Technological change seems likely to drastically reduce its relative labour intensity and the comparative advantage it draws from particularly cheap and repressed labour, whereas exports will soon lose their Least Developed Countries preferential trade status in advanced countries because of the recent graduation of Bangladesh to (lower) middle-income country status. As a matter of fact, such a slowdown can already be observed in the volume of exports – since the 2008 crisis and particularly the 2013 Rana Plaza accident where 1,100 workers died when their factory collapsed. The slowdown in employment growth is even more pronounced.

If the Bangladesh economy seems to share many features and follow the same path as the Southeast Asian newly industrialised countries at the time of their take-off, there is a risk that its growth rate decelerates and even that its structural transformation dynamic gets jammed. At the same stage of development in the benchmark countries, the manufacturing sector was larger and was growing faster. In short, the growth engine was more powerful. Moreover, it was gaining more power still through diversification, both within their initial area of excellence (RMG and other labour-intensive exports) and without. On the contrary, manufacturing exports in Bangladesh tend to concentrate everyday more on RMG products, and within RMG, on a limited set of product lines.

This issue of the diversification of labour-intensive manufacturing exports is the main challenge that Bangladesh will soon face in trying to sustain its rates of growth. This has been recognised, now for quite some time, by observers and policymakers. On several occasions and in several official documents, the government has committed to adopting such a strategy. But no tangible results were delivered yet, even though the continuation and the needed acceleration of the structural transformation depends on this diversification. From the point of view of the relationship between development and institutions prioritised in this volume, the question is: what is the institutional cause of this apparent blockage of measures that would benefit the national community?

Various factors can be mentioned, including the endemic lack of a clear development strategy and implementing capacity, a culture of business-government informal ‘deals’, a notoriously corrupt public and private financial system, and very limited public resources due to an exceptionally low average tax rate. In that context, a major impediment to a strategy of diversification seems to be the size of the RMG sector and its critical importance, up to now, in the overall development of the country. As a result, the leverage that it can bring to bear on the government is particularly strong, leading to the pre-emption of the development of other sectors of the economy and the monopolisation of public support and credit. On the other hand, the lack of diversification within the RMG sector itself seems to result from a strong specialisation in those product lines that make most use of exceptionally low labour costs.

In summary, there is a risk that, even though it has been rather effective over the last two or three decades, the growth engine that feeds the structural transformation of the Bangladeshi economy will slow down in the close future. There are some signs that this has already started, especially with respect to job creation. Remedying that situation would require the RMG to increase its global market share by expanding the scope of its activity or supplementing it by other lines of labour-intensive manufacturing exports. The first option would require a substantial improvement of the competitiveness of the RMG sector, which has relied until now mostly on the low cost of labour and poor working conditions imposed through the co-option of trade unions. Enhanced RMG competitiveness or diversification of the export manufacturing sector also requires progress to be made in various areas: (i) better production infrastructure in a country where land is particularly scarce – a priority that is acknowledged by the present government; (ii) a more educated and skilled labour force, not only in quantity but also in quality terms, which requires significant progress on the latter; (iii) the laying down of clear and well-thought-out development policies, and the setting of an effective bureaucracy apparatus to implement them, rather than the reliance on informal deals that essentially favour dominant economic actors; and (iv) an efficient and non-corrupt financial system. It is not clear, at this stage, that all these requirements for the pursuit of an autonomous and fast structural transformation of the economy and the society, possibly one that is less reliant on migration, will soon be met.

B Tanzania: An Uncertain Growth Engine

After a difficult transition from a socialist development experiment to a market economy, growth has proceeded at a rather satisfactory rate of 6 per cent annually over the last two or three last decades in Tanzania. But population growth has curtailed that rate by a little less than half when considering GDP per capita. Since the turn of the millennium, living standards have approximately doubled and poverty has fallen, although at a slower pace lately. The structure of the economy has also changed with the GDP-share of agriculture falling in favour of services and, to a lesser extent manufacturing. The share of agriculture in employment has fallen too, but it is still high at 70 per cent. Somewhat surprisingly, however, labour was reallocated primarily towards the construction sector, retail trade and hospitality services with a noticeable fall in labour productivity in the latter two sectors.

Such an evolution is hardly consistent with a powerful engine of growth moving low-productivity agricultural workers to higher-productivity jobs in the rest of the economy. It resembles more a process of demand-driven growth where income gains, partly stemming from the expansion of mining (gold) and favourable changes in the terms of trade, are spent on domestic production, including construction investments and services.

Although limited, the relative increase in the output and employment shares of the manufacturing sector sends a positive signal. That it has come with a significant expansion of labour-intensive manufacturing exports is especially encouraging. But it is still too slow, and the sector is too small to have a major impact overall. There was also some promising progress in tourism before the COVID-19 pandemic struck.

The fall in the employment share of the agricultural sector conceals a limited reallocation of labour to the rest of the economy and, because of population growth, an absolute increase in the size of the agricultural sector. There were 10 million workers in the agricultural sector in 2000 – 82 per cent of the whole labour force. Demographic growth would have raised this figure to 17 million by 2018, but because of net migration to other sectors or to foreign countries, they were only 15 million. Discounting foreign outmigration, the absorption capacity of the non-agricultural part of the economy was thus quite limited. Somehow, the structural transformation worked backwards since population growth, at the rate of 2.8 per cent annually, overcame job creation in the dynamic part of the economy. Things would even look worse if part of the observed increase in the employment share of urban informal sectors were considered as participating in the accumulation of surplus labour in the non-agricultural part of the economy.

The preceding argument needs to be seriously qualified, though. As already pointed out, the Lewisian model presupposes the full utilisation of land in the context of densely populated countries, and it is in this specific context that the concept of an unlimited labour supply in the rest of the economy makes sense. In most African countries, including Tanzania, Benin, and Mozambique in the present project, this condition does not seem to be satisfied, so that the rural population may grow without labour productivity falling. If it did so, the low absorption capacity of the non-agricultural sector would not be a problem, and the economy’s structural transformation could rest not only on the growth of the non-agricultural sector but also on an extension of the agricultural sector and on agro-industrial development.

It turns out that the Tanzanian agricultural labour productivity has been increasing during the period under analysis without it being possible to distinguish autonomous gains in yields per hectare and changes in the cultivated area. In any case, the average labour productivity in agriculture remained much lower than in the rest of the economy – the gap even slightly increased – which means that the structural transformation argument above remains correct in the sense that absorbing agricultural labour contributes to growth and to the reduction of agricultural low-productivity pockets. From that point of view, it remains the case that the absorption capacity of the non-agricultural sector was limited, and Tanzanian development has been little transformative. On the other hand, this discussion, and the evidence on productivity in agriculture suggest that this sector holds development opportunities that may presently be underexploited.

The situation of Tanzania today clearly differs from that of Taiwan or South Korea when those countries were at the same level of real income per capita – see Table 7.1. If the GDP share of agriculture is of the same order, the employment share is much higher in Tanzania, whereas the share of manufacturing in GDP and in employment are substantially smaller. These contrasts imply that differences in labour productivity between agriculture and other sectors are more pronounced in Tanzania, this being particularly true for the industrial or manufacturing sectors. This suggests that the structural transformation was already more advanced among the Asian tigers when they were at the same level of GDP per capita as Tanzania today. Of course, this advantage essentially reflects their higher level of industrialisation and the various circumstances and conditions that made it possible, including the accumulation of physical and human capital or their institutional setting.

The above type of transhistorical comparative exercise should obviously be interpreted with caution. There is no reason to expect history to repeat itself across countries or regions and it would be naïve to believe that the only pathway to development and structural transformation is the one followed some fifty years ago by the two East Asian countries. This being acknowledged, the above comparison provides diagnostic insights that can be helpful in gauging, almost mechanically, the development potential of Tanzania, and in highlighting the consequences of a missing engine of long-run growth.

Even with continuing favourable terms of trade, and possibly with the benefit of rents accruing from presently untapped reserves of natural gas, Tanzania’s known natural resources are not sufficient to ensure the future prosperity of the country. Moreover, the labour content of such a strategy would be limited. To be effective and sustainable, especially in view of the fast population growth expected for still a few decades ahead, the structural transformation must additionally rely on a solid growth engine based upon the production of labour-intensive goods and services, the demand of which is not constrained by the size of the domestic economy. Obvious candidates are labour-intensive manufacturing exports, agroindustry, and tourism, the two latter corresponding to clear comparative advantages of Tanzania. As a matter of fact, the last three administrations committed to pursue a development strategy based upon industrialisation and export diversification. Yet, as of the late 2010s, results have been quite modest.

What can explain the limited success of this industrialisation strategy? Many factors should be mentioned. The most important one seems to be the difficulty of disciplining business. The big business sector is highly monopolistic, and, because of the fractionalisation of power within the dominant party, it had, until recently, a powerful leverage on state decisions that would go against their short- or medium-run interest. Things may have changed with the Magufuli administration in the late 2010s but, even then, the relationship between state and business was a difficult one. In theory, appropriate incentives, duly conditioned on results should permit to align business interests with the government’s strategy. Such incentivising policies may infringe international WTO trade rules, which did not exist at the time of the East Asian industrialisation. Yet the subsidisation of credit at the firm level, duty exoneration on inputs, or the provision of critical infrastructure, are perfectly legal, and may help Tanzanian firms to become competitive on foreign markets where they are not present. However, managing such incentives and their conditionality on results requires a skilled and uncorrupted administrative apparatus, which may not be up to the task in Tanzania.

Other factors that hinder the diversification of exports include the limited public resources arising from a low overall level of taxation, the slow accumulation of soft capital – progress has been made in enrolling nearly all children of schooling age, but learning outcomes remain disappointing – and the slow rate of infrastructure building – until recently, Tanzania was a laggard in energy production and distribution, ports, roads, and rail transport facilities. With respect to the agroindustry, the complexity and ineffectiveness of the law governing land user rights is also often cited as a major impediment to commercial production and exports.

Moreover, excessive financial dependency on foreign countries or organisations may be considered as a source of uncertainty for future development. Foreign assistance has continued to account for 5–8 per cent of GDP since the late 2000s. One can therefore worry about what would happen to public investments if the volume of aid were to fall in line with the recent announcements made by several donors.

In summary, Tanzania’s growth performance since the turn of the new millennium provides reasons to celebrate, although optimism must be tempered. Tanzania is still in the middle of the dualistic stage of development and the problem is whether it has the potential to reach in the foreseeable future the next stage of the structural transformation where the cross-sectoral labour productivity gap starts narrowing down. The evidence suggests that it lacks a clearly identifiable engine of sustainable long-run growth, and, more worryingly, an institutional setup appropriate to develop such an engine and meet the challenges ahead. More will be said about these institutional aspects in Chapters 8 and 9.

C Benin: Informal Growth as a Delusory Development Strategy

Benin’s development over the last few decades has been characterised by modest growth performance and a rather atypical sectoral structure of employment. Although some acceleration has been observed during the last five years, the average annual GDP growth rate has been slightly below 5 per cent since 2000. With an almost 3 per cent rate of growth of the population, income per capita has grown rather slowly, and in any case at a slower pace than in the rest of the continent. Concerning the sectoral structure of the economy, it can be seen in Table 7.1 that the share of agriculture is of the same order as in South Korea and Taiwan when those economies were at the same level of income, but also that the share of the manufacturing sector is well below that of these Asian tigers. Benin’s sectoral structure of GDP is close to that of Tanzania and, as a matter of fact, that of most sub-Saharan low-income or lower-middle-income countries. Where Benin is atypical relative to both sub-Saharan countries and the Asian comparator countries, however, is in the structure of employment. It exhibits a substantially smaller proportion of the labour force employed in agriculture – and therefore a higher proportion in other sectors – and a somewhat higher average labour productivity in that sector relative to the whole economy. The latter is partly the consequence of the importance of cotton production and exports in the Beninese economy, even though productivity in this activity has gone through sharp cycles since the turn of the century.

The modest rate of growth of the Beninese economy has its roots in an investment rate which has been below 20 per cent over the 1995–2017 period, except in the very last years of that period. As a matter of fact, the overall growth of GDP owes more to the movement of labour out of low- productivity agriculture than to sectoral productivity gains, except perhaps in agriculture, which benefits from a rather weak population pressure on available land. In effect, productivity has gone down in all non-agricultural sectors due to population growth, limited investment, and the inflow of labour coming from agriculture.

Despite the slow rate of growth, the structural transformation of the Beninese labour force has been substantial, but in a direction and according to patterns that are quite peculiar. In the decade from 2006 to 2015, it is estimated that 30 per cent of the agricultural labour force went to work in other sectors. As population growth was slightly higher, the total volume of agricultural employment went slightly down. What makes this restructuring so particular, however, is that, instead of going to work predominantly in the modern part of the economy, most workers went to the commerce and service sectors, which are the lowest productivity sectors outside agriculture. In commerce, new entries caused such a significant decrease in the mean income that it became hardly higher than in the agricultural sector.

There is ground to believe that the transfer of informal employment from agriculture to the rest of the economy has been caused by the largely informal cross-border trade (ICBT) with Nigeria rather than by the growth of the informal non-agricultural sector in tandem with the development of the formal sector (as observed elsewhere). This is a major specificity of the Beninese economy and the consequence, as well as a possible cause, of slow formal development. ICBT was first encouraged by differences in tariff and non-tariff barriers between Nigeria and Benin, which, as a member of the West African Economic and Monetary Union (WAEMU), applies the Union’s rules. These differences create arbitrage opportunities for products legally imported into Benin and re-exported to Nigeria, where they enter illegally, and vice-versa for certain products smuggled illegally from Nigeria, where they are cheaper, into Benin. Such informal, and mostly illegal, trade activity is estimated to account for a little more than 10 per cent of GDP, only slightly less than official cotton exports, and to employ directly at least 2 per cent of the labour force, but much more indirectly. Both figures are thought to have sizeably increased over the last ten years.

By itself and through its upward and downward linkages, ICBT has huge effects on the Beninese economy and society. First, it contributes to increasing the incidence of informality and to nurturing a culture of corruption. Informality follows from the illegal nature of the activity, while corruption is used to buy the complicity of state executives and bureaucrats at various levels of the administration (including customs officers), and to obtain credit facilities from banks. This is particularly true for the large-scale smuggling of gas and other materials from Nigeria. Second, ICBT displaces some formal activities and diverts entrepreneurs from potentially more socially profitable lines of formal business. Several smuggled products outcompete domestic producers, most notably in gas distribution and cement production. Third, the failure of the state to curb this illegal activity entails a loss of intervention capacity in other areas indirectly affected by the cross-border trade. For instance, incentives to develop other activities are rendered inoperative, despite the presence of a dynamic entrepreneurial class. Fourth, at the macro level, illegal trade with Nigeria makes Benin dependent on the former’s oil revenues and subject to oil price fluctuations in international markets. The macroeconomic shortcomings of this dependence are well known, especially for a country, which being a member of the WAEMU has adopted a fixed exchange rate system.

Cross-border trade with Nigeria is a revenue-generating opportunity and it is natural that some entrepreneurs have been eager to seize it. Yet its overall contribution to development may end up being negative, because of the informality and the culture of corruption that it has brought about, the uncontrolled smuggling that it has triggered, its unsustainability, and the marked dependence that it has created vis-à-vis Nigeria’s trade policy. In effect, the ICBT activity has very much expanded between 2005 and 2014, due to the high price of oil. Yet GDP has not grown much faster during that period and poverty has changed little.

The fact that Nigeria recently decided to close its border with Benin, and has effectively stuck to that decision, is a sore reminder of the high economic dependency of Benin on its giant oil-exporting neighbour.

As mentioned earlier, cotton is the main formal activity in Benin, representing 12 per cent of GDP, and providing most of the country’s formal export revenues. The organisation of the whole sector, and the respective roles of the private and public sectors, have gone through several changes over time, with direct effects on production and exports. Except for farming, the whole chain of production is structured as a monopoly and has been very much under the control of a single business group, headed by Patrice Talon, an entrepreneur who was elected president of the Republic several years ago and has just been re-elected. The monopolistic structure of the cotton sector necessarily entails significant efficiency costs. It is fair to recognise, however, that the sector has done rather well since its organisation has been stabilised, and this despite the monopolistic organisation of input provision, ginning, and commercialisation.

Since cotton exports or cross-border trade can hardly be conceived as powerful and sustainable vectors of development, it must be acknowledged that there is no engine of growth nor any real structural transformation process presently at work in the Beninese economy. For a while, it may have been the case that one could earn more by selling bottles of smuggled gas on roadsides than by working in the family farm, but this is not what development is about and such a restructuring is essentially unstable. More seriously, it is not clear that an effective growth engine able to trigger genuine structural change is about to be developed.

Benin has clear comparative advantages in agriculture, not least because there is plentiful land available, particularly in the northern part of the country, which has been largely neglected by successive governments in Cotonou. Developing agro-industrial exports is a real possibility as some encouraging starts are attesting. However, to push these further and to create an impetus that can spread to other areas and other lines of products requires a better provision of public goods, including a competent and non-corrupt bureaucracy, better infrastructure, more investments in quality schooling, and, most importantly, a clear and consistent development strategy.

The economic diagnosis about the stalled structural transformation – and about the kind of wrong-headed change that has taken place – is evident. The nature of the development strategy that should be pursued, whether on the agricultural side, by looking for complements to cotton, or possibly on the manufacturing side, by substituting domestic production for smuggled goods, is equally clear. The issue is why no such strategy had been implemented.

The answer must be found in the governance of the country, at least until Talon became president. There were then two major sources of rents, cotton exports and ICBT with Nigeria, and a rent-sharing agreement had been reached between the oligarchs who controlled these two sectors, including Talon, and those in power. Here is a perfect case of state capture. The equilibrium between the main players of that rent-sharing game got disrupted at some point, which led the oligarchs to compete for political power. One of them won. Precisely because political and economic power are now in the same hands, the nature of the equilibrium has changed, and new development strategies may emerge.

Although controversial, the present administration seems to be making progress in that direction. But there is still a long way to go before a genuine structural transformation and a definitive dent on poverty can take place in the country.

D Mozambique: Natural Resource Curse or Structural Transformation?

In comparison with other case study countries and a fortiori with South Korea and Taiwan, the combination of development advantages and shortcomings in Mozambique is quite specific. First, both its geography and ethnic composition are extremely fragmented. The country extends over 2,300 km from north to south and its population includes ten main ethnic groups, rather clearly differentiated by geographic region. From the latter point of view, Mozambique is comparable to Benin, except that the groups are physically more distant from each other, their isolation being amplified by a limited development of transport infrastructure. Second, the country obtained its independence much later than other African countries – as a matter of fact, not long after Bangladesh in 1975 – but it fell quickly into a long civil conflict which paralysed economic development for fifteen years. Third, similar to other case study countries in the Institutional Diagnostic Project (IDP) project, Mozambique adopted a socialist approach to development at independence, which delivered poor results, especially in the context of the domestic conflict. A transition towards a market economic system was made under the supervision of the international financial institutions, yet it proved more difficult and painful than elsewhere because it was launched at a time when the conflict was still ongoing.

Despite these hindrances, and largely thanks to an unusually high level of foreign development assistance, the Mozambican economy was able to grow at a fast rate until a few years ago. Growth was first triggered by the recovery from the civil war, and it then proceeded via more standard economic mechanisms. GDP per capita has thus grown at a little more than 4 per cent annually since the turn of the new millennium, and until 2016 when a major economic crisis struck for reasons which are detailed below.

Focusing on the last two decades, when the economy and the population have fully recovered from the civil conflict, two very different structural growth regimes were observed. Both the sectoral structures of GDP and employment varied little during the 2000s, except for some progress of the manufacturing sector. Overall growth originated in labour productivity gains across the board, including in agriculture. Things then changed radically during the 2010s: aggregate growth resulted essentially from a major sectoral restructuring away from agriculture, both in terms of GDP and employment, and in favour of non-agricultural sectors, except manufacturing.

This structural transformation has indeed been significant. During the six-year period from 2009 to 2015, just before the recent crisis, 1.5 million workers left agriculture to go to work in the rest of the economy. They represent 16 per cent of what would have been the agricultural labour force at the end of the period. This outmigration from agriculture has been so large that it overcame population growth and total agricultural employment fell. Yet the difference with the Southeast Asian countries and Bangladesh, and to a lesser extent Tanzania, is that the main sector of destination for those workers leaving agriculture was not manufacturing but private services, a sector where informal low-productivity jobs coexist with formal high- or median-productivity jobs. The issue then arises of the type of job taken up by agricultural workers, whether belonging to the former or to the latter category.

That labour productivity fell drastically in the private service sector suggests that employment increased for both types of job, probably more in the subsector based on informal low-productivity jobs. Yet the average labour productivity in the private service sector remained much above that in agriculture, so that the structural transformation mechanism kept working even though with some uncertainty about the exact nature of the process.

The manufacturing sector that came to represent 14 per cent of GDP in the 2000s did not contribute to the absorption of low-productivity labour in agriculture. This is not surprising given its lines of production. A major part of its initial growth has actually come from the production of aluminium made possible by the availability of cheap hydroelectric power on the Zambezi River. But this activity, very much akin to the exportation of natural resources, employs only a small number of workers and has limited linkages with the rest of the economy. Moreover, its capacity stopped expanding around 2008, and since few other lines of manufacturing activity have developed afterwards, the GDP share of the manufacturing sector has fallen continuously since then. Over the last decade, the dynamic part of the Mozambican economy has been the extraction of coal and natural gas. The latter is expected to expand drastically in the future when the huge reserves discovered in 2010 will enter into full exploitation. Together with aluminium and electricity sales to neighbouring countries, coal and natural gas represent today some 70 per cent of total exports and their share of GDP may be estimated at around 12 per cent.

It is thus fair to say that Mozambique has become an exporter of natural resources and has tended to live on the related rent over the last few years. This explains why slow progress has lately been made in the production of tradeable goods, since the largest part of domestic growth has been accommodating the rent-based increase in the aggregate demand for non-tradeables, including private services. Quite telling in this respect is the fact that both agricultural and manufacturing output per capita have stagnated since 2010, and even somewhat earlier for manufacturing.

A possible reason for the lack of dynamism of the manufacturing sector is the absence of an entrepreneurial class in Mozambique. After independence, a period characterised by central planning, bureaucrats came to oversee the production apparatus, as Portuguese entrepreneurs had left the country. When the transition to a market economy took place a little before the end of the civil conflict, production units were privatised in favour of political personnel with little or no business experience and relying more on political connections and rent creation than commercial flair. Now that the country can live on the rent arising from natural resources, incentives for the appearance of an ambitious class of industrial entrepreneurs are weak and might become even weaker in the future. The demand arising from the rent will be mostly addressed to domestically oriented sectors such as services and construction and, presumably, it will mostly benefit the urban part of the country, and Maputo, the capital city, in particular.

The prospect of huge rents related to the future exploitation of natural gas has also exacerbated the appetite of rent-seekers and revealed the extent of corruption in Mozambique, at the same time as the ineffectiveness of the state apparatus to control it. A major scandal struck in 2016 about a US$2 billion embezzlement involving senior officials. It led donors to cancel foreign assistance payments, which plunged Mozambique into a deep financial and economic crisis. Yet the worst damage was created by the surging awareness of the pervasiveness of rent-seeking and corrupt practices, and their prevalence over entrepreneurship and bureaucratic effectiveness.Footnote ¹²

Finally, we need to turn our attention to the agricultural sector and the issue of poverty, which affects more than 60 per cent of the population and is concentrated in rural areas. Given that the present and future engine of growth lies in natural resources whose exploitation is based on capital-intensive techniques, there is little hope for a quick absorption of low-productivity agricultural labour by the exporting sector, and for rapid advances on the poverty front. Can we witness a repetition of the structural transformation scenario that was observed in the early 2010s when a huge outflow of agricultural workers towards the service sector was observed? There is great uncertainty in this respect. Moreover, it was seen above that there was some ambiguity about the productivity of the jobs created in the service sector. Given the present size of the agricultural labour force, its low level of incomes and the absence of an autonomous labour-intensive growth engine outside agriculture, any reasonable development strategy must include an agricultural component.

An important challenge that Mozambique will face in the future thus lies in its capacity to use its forthcoming rents from gas exports to boost the traditional agricultural sector. This alternative to industrial development as the main engine of structural transformation, which befits land-abundant countries, must find its way into the minds of Mozambican policymakers if they want to make a real dent in poverty and establish a broader base for the country’s development. However, agricultural or rural development in Mozambique is made more difficult than in most other countries by the geographic stretch and the ethnic fragmentation of the country. Combined with largely insufficient and inefficient transport infrastructure, both factors reduce labour mobility and limit the gains that could be obtained from inter-regional trade. Agricultural productivity increased at the same pace as the rest of the economy in the first decade of this century, when the recovery from civil war times was most likely completed. Its pace slowed down since then. Efforts should be made to revive this earlier period, possibly accelerating productivity gains while avoiding Dutch disease phenomena, which will unavoidably manifest themselves as natural resource exports increase.

In summary, the key development issue in Mozambique is whether existing institutions and the structure of political power will allow a structural transformation of the country that will simultaneously absorb part of rural labour, increase agricultural productivity, and expand local markets through a deeper physical integration of the country. The recent evolution of the economy and the reappearance of social and political tensions, including recent terrorist attacks in the coastal area facing the offshore gas fields, are worrying in this regard.