12 Chapter 10 — Growth, Development, and Structural Transformation
From Solow to Lewis: what economies must do to converge
“The consequences for human welfare involved in questions like these are simply staggering: once one starts to think about them, it is hard to think about anything else.” — Robert Lucas (1988)
“The process of Creative Destruction is the essential fact about capitalism.” — Joseph Schumpeter, Capitalism, Socialism and Democracy (1942)
The Map
The preceding chapters analysed macroeconomic stabilisation in the short run: the business cycle, unemployment, fiscal and monetary policy. This chapter changes horizon entirely. The question is not how to prevent the economy from falling below its potential but how to raise that potential itself over decades. Why are some countries today 50 times richer than they were in 1800, while others have barely doubled their living standards? Why have South Korea and Taiwan transformed from poor agrarian societies to industrial powerhouses in a single generation, while economies starting from comparable points in the 1960s have stagnated? What are the mechanisms through which structural transformation happens, and what can policy do to accelerate or distort them?
The chapter advances in seven movements, each building on the previous. First, the standard vocabulary of growth accounting and the Solow framework. Second, endogenous growth theory and what it contributes beyond Solow. Third, convergence, clubs, and the middle-income trap. Fourth, the dual economy model, premature deindustrialisation, and why the industrial pathway remains central. Fifth, structural transformation mechanics: the shift-share decomposition and the product space. Sixth, the theory of industrial policy: four conditions, four roles of the state, and the growth diagnostics method. Seventh, Morocco’s position: where it stands in the product space, what the binding constraints appear to be, and what trajectory the evidence suggests.
13 The Great Divergence: Why Now and Why Some
Before analysing the mechanisms of growth, one must grasp the historical scale of the problem. In 1776, when Adam Smith described his pin factory, a French peasant and an Indian peasant and a Chinese peasant lived at broadly comparable material standards. Nutrition, shelter, life expectancy, and material comfort were similar across civilisations: different in detail, convergent in level. Two centuries later, the average resident of a rich country lives at a standard 40-50 times higher than the average resident of a poor country. This is not a difference of degree but a difference of kind.
Angus Maddison spent decades reconstructing two millennia of economic data. His central finding is that this divergence is recent and concentrated: before the Industrial Revolution, the gap in income per capita between the richest and poorest regions of the world was approximately three to four times. Today it is forty to fifty times. What Simon Kuznets called modern economic growth (sustained and accelerating per capita income growth, associated with urbanisation and industrialisation) is not a universal spontaneous phenomenon: it is a historically constructed transformation, unequally distributed and politically contested.
The question that frames this entire chapter is therefore not why some countries are rich — we have a fairly good understanding of that — but why the transformation generalised in some places and not others, and what this implies for countries that have not yet completed it. Before 1800, the world was relatively uniform in poverty. After 1800, it bifurcated into two trajectories. Understanding the bifurcation mechanism is the precondition for understanding whether and how late-developing countries can cross to the higher trajectory.
The Maddison evidence also provides the long-run context for Morocco. In 1900, under the waning Alaouite sultanate and encroaching European influence, Morocco’s GDP per capita was comparable to many other economies in the Mediterranean and Middle East. Colonial integration into the French economic sphere (1912-1956) did not accelerate Moroccan convergence toward European living standards; it organised Moroccan territory for French agricultural and mineral extraction while limiting the industrial development that would have built productive capability. The post-independence trajectory has been real growth (Morocco’s income per capita today is multiple times its 1956 level), but the inherited productive structure, shaped by colonial extraction logic, created the starting conditions for the development challenges this chapter analyses.
14 Part I — The Grammar of Growth: Capital, Labour, Technology
14.1 The production function and its three sources
The standard framework begins from a deceptively simple representation. An economy’s output \(Y\) depends on the factors it mobilises and how efficiently it combines them:
\[Y = A \cdot F(K, L)\]
Capital \(K\) encompasses machines, infrastructure, equipment, and software: the physical tools that amplify what labour can accomplish. Labour \(L\) is the working population and its hours. \(A\) is the total factor productivity (TFP): the efficiency with which all inputs are combined. A worker with a spade and a worker with a tractor have radically different productive capacities even working the same hours. Two factories with identical machines achieve different results if one is well-organised and the other is not. \(A\) captures all of that invisible productive efficiency.
The Cobb-Douglas functional form, widely used in empirical work:
\[Y = A \cdot K^\alpha \cdot L^{1-\alpha}\]
decomposes output into factor contributions weighted by their income shares. The parameter \(\alpha\) represents capital’s share of national income (empirically around 0.3-0.35 in most economies), and \(1-\alpha\) represents labour’s share. The growth accounting methodology uses this framework to decompose observed growth into its sources:
\[\hat{Y} = \hat{A} + \alpha \hat{K} + (1-\alpha) \hat{L}\]
where hats denote growth rates. If capital grows at 4% and labour at 2%, their combined contribution is \(0.35 \times 4 + 0.65 \times 2 = 2.7\) percentage points. If the economy grows at 5%, the Solow residual, TFP growth, is 2.3 percentage points.
Solow’s own calculation (1957) found that approximately 87.5% of US growth per capita between 1909 and 1949 was attributable to technical change, with only 12.5% to capital accumulation. This number shocked everyone, including Solow. It says something fundamental: the wealth of nations comes primarily not from tools but from what we know how to do with tools. An economy with the same machines but better management, better labour organisation, and better resource allocation will produce more without any additional investment.
The residual has been called, with gentle irony, “the measure of our ignorance.” What exactly does it contain? Genuine technological innovation; quality improvement of capital unmeasured in statistics; better allocation of resources between sectors; institutional improvement that releases productivity; reduced corruption; and measurement error. This ambiguity is analytically inconvenient but intellectually valuable: it forces the search for what hides behind the residual. That search produced the endogenous growth theories of the 1980s-1990s.
14.2 Endogenous growth: making the residual explicable
The Solow model takes TFP growth as exogenous, falling from the sky at a fixed rate, unexplained by the model itself. Paul Romer (Nobel Prize 2018), Robert Lucas, Philippe Aghion, and Peter Howitt developed theories that endogenise the residual, making innovation, human capital accumulation, and knowledge spillovers the explicit results of economic choices rather than unexplained manna.
Romer’s model (1990) is built on a crucial insight about the economics of ideas: ideas are non-rival goods. A piece of knowledge, unlike a machine, can be used by an unlimited number of people simultaneously without being diminished. When OCP’s research unit develops a customised fertiliser formulation for Sahelian soils, that knowledge can be applied to Saharan soils, Sahelian soils in different countries, and further developed without the original formula being depleted. This non-rivalry means that ideas, unlike physical capital, do not face diminishing returns. Once created, they can be infinitely replicated at near-zero marginal cost.
This creates the possibility of sustained long-run growth without diminishing returns: each new idea opens possibilities for further ideas, and the stock of knowledge grows without bound. The policy implication is striking: anything that encourages idea-generation, intellectual property rights, R&D subsidies, open science infrastructure, university-industry linkages, can raise the long-run growth rate, not just the level of output.
Lucas’s model (1988) makes human capital accumulation the engine of growth. Workers embedded in dense urban environments with many other skilled workers are more productive not just because of their individual skills but because of human capital externalities, specifically the knowledge spillovers from learning, imitating, and collaborating with skilled peers. This is why a skilled worker in Casablanca earns more than an equally qualified worker in a rural area: not (entirely) because of employer preferences but because the density of interaction generates genuine productivity gains. The implication: urbanisation and educational investment are complementary engines of long-run growth, not alternatives.
Aghion and Howitt’s Schumpeterian model (1992) formalises Schumpeter’s insight that innovation proceeds through creative destruction: new products and processes displace old ones, creating growth while simultaneously destroying the rents of incumbent producers. The model generates the prediction that competition and growth are positively related (the threat of displacement by new entrants stimulates innovation effort) up to a point, but very intense competition that eliminates all rents may discourage the long-horizon investment that radical innovation requires. This non-monotonic relationship between competition and innovation explains why competition policy in innovation-intensive sectors is more complex than in standard product markets.
For Morocco, the endogenous growth perspective adds a crucial dimension to the standard capital accumulation story. Morocco’s infrastructure investment, roads, ports, industrial zones, is necessary but not sufficient. What generates long-run growth is the expansion of the knowledge base: the number of engineers who can design industrial processes rather than just operate them, the density of inter-firm linkages that diffuse technical knowledge, the quality of university-industry connections that translate research into productive capability. These are the investments that shift Morocco’s production frontier rather than simply moving closer to an existing frontier.
14.3 Institutions and growth: Acemoglu and Robinson’s inclusive-extractive distinction
If endogenous growth theory explains how knowledge and human capital generate sustained growth, it does not fully explain why some countries invest in knowledge while others do not. Daron Acemoglu and James Robinson’s framework (Why Nations Fail, 2012) fills this gap with a political economy argument: growth is determined by institutions, and institutions are determined by the political settlement between those who gain from extracting wealth from others and those who would benefit from productive wealth creation.
Their central distinction is between inclusive institutions, which distribute power and economic opportunity broadly, protect property rights for all, enforce contracts impartially, and allow creative destruction to displace incumbents, and extractive institutions, which concentrate power and wealth in a narrow elite, protect incumbents from competition, and suppress the creative destruction that threatens established interests. Inclusive institutions generate growth; extractive institutions generate rent extraction that may produce short-run prosperity for the elite but systematically fails to generate broad-based long-run development.
The historical evidence is stark. The divergence between North and South Korea after the division of 1945, two populations with identical culture, language, religion, and history, is almost entirely explained by institutional difference. Under inclusive institutions, South Korea created one of the most rapid catch-up trajectories in history. Under extractive institutions, North Korea generated stagnation and periodic famine. The contrast between Nogales, Arizona and Nogales, Sonora, same geography, same culture, same history, different side of the US-Mexico border, radically different living standards, makes the same point at smaller scale: what differs is the institutional environment on each side of the line.
For Morocco, the Acemoglu-Robinson lens illuminates several structural features. The colonial period created extractive institutions par excellence: a dual economy with one set of rules for European settlers and another for the Moroccan population, concentration of modern productive assets in colonial hands, suppression of Moroccan entrepreneurship in sectors competing with French interests, and a legal structure designed to facilitate resource transfer to the metropole rather than Moroccan productive development. Independence did not automatically generate inclusive institutions: it transferred institutional control to a Moroccan elite that maintained many of the concentrated structures of the colonial economy while adding a layer of nationalist political economy.
The contemporary challenge is captured in Mohamed Saadi’s analysis of Moroccan crony capitalism: a system where political connection remains a significant determinant of business success, where regulatory capture protects incumbents in concentrated sectors, and where the barrier between political power and economic advantage is insufficiently thick. This is not the full extractive-institution pathology of Acemoglu and Robinson’s worst cases, Morocco has competitive elections, a functioning judiciary, and a broadly respected rule of commercial law. But the distance from fully inclusive institutions is real and measurable in the gap between Morocco’s ECI and what its income level would predict: extractive elements in the institutional structure suppress the productive experimentation and creative destruction that inclusive institutions would generate.
14.4 The Harrod-Domar model: the primacy of savings
Before Solow, Harrod and Domar’s model shaped two decades of development planning with its simple but consequential prediction: the growth rate equals the savings rate divided by the capital-output ratio (\(g = s/v\)). The implication drove development policy for decades: if developing countries could not generate sufficient domestic savings, they needed foreign capital to fill the “savings gap” and achieve a target growth rate.
This framework underlay the rationale for foreign aid and structural adjustment lending: if the binding constraint is savings, injecting capital should produce predictable growth. The post-WWII experience revealed the model’s limitation: capital was often not the binding constraint. Economies with high capital inflows grew slowly if institutions were weak, technical capabilities absent, or complementary infrastructure lacking. The model was useful for identifying one necessary input but failed to identify which input was actually binding at any given moment, a problem that the growth diagnostics approach directly addresses.
15 Part II — Convergence, Clubs, and the Middle-Income Trap
15.1 Conditional convergence: the promise and its conditions
The Solow model contains an intuitive promise: poor countries should grow faster than rich ones. In poor countries, capital is scarce and its marginal return should be high, attracting investment and accelerating growth. In rich countries, capital is abundant and its marginal return is lower. This logic predicts convergence: retardataire economies should catch up with advanced ones.
The historical record refutes this simple prediction spectacularly. Countries rich in 1960 were generally also rich in 2024. Many poor countries remained poor or converged only modestly. There is no unconditional convergence in the global data. If anything, income gaps between the richest and poorest countries widened over some periods.
Robert Barro and Xavier Sala-i-Martin documented something more precise: conditional convergence. Within groups of countries sharing comparable characteristics, similar education levels, similar macroeconomic stability, similar institutional quality, poorer countries do converge toward richer ones. The estimated convergence speed is approximately 2% per year, the “iron law of convergence.” This may seem encouraging, but it implies a half-life of approximately 35 years: a country whose income is half its steady-state level takes more than three decades to close half that gap, and only if conditions remain stable throughout.
The critical qualifier is conditional: convergence occurs toward each country’s own structural equilibrium, not toward a common global one. The structural equilibrium depends on the quality of institutions, the depth of human capital, the sophistication of productive structure, and the quality of macroeconomic governance. Improving these conditions raises the structural equilibrium itself, which is how countries escape their current convergence trajectory and join a higher-income club.
15.2 Clubs of convergence and the middle-income trap
Convergence clubs, groups of countries converging among themselves at different income levels, are the empirical reality. Many African and Arab economies converge among themselves toward a relatively poor structural equilibrium, without approaching advanced economies. For a country to escape its current club and join a higher-income one requires not just accumulating more capital but transforming the structural conditions that determine where it converges.
The middle-income trap illustrates this mechanism with particular clarity. Economies like Brazil, Mexico, South Africa, and Thailand reached middle-income levels without managing the transition to high-income status. Their growth slowed, as if a restoring force maintained them in an intermediate zone. The economic explanation identifies several mechanisms. In the first phase, competitiveness rests on low wages, which attracts labour-intensive manufacturing. Then wages rise with prosperity, eliminating this advantage. Transitioning to quality and innovation-based competitiveness requires institutions, firms, and capabilities that the economy has not yet built. The trap is precisely the inability to make this qualitative leap at the moment when the cost advantage is eroding.
Morocco’s positioning in these convergence clubs deserves precise analysis. In 2024, Morocco’s GDP per capita at purchasing power parity (approximately 9,066 international dollars in 2021 constant prices) sits in the lower-middle-income category, above many poor countries but far from advanced economies. Several indicators suggest the trajectory is moving toward the higher-income club, but not quickly enough: real export growth in automotive and aerospace, improving ECI scores, and infrastructure of international quality are positive signals. But the structural warning signs are also visible: a manufacturing share of GDP around 14-15%, graduate unemployment above 19%, and a persistent current account deficit that reflects insufficient export diversification. The question is whether the ongoing structural transformation will accelerate enough to avoid the middle-income trap, or whether it will stagnate in the comfortable but ultimately inadequate pattern of assembly-dependent growth.
16 Part III — Structural Transformation: The Engine of Development
16.1 Lewis’s dual economy: the mechanism of reallocation
W. Arthur Lewis’s 1954 model of economic development with unlimited supplies of labour remains the most influential framework for understanding how structural transformation generates growth in the early stages of development.
Lewis’s model posits an economy divided into two sectors. The traditional sector (agriculture, subsistence activity, informal services) is characterised by a marginal product of labour so low, sometimes near zero in conditions of surplus population, that removing a worker reduces output negligibly. This is disguised unemployment: workers are employed but their marginal contribution is minimal. The modern sector (formal industry, export services) pays a higher institutional wage and uses labour at its full marginal productivity.
The engine of growth is the transfer of labour from the traditional to the modern sector. Because the marginal product of labour is low (or zero) in the traditional sector, each transfer generates a net productivity gain: the transferred worker produces much more in the modern sector than she did in agriculture. This gain does not require any new technology or additional capital, it comes purely from reallocating a factor from under-use to productive use. This reallocation gain can sustain rapid growth for decades without requiring the productivity improvements that drive long-run growth in advanced economies.
The process continues as long as the traditional sector has surplus labour to supply to the modern sector at roughly the subsistence wage. This is Lewis’s unlimited supplies of labour phase: the modern sector can expand without bidding up wages, generating high profits that are reinvested, financing further expansion, attracting more workers, and so on in a self-sustaining cumulative dynamic.
The Lewis turning point, when surplus labour is exhausted and wages begin rising, marks the transition from extensive to intensive growth. After the turning point, growth requires genuine productivity gains rather than reallocation from underuse to use. This transition is considerably more difficult and explains the growth slowdown that many middle-income countries experience as they approach and cross the Lewis turning point.
For Morocco, approximately 29.6% of employment remains in agriculture in 2023, indicating that the traditional sector still holds substantial surplus that could, in principle, fuel reallocation gains. But capturing these gains requires the modern sector to create enough formal productive employment to absorb the transfer. Morocco’s challenge is that the informal urban sector often absorbs workers leaving agriculture without significantly raising their productivity, a pattern of modernisation without development: workers move from low-productivity agriculture to low-productivity urban informality. For the Lewis mechanism to operate, the formal modern sector must generate employment at sufficient scale and productivity to make the reallocation genuinely transformative.
16.3 Prebisch-Singer and the terms of trade: the structuralist warning
Raúl Prebisch and Hans Singer, working independently in 1950, formulated the thesis of the secular deterioration of the terms of trade for primary commodity exporters, drawing on two structural asymmetries between industrial centre and primary periphery.
On the demand side, Engel’s Law applies internationally: as incomes grow, the share spent on food and raw materials falls, while the share spent on manufactured goods and services rises. This means that as world income grows, demand for primary commodities grows more slowly than demand for manufactured goods, structurally limiting the export revenues of commodity-specialised economies.
On the supply side, the distribution of productivity gains is asymmetric. In the industrial centre, productivity gains from technical progress are captured as higher wages and profits, because organised labour markets and concentrated market structures allow workers and firms to capture the gains from efficiency. In the commodity periphery, productivity gains transmit into lower prices, because competitive commodity markets and weak labour organisations prevent capture. Consequently, the periphery must progressively sell more commodity exports to buy the same quantity of manufactured imports. Trade, far from equalising living standards as Ricardo’s comparative advantage promised, structurally transfers value from periphery to centre.
The empirical record is mixed over different periods and commodity groups. But the core insight has proved analytically durable: commodity specialisation exposes economies to terms-of-trade deterioration and devastating price volatility, and the project of industrialisation as a pathway out is well-founded precisely in this analysis.
For Morocco, the Prebisch-Singer lens is directly operational. Morocco’s phosphate endowment is a structural comparative advantage, but raw phosphate rock is a commodity whose price is volatile and whose terms of trade relative to manufactured imports could deteriorate over long horizons. OCP’s strategic transformation from raw rock exporter to processed fertiliser producer to customised agri-nutrition supplier is, at its core, a response to the Prebisch-Singer diagnosis: move up the value chain before the commodity terms of trade turn against you. The battery materials trajectory, positioning Moroccan phosphate in the electric vehicle supply chain as cathode material, extends this logic to the 21st century technology cycle. The structuralist prescription and the product space strategy converge on the same industrial policy conclusion.
16.4 Kaldor’s laws and the engine of manufacturing growth
Nicholas Kaldor articulated three empirical regularities about economic growth, known as Kaldor’s three laws, that provide the theoretical foundation for the centrality of manufacturing in the development process.
Kaldor’s First Law: the growth rate of GDP is positively correlated with the growth rate of manufacturing output. Manufacturing is the engine sector: its expansion generates broader economic dynamism that agriculture and services do not generate at the same intensity.
Kaldor’s Second Law (Verdoorn’s Law): manufacturing productivity growth is positively correlated with manufacturing output growth. This is the dynamic increasing returns phenomenon: as manufacturing expands, it generates learning-by-doing, economies of scale, and process improvements that raise productivity, creating a cumulative advantage for early industrialisers. The empirical coefficient (Verdoorn coefficient) is typically 0.5: a 2% faster manufacturing growth generates approximately 1% faster manufacturing productivity growth. This self-reinforcing dynamic explains why, once launched, industrial catching-up can be very rapid.
Kaldor’s Third Law: the growth of manufacturing output generates productivity growth in non-manufacturing sectors by pulling in labour from lower-productivity activities (the Lewis mechanism) and by generating demand for services and input supply that raises productivity in those activities too. Manufacturing is the sector with the strongest positive linkage effects on the rest of the economy.
Together, Kaldor’s laws provide a theoretical basis for what the empirical shift-share evidence confirms: manufacturing generates growth and productivity through multiple channels simultaneously that no other sector replicates. The implication for Morocco is direct. A development strategy that relies on services, tourism, and agriculture without building a substantial manufacturing capability is structurally handicapped: it forgoes the Verdoorn dynamics and the cross-sectoral productivity effects that manufacturing uniquely generates.
17 Part IV — The Product Space and Economic Complexity
17.1 Capabilities, not sectors
Ricardo Hausmann, César Hidalgo, and Bailey Klinger proposed a radically different way of thinking about economic diversification. Their central insight: what matters in development is not the sector to which a product belongs in statistical classifications, but the capabilities it mobilises, tacit knowledge, organisational routines, standards, supplier networks, institutions, local social capital. Two products are “close” in the product space if they mobilise similar capabilities, and “distant” if they require radically different ones.
The product space is the graphic representation of this network of proximities. Products are nodes; links represent the statistical probability that a country exporting one also exports the other. Observing this network across all countries over decades established a major stylised fact: economies move through the product space by adjacency, developing products close to those already mastered before progressing toward more distant and sophisticated ones. “Long jumps”, entering sectors radically different from current capabilities, rarely succeed because they require capabilities the economy does not yet possess.
The centre of the network is dense: semiconductors, precision instruments, fine chemicals, machine tools, products requiring many shared capabilities and close to many other sophisticated products. The periphery is sparse: minimally processed agricultural goods, raw minerals, basic textiles, products requiring few capabilities and far from high-value segments.
A country positioned at the periphery of the product space has few natural doors toward complex activities. A country near the centre can pivot easily toward new sophisticated segments. This asymmetry explains why product space analysis is now central to development strategy: it maps not just what a country currently does but what it could plausibly do next, and what investments would open the most opportunities.
17.2 The Economic Complexity Index and Morocco’s position
Hidalgo and Hausmann’s Economic Complexity Index (ECI) measures the diversity and ubiquity of a country’s exports, in summary, the richness of its capability fabric. High-ECI countries (Japan, Germany, Switzerland, South Korea) export many products that few countries know how to produce. Low-ECI countries export few products, and those products are produced by many other countries.
Morocco displays a slightly positive and growing ECI since the 2010s, a sign of progressive complexity upgrading, but remains below the level expected for its income per capita. This complexity gap means that Morocco’s export basket is less diversified and sophisticated than its relative wealth would suggest. This gap is precisely the space that industrial policy must fill.
In 1965, Morocco exported primarily raw phosphate and citrus fruits: a firmly peripheral position in the product space with few links toward dense manufacturing segments. South Korea, at the same period and comparable income level, exported textiles and light goods, modest products, but located in a zone of the network opening doors toward light mechanics, electronics assembly, steelmaking, then automobiles and semiconductors. Sixty years later, the contrast is stark: South Korea occupies the core of the global product space. Morocco has progressed, the emergence of automotive (wiring, then Renault-Tanger assembly and Stellantis-Kénitra) and aerospace (Safran, Boeing, Bombardier at Casablanca-Nouaceur) represents entry into medium-density segments, but remains at mid-space, with an export base still concentrated on phosphates and derivatives, automotive assembly, and minimally processed agri-food.
The strategic question is not “which sector to invest in” but “which products in the product space are simultaneously reachable from current capabilities and open doors toward more sophisticated products?” From automotive wiring, credible adjacencies include: embedded electronics, mechatronics, electrical equipment for electric vehicles, battery management systems. From phosphates and fertilisers: specialty chemicals, agrochemistry, specialised fertilisers for differentiated African markets. From aerospace assembly: precision machining, maintenance-repair-overhaul (MRO), lightweight composites.
What would be a jump too long at this stage: advanced semiconductors, industrial robotics, biotechnology. The distance in the product space reflects the current absence of the intermediate capabilities these require.
18 Part V — Industrial Policy: Theory, Conditions, and the State’s Four Roles
18.1 Ha-Joon Chang: kicking away the ladder
Ha-Joon Chang’s Kicking Away the Ladder (2002) makes an argument that is simultaneously historically documented and politically explosive: all currently rich countries used activist industrial policy, protection, and state intervention during their development phase, and then, once developed, imposed free trade and market liberalisation on late-developing countries precisely by removing the policy tools that had enabled their own development.
Britain protected its textile industry for centuries before achieving comparative advantage in it, then preached free trade once that advantage was secure. The United States had the highest average tariff rates in the world throughout the 19th century while industrialising, and used import substitution industrialisation as aggressively as any 20th-century developing country, then promoted free trade only once its manufacturing supremacy was established. Germany under List’s influence explicitly rejected free trade in favour of infant industry protection. Japan, South Korea, and Taiwan all used protection, selective credit allocation, and industrial targeting extensively and successfully. The historical record is unambiguous.
The policy implication for Morocco is politically uncomfortable but analytically inescapable. The prescriptions of international financial institutions and trade agreements, reduce tariffs, eliminate export subsidies, open capital accounts, enforce intellectual property rights consistent with TRIPS, are in many respects the policies that developed countries refused to adopt during their own industrialisation. This is not a justification for unlimited protection: the failure modes of industrial policy are real and well-documented. But it is a justification for Morocco designing its industrial policy on the basis of what the historical record shows to have worked, rather than what developed-country institutions prescribe based on models calibrated to their own current circumstances.
18.2 Absorptive capacity and the technology gap
Wesley Cohen and Daniel Levinthal’s concept of absorptive capacity (1990) resolves a puzzle that technology diffusion theory often elides: if advanced technology is available globally, why don’t all countries simply adopt it and converge rapidly?
The answer: to assimilate technology from abroad, it is not enough to have access to it. One must already possess sufficient knowledge and competence to understand, adapt, and eventually improve upon it. Learning requires prior learning: absorptive capacity is itself the product of prior investment in education, R&D, and practical industrial experience.
Paradoxically, you need to already know a great deal to be able to learn more. A factory that imports German machinery runs it but may not understand why the tolerances are set as they are, or how to adjust the process when inputs change, or how to diagnose a systematic quality defect in the output. Understanding these things requires engineers trained in the relevant industrial process, which requires universities teaching applied engineering in collaboration with industry, which requires industry sophisticated enough to engage with universities as partners. Without absorptive capacity, the most modern technology remains a black box that is operated but not learned from.
This is what separates East Asian catching-up from blocked catching-up. South Korea and Taiwan invested heavily in engineering education, industrial training systems, and reverse engineering of foreign technologies before the technology transfer agreements that brought Japanese and American firms in. When foreign firms arrived, Korean engineers could decode what they saw and gradually absorb the knowledge. In many other developing countries, foreign investment arrived without the domestic absorptive capacity to transform technology access into technological mastery. The factories operated; the technology remained foreign.
For Morocco, this analysis adds a critical dimension to the product space and industrial policy discussion. The adjacencies available in the product space from current positions (battery materials from phosphates, embedded electronics from automotive wiring, composites from aerospace assembly) are only reachable if Morocco builds the engineering absorptive capacity to exploit them. This means targeted investment in applied engineering education at UM6P and the national engineering schools, in industrial research partnerships with the automotive and aerospace clusters, and in apprenticeship systems that create practical learning pathways for young technicians. The product space maps what is possible; absorptive capacity determines whether it is accessible.
18.3 The Big Push and coordination failures
Paul Rosenstein-Rodan and Ragnar Nurkse identified a fundamental mechanism that explains why private markets systematically underinvest in structural transformation: coordination failures. A modern factory needs roads, energy, trained workers, suppliers, markets, and financing. If each actor waits for the others to invest first, the economy can remain trapped in a low-investment equilibrium even when all the investments would be collectively profitable.
The Big Push insight is directly applicable to Tanger Automotive City: by concentrating Renault, its first-tier suppliers, infrastructure, energy, and worker training in a dedicated zone, the Moroccan state created the complementarities that no single private investor could have assembled alone. The lesson from Hirschman applies here: what matters is not just the anchor firm but whether it triggers genuine upstream and downstream linkages that build productive capability in the surrounding economy.
18.4 Hirschman’s linkages: the criterion of selection
Albert Hirschman proposed in 1958 a way of evaluating sectors that cuts across the simple profitability calculation. His question was not “how much does this sector earn?” but “what chains does it trigger?” A sector creates backward linkages when it generates demand for inputs, components, and technical services, pulling a galaxy of suppliers behind it. It creates forward linkages when its products serve as inputs to other industries. It generates learning effects when the routines it imposes, quality, metrology, planning, maintenance, diffuse into other sectors.
When these mechanisms co-activate and mutually reinforce, Hirschman speaks of a “multidimensional conspiracy in favour of development”: a sheaf of effects that spill far beyond the direct value added of the sector. This is the logic of inducement, initial investment induces complementary investments, that distinguishes a “teaching sector” from an “enclave sector.”
The distinction between the two is not a question of sectoral label but of context, timing, and governance. Textiles in 18th-century England served as a laboratory for mechanisation, factory organisation, and logistics: a teaching sector par excellence. The same textiles, introduced late into a country without machinery industry, without supplier networks, without design transfer, can be an export enclave at low wages that diffuses nothing. Automotive can build an ecosystem of suppliers and competencies, as in Germany or Korea, or remain an assembly of imported parts on a line built by a foreign investor without local roots, as in some export processing zones.
Hirschman’s test for Morocco is therefore: what happens if the lead firm leaves? If its departure would leave behind a fabric of competent suppliers, reusable skills, embedded know-how, and local innovation capacity, then a territorial productive system has constituted itself. If it would leave an empty hangar and unemployed workers, the district was a satellite platform, dependent and moveable. Morocco’s automotive sector is moving toward the first outcome but has not fully arrived. Morocco’s offshore services and lower-tier textile assembly remain closer to the second.
18.5 The four conditions of industrial policy
The literature on comparative industrialisation, drawing on the East Asian successes (Korea, Taiwan, Singapore), the partial successes (Malaysia, Thailand, Brazil), and the failures (many Sub-Saharan Africa industrial programmes, some MENA attempts), converges on four conditions that separate industrial policy that generates learning from industrial policy that generates rent.
Condition 1: Real coordination failure. The intervention must address a genuine market failure, typically a coordination problem (the complementary investments that private markets cannot simultaneously provide), a learning externality (the pioneer who bears the cost of discovery allows imitators to free-ride without bearing that cost), or a capital market failure (long-horizon industrial investment that private finance cannot accommodate). Industrial policy that does not address a real coordination failure is simply resource allocation by political preference, generating no productivity benefit and creating rents.
Condition 2: Institutional capacity. The state must have the administrative capability to select promising sectors, design appropriate instruments, monitor performance, and resist capture. This capacity is not generic; it is sector-specific. A state competent to regulate telecoms may not be competent to administer an automotive supplier development programme. Building institutional capacity is itself an investment in development, and one that tends to be chronically underfinanced in countries that most need it.
Condition 3: Conditional reciprocity. Support must be conditional on measurable performance, export growth, productivity improvement, technology upgrading, local content increase. The East Asian developmental states were stringent about this discipline: firms that received protection, credit, or incentives were required to demonstrate performance within defined timeframes or lose the support. Without this conditionality, industrial policy becomes unconditional protection, creating incentives to lobby for continued support rather than to improve performance.
Condition 4: Credible exit mechanism. Support must have a defined end condition, either reaching the target performance level or being cut for failure to progress. Infant industry protection that has no credible sunset clause becomes permanent protection, converting learning subsidies into permanent rent transfers. The political economy challenge is that by the time a sunset clause is scheduled to activate, the supported firms have become politically organised to resist exit. Designing credible pre-commitment mechanisms is among the most difficult aspects of industrial policy governance.
18.6 The Kuznets curve and productive versus rentier inequality
Simon Kuznets’s inverted-U hypothesis (1955) predicts that income inequality first rises during early industrialisation before falling as development matures. The mechanism: as workers migrate from low-wage agriculture to higher-wage urban industry, earnings inequality initially increases (because the distribution widens between those who have made the transition and those who have not) before narrowing as the transition completes, the industrial workforce unionises, and the state expands redistributive capacity.
Empirically, the Kuznets curve is not universal, it is a statistical tendency that holds for some development trajectories but not all, and its parameters depend heavily on institutional configuration. What the framework usefully contributes is a distinction between two types of inequality that have very different economic and social implications.
Productive transitional inequality accompanies genuine structural transformation: engineers earn more than farmers in an industrialising economy because their skills are genuinely scarcer and their productivity genuinely higher. This inequality is socially acceptable if it is transitional (the educational pathway to becoming an engineer is accessible) and if growth is broad-based enough to raise all incomes even while gaps widen at the top. Rentier permanent inequality reflects capture of resources by politically connected groups rather than genuine productivity differences: connected businesses pay below-market rates for land, receive exclusive licences that prevent competition, and capture public contracts through relationship rather than merit. This inequality is economically inefficient and socially corrosive, it compounds over generations rather than narrowing as Kuznets predicted.
Morocco’s inequality profile combines both types. The productivity-based inequality between formally educated urban professionals and informal rural workers reflects genuine structural transformation in progress, and is, in principle, addressable through educational investment and formal sector job creation. The rent-based inequality documented by Saadi in Moroccan cronyism, sectors where political connection determines market access, regulatory outcomes, and public contract allocation, is a different phenomenon requiring different instruments: competition policy, regulatory independence, public procurement transparency, and governance reform.
The Nouveau Modèle de Développement (NMD, 2021) diagnosed both dimensions explicitly. Its central finding, that Morocco’s growth model has been insufficiently inclusive, generating macroeconomic aggregates that mask a distribution of gains heavily skewed toward the formally educated and politically connected, converges on the same conclusion from a policy development perspective. The NMD’s recommendations for structural transformation, educational reform, social protection expansion, and private sector competitiveness improvement map directly onto the analytical framework of this chapter.
18.7 Peter Evans’s four roles of the developmental state
Peter Evans (Embedded Autonomy, 1995) argued that the question “more or less state intervention?” entirely misses the point. What distinguishes industrial policy success from failure is not the quantity of intervention but its nature and appropriateness to the sectoral context. He identifies four distinct state roles, each corresponding to a specific type of market failure at a specific stage of industrial maturity.
The Custodian (gardien) preserves market contestability: applying competition law, dismantling unjustified rents, guaranteeing incentive transparency, preventing abuse of dominant position. Without a functioning Custodian, rents proliferate, industrial protection becomes permanent, and zombie industries survive at the cost of dynamic innovators. But Evans identifies a symmetric danger: the heavy procedural Custodian, with permit systems, quotas, redundant certifications, consumes disproportionate administrative capacity for limited benefit. In Morocco, the ANRT plays this role in telecoms with uneven success on the antitrust dimension; the Conseil de la Concurrence exercises it transversally but lacks adequate investigative resources and robust leniency procedures.
The Producer (demiurge) creates productive capacity directly when private capital cannot or will not assume the required risk and investment scale. The canonical example is POSCO in South Korea: in the 1960s, private Korean capital was absent, multinationals refused to engage, yet cheap steel was the prerequisite for automobiles, shipbuilding, and appliances, the pillars of the Korean export model. POSCO did not just produce steel: it produced engineers, managers, qualified subcontractors, a dedicated university (POSTECH), and technology transfer partnerships with U.S. Steel. The investment seeded an entire ecosystem. OCP’s trajectory, from raw phosphate extraction to integrated fertiliser production to African agri-nutrition strategy, illustrates how a public producer can evolve from extractive enclave toward systemic anchor, provided strategic vision and sufficient institutional autonomy allow the transformation.
The Midwife (accoucheur) induces the emergence of private entrepreneurs and new sectors by reducing uncertainty and orchestrating complementary institutions. Rather than creating firms directly, the Midwife sends credible signals (this sector has a future here, these are the required standards, this infrastructure will be built on this schedule), coordinates (connecting potential suppliers with potential clients, co-investing in shared technology centres), and provides temporary conditional protection for infant industries. Singapore’s Economic Development Board played this role exemplarily for electronics in the 1960s-1970s: not creating public electronics factories but negotiating with Texas Instruments and Hewlett-Packard, demanding capability transfers in exchange for incentives, and progressively raising the sophistication requirements of implanted activities. Morocco’s automotive PAI is essentially Midwife policy: Tanger-Med construction, equipped free zones, dedicated training (IFMIA), fiscal facilitation, and negotiated progressive local content commitments with Renault and Stellantis. The result, over 250 equipment suppliers installed, automotive as first export sector, validates the approach.
The Husbandman (éleveur) assists once private actors have entered a sector, helping them upgrade, improve processes, certify products, access new markets, and finance R&D. The Husbandman takes on the complementary tasks that private firms would underinvest in because learning returns are not fully appropriable by the investor. Morocco has Husbandman instruments: Maroc PME programmes, sectoral technical centres (automotive, aerospace, textile), CNRST research co-financing, OFPPT continuous training. But two structural limits persist. First, conditionality is insufficient: modernisation support is often decoupled from measurable performance milestones. Second, evaluation is rare: few programmes undergo rigorous impact analysis to distinguish what works from what does not.
Evans’s synthesis rule: states should not choose one role and apply it universally. They should adapt the role to the sectoral context and to their own administrative capabilities. At a given institutional capacity, the Midwife → Husbandman sequence is often dominant: it requires fewer scarce resources than the Producer, is less exposed to capture risks than the heavy procedural Custodian, and creates self-reinforcing private dynamics. But when a critical upstream asset is absent and the private sector cannot provide it (steel, utilities, fundamental applied research), the targeted Producer role remains justified, provided the exit trajectory is planned from the outset.
19 Part VI — Growth Diagnostics: Finding the Binding Constraint
19.1 The HRV decision tree
Ricardo Hausmann, Dani Rodrik, and Andrei Velasco (2005) proposed a method for identifying the binding constraint on private investment and growth in a specific country at a specific time. Their insight: rather than applying a standard reform package (improve governance, liberalise trade, reduce regulation), ask which specific constraint, if relaxed, would have the largest impact on the return to private investment.
The method proceeds as a decision tree. The central question: is private investment low because the return is low or because the cost of finance is high?
If the cost of finance is high, is it because domestic savings are insufficient (savings constraint) or because international finance is inaccessible (country risk constraint)?
If returns are low, is it because of appropriability problems (firms cannot keep the returns they generate, due to corruption, predatory taxation, unclear property rights, or crime) or because of coordination failures (the complementary inputs, infrastructure, skills, inputs, that would make investment productive are absent)?
Each branch of the tree generates diagnostic tests, price signals, firm behaviour, growth responses to reforms, that allow ruling out incorrect hypotheses and converging on the most likely binding constraint.
Price signals provide the first diagnostic layer. If borrowing rates are very high relative to comparators, financial constraint is likely. If returns to education are much lower than in comparators, the bottleneck is not human capital but productive demand for it. If firms extensively use private generators for electricity, the infrastructure constraint is revealed by the private cost of circumventing it. Firm behaviour provides the second layer. If firms that do manage to access finance grow disproportionately faster than peers, finance is likely binding. If the most capable firms invest elsewhere rather than expanding locally, appropriability problems may dominate. Growth responses to past reforms provide the third layer: if financial liberalisation reforms did not trigger investment surges, finance was not the binding constraint.
19.2 Thirlwall’s constraint: the balance of payments as long-run ceiling
A.P. Thirlwall (1979) identified a fundamental external constraint on long-run growth that complements the production-function frameworks from a demand-side perspective: the balance of payments constraint. While Solow focuses on supply-side limits to growth (capital and TFP), Thirlwall shows that demand-side external constraints can be equally binding for small open economies. A country cannot sustain a growth rate that continuously deteriorates its current account, because at some point the external financing dries up and a painful adjustment forces growth below potential.
Thirlwall’s law derives the maximum sustainable growth rate:
\[g^* = \frac{\varepsilon \cdot g_w}{\pi}\]
where \(\varepsilon\) is the income elasticity of exports (how much export volume grows for each 1% increase in world income), \(\pi\) is the income elasticity of imports (how much import volume grows for each 1% increase in domestic income), and \(g_w\) is world income growth.
If \(\varepsilon = 1.2\), \(g_w = 3\%\), and \(\pi = 1.8\), the maximum sustainable growth rate is \(\frac{1.2 \times 3}{1.8} = 2\%\), well below what Morocco needs to create jobs for its demographic cohort. Accelerating growth sustainably requires either raising \(\varepsilon\) (exporting products whose demand grows faster as world incomes rise, moving up the product space toward income-elastic goods) or reducing \(\pi\) (substituting imports in strategic sectors, building domestic supply chains that reduce the import content of growth). Both require exactly the structural transformation and industrial policy that this chapter analyses.
19.3 The growth diagnostic applied to Morocco
Applying the HRV framework to Morocco’s specific constraints produces a differentiated and operationally useful diagnosis.
Is low investment due to high financing cost or low returns? Signal analysis: real borrowing rates in Morocco are not anomalously high by regional MENA standards. The financial reforms of the 1990s did not generate a private investment surge in manufacturing, suggesting financial cost is not the primary binding constraint. The constraint appears to be on the return side.
If returns are low, is it appropriability or coordination failures? Morocco’s macroeconomic stability, relatively secure property rights, and functioning judicial system make classic appropriability problems (predatory taxation, confiscation) a second-order concern. The evidence points more strongly toward coordination failures: the complementary inputs that would make industrial investment productive, qualified technical workers for specific industrial processes, local supplier networks capable of meeting international quality standards, logistics infrastructure connecting industrial zones to input markets, and applied research that bridges the gap between engineering education and industrial practice, are insufficiently developed relative to what would justify the level of private industrial investment Morocco needs.
The multiplicity diagnosis: unlike the clean single-constraint result that the standard HRV tree idealises, Morocco’s binding constraints appear multiple and interlocked. Low local content in export sectors reflects insufficient local suppliers. Insufficient local suppliers reflect insufficient investment in industrial skills. Insufficient industrial skills investment reflects low expected returns if industrial employers are few. Low industrial employers reflect low local content and therefore few supplier opportunities. This multidimensional vicious circle is precisely what the thick diagnostic approach, combining quantitative signals with enterprise surveys, sectoral case studies, and political economy analysis, seeks to identify the entry point for breaking.
The policy implication is that generic business climate improvements (one-stop-shop registration, administrative simplification, contract enforcement speed) are necessary but insufficient: they do not address the specific coordination failures in industrial upgrading. The most promising interventions would simultaneously address complementarity development (local supplier capability programmes with measurable quality milestones) and performance discipline (conditional public support with monitoring and exit mechanisms), rather than additional horizontal tax incentives that do not change the underlying complementarity structure.
Morocco’s enterprise surveys in the automotive sector are consistent with this diagnosis. The constraint most frequently cited by international tier-1 suppliers considering deepening their Moroccan sourcing is not the regulatory environment (already internationally competitive) nor financing cost (adequately managed through investment code instruments) but the availability of qualified local second-tier suppliers: Moroccan firms with the technical capability, quality management systems, and financial depth to supply complex components reliably at international standards. This is a textbook complementarity failure. Anchor firms want to source locally; local firms want to supply; but the knowledge and quality gap between international standards and current local capability prevents the market from clearing. Evans’s Husbandman role is designed for exactly this, not protecting current suppliers from competition but co-investing in the capability-building that closes the gap.
The electricity sector offers a contrasting example of an appropriability constraint. Renewable energy investment was initially constrained by a regulatory architecture that prevented independent producers from selling directly to industrial consumers. Investors could not capture the return because revenues were directed through ONEE rather than the producer. The 2021 reform authorising direct energy sales between renewable producers and large industrial consumers addressed precisely this constraint and triggered rapid expansion of project development. The diagnostic logic predicted this: private generators proliferating (revealing high willingness-to-pay for reliable energy), source identified as regulatory rather than technical, minimal targeted reform designed rather than comprehensive restructuring.
19.4 Morocco’s development trajectory: a synthesis
Applying the full analytical toolkit of this chapter to Morocco produces a nuanced verdict, not optimism or pessimism, but a structured map of where the country stands and what the analytical evidence implies about what must change.
In product space terms, Morocco has moved from a peripheral position (raw phosphates, citrus, basic textiles) toward a mid-space position (automotive assembly and components, aerospace subcontracting, processed fertilisers). The ECI trajectory is positive. But the distance between Morocco’s current product space position and the dense core of sophisticated manufacturing remains large. From current positions, the credible adjacencies in electric vehicle components, specialty chemicals, and precision aerospace manufacturing represent the next logical steps, not leaps to semiconductors or advanced robotics.
In structural transformation terms, the McMillan-Rodrik decomposition for Morocco shows that the between-sector reallocation component has been positive but insufficient: workers have moved from agriculture toward formal industry and modern services, but the pace and scale of formal manufacturing job creation has not matched the scale of agricultural labour release. The informal sector has absorbed a significant share of the transfer, limiting the productivity gains that the Lewis mechanism would otherwise generate. Premature deindustrialisation is a risk: Morocco’s manufacturing employment share has not grown as fast as its income level would predict based on historical comparators.
In institutions terms, Morocco has made real but incomplete progress toward inclusivity. The 2011 constitutional reform, the 2015 Régionalisation Avancée, and the 2021-2025 social protection reform all represent institutional investment in broader participation in growth. The Nouveau Modèle de Développement’s diagnosis of an insufficiently inclusive growth model acknowledges the rentier inequality problem. But the pace of institutional transformation in competition policy, public procurement transparency, and regulatory independence has not kept pace with the pace needed to support the creative destruction that further economic complexity requires.
In growth diagnostics terms, the binding constraint appears to be a combination of complementarity failures (insufficient local supplier capability relative to anchor firm demand) and insufficient absorptive capacity (engineering education not yet producing the graduate volume and quality needed for the planned industrial trajectory). Both are addressable through coherent industrial policy, but they require sustained investment in specific institutional mechanisms rather than generic business climate improvement.
The window of opportunity is real but time-bounded. The nearshoring dynamics post-COVID, Morocco’s geographic advantage for European supply chains, the renewable energy resource base, and the phosphate positioning for electric vehicle battery materials all create conditions for accelerated structural transformation. But this window will not stay open indefinitely: automation advances, competing countries are also improving their capabilities, and the EV transition’s demand for battery materials depends on technology trajectories that Morocco does not control. Acting now, with the coherent industrial policy logic this chapter has developed, is a precondition for making the transition from semi-periphery to industrialised economy within the current generation.
20 Conclusion: Growth as Construction, Not Gift
This chapter has traversed two centuries of economic thought on growth and development, from Smith’s pin factory to Hausmann and Hidalgo’s product space algorithm. Three structural tensions organise the corpus.
Accumulation versus transformation. Standard models measure capital and labour accumulation. What remains unexplained, the Solow residual, is precisely what matters most for long-run development: institutions, knowledge, organisational capabilities, and the quality of the knowledge base. Endogenous growth theories attempt to explain the residual; developmentalist approaches locate it in irreducible historical and political structures.
Convergence versus divergence. Standard models predict conditional convergence at approximately 2% per year. Economic history shows persistent divergences, between nations, between regions, between groups. The balance of payments constraint, unequal exchange, premature deindustrialisation: all are mechanisms that explain why convergence remains the exception rather than the rule.
Market versus state. Developmentalist approaches from Lewis to Hirschman, from Prebisch to Rodrik, share a conviction: structural transformation does not happen spontaneously. It requires active industrial policy, capable governance, and credible institutions. But these same institutions can capture the rents they were meant to allocate.
For Morocco, these tensions define the development agenda: industrialise without remaining trapped in assembly; diversify without dispersing scarce resources; invest in human capital without losing a generation to waiting. The four-condition framework for industrial policy, Evans’s four state roles, and the HRV growth diagnostics are not academic tools but operational instruments for navigating these tensions with analytical precision rather than ideological autopilot.
21 Further Reading
Solow, R.M. (1956). “A Contribution to the Theory of Economic Growth.” Quarterly Journal of Economics. The foundational neoclassical growth model. Short and still essential.
Romer, P. (1990). “Endogenous Technological Change.” Journal of Political Economy. The founding paper of endogenous growth theory. Ideas as non-rival goods, the key intellectual innovation.
Lewis, W.A. (1954). “Economic Development with Unlimited Supplies of Labour.” The Manchester School. The most influential paper in development economics of the 20th century. Unchanged reading 70 years later.
Rodrik, D. (2007). One Economics, Many Recipes. Princeton University Press. The most rigorous argument for context-sensitive industrial policy. Chapter 4 on growth diagnostics is required reading.
Hausmann, R. and Hidalgo, C.A. (2009). “The Building Blocks of Economic Complexity.” PNAS. The product space and ECI. Empirical and accessible, complement with the Atlas of Economic Complexity website.
Hirschman, A.O. (1958). The Strategy of Economic Development. Yale University Press. Linkages, inducement mechanisms, and the logic of choosing sectors. The most operationally useful text in development economics.
Evans, P. (1995). Embedded Autonomy: States and Industrial Transformation. Princeton University Press. The four roles of the state and the concept of embedded autonomy. Essential for evaluating any industrial policy programme.
McMillan, M. and Rodrik, D. (2011). “Globalization, Structural Change, and Productivity Growth.” NBER Working Paper. The shift-share decomposition and the premature deindustrialisation evidence. The empirical reference on structural transformation mechanics.
Nouri, F. (à paraître). “Why Catch-Up Remains Rare.” The four-condition framework for industrial policy, with comparative evidence from Korea, Malaysia, and Morocco. The primary source for the four conditions presented in this chapter.
Atlas of Economic Complexity. atlas.cid.harvard.edu, interactive visualisation of the product space and Morocco’s position within it. Use it in parallel with this chapter.