Explanation:

 the risk of a Late Convergence Stall needs to be taken seriously because of four headwinds: the hyper-globalization repudiation, thwarted/impeded structural transformation, human capital regression induced by technological progress, and climate change-induced agricultural stress.

A. HYPERGLOBALIZATION REPUDIATION
Developing countries that came late to convergence now face a very different global trading environment from their predecessors. Early convergers benefited from the process of rapid globalization or hyper-globalization, reflected in dramatic increases in the world trade-GDP ratio. As a result, Japan, South Korea and China were all able to post average export growth rates of over 15 percent for the thirty years of their convergence periods. But this globalization has led to a backlash in advanced countries reflected in the decline in world trade-GDP ratios since 2011. This means that the trading opportunities available to the early convergers, specifically the ability to export at double digit rates of growth for three to four decades consistently, may no longer be available.
One way of understanding the potential impact of the hyperglobalization repudiation is to seek recourse to the gravity model of trade. Basic gravity theory implies that smaller countries tend to trade more than larger ones. A world made up of two equal-size countries will experience more trade than a world in which the larger country accounts for 95 percent of world output. Over time, the world is becoming more equal in the distribution of the underlying output. That is the consequence of convergence. Therefore, if there is convergence, the gravity model suggests there will also be increased trade.
Since 2000, as more countries have started catching up with the rich, world output has become more dispersed. Taking the list of top 50 countries (excluding oil exporters) and calculating the distribution of world output suggests that in 2016 there are about 9.6 country-equivalents in the world. Now, for one or a few countries such as India, there need not be such an external constraint on growth going forward but for lower and middle income countries as a whole as a whole there may well be. The question is whether politics, especially in advanced economies and China, might be able to sustain such an increase in trade. Recall that politics in advanced countries is moving defacto in the direction of seeking and forcing lower trade-GDP ratios.
B. THWARTED STRUCTURAL TRANSFORMATION: GOOD GROWTH AND SUSTAINABLE GROWTH
Successful development requires two kinds of structural transformations:
1) a shift of resources from low productivity to high productivity sectors (as highlighted by Sir Arthur Lewis); and
2) a larger share of resources devoted to sectors that have the potential for rapid productivity growth.
In many cases, however, resources do not shift in this way. They shift instead from informal, low productivity sectors to ones that are marginally less informal/more productive. These are cases of “thwarted structural transformation”.
Rodrik (2015) identifies manufacturing as a critically important sector for ensuring successful transformations. This sector exhibits unconditional convergence toward the world frontier, so that it can become an escalator for rapid growth – if countries manage to get on to it. This is why “premature de-industrialization,” the tendency for manufacturing in late convergers to peak at lower levels of activity and earlier in the development process, is such a cause for concern.
Are late convergers particularly vulnerable to thwarted transformation? To assess this, Rodrik’s identification of structural transformation with manufacturing is broadened. In particular, based on the detailed study of India by Amirapu and Subramanian, dynamic sectors are those with high levels of productivity and potential for unconditional convergence. Such a list comprises manufacturing, finance, telecommunications, and professional services. The Groningen database (Timmer, de Vries, & de Vries, 2014) is then used to do the shift-share analysis of Rodrik and decompose overall productivity growth into “good” (i.e., involving desirable structural transformation) and “less good” growth (e.g., in hotels, restaurants, transport, etc.). Therefore, good growth comprises growth accounted for by labor share shifts into these good sectors and their productivity growth. Therefore, good growth comprises growth accounted for by labor share shifts into these good sectors and their productivity growth. To motivate the argument before presenting the broader stylized facts, we compare good and less good growth in China and India since 1980. Interestingly, China’s good growth persists in both periods; India’s share of good growth declines in the second period. Both are of course positive outliers to the relationship itself, raising the possibility that while the general pattern is that good growth is necessary for sustained growth, China and India might defy this pattern. However, it would more prudent not to rely on permanent exceptionalism.
C. HUMAN CAPITAL REGRESSION
In some ways, there is one key difference between early convergence based on manufacturing and late convergence against the strong headwinds of automation and the globalization backlash. And that relates to human capital. In early convergence, it was the alignment of human capital endowment (educated but relatively unskilled labour) with the sector associated with structural transformation, namely manufacturing, that allowed for the percolation and spread of dynamism to the rest of the economy. Shifts in labor, the so-called Lewisian transformation from farm to factory, were possible because of this co-incidence: growth and structural transformation based on comparative advantage. Not only have they failed to provide even the basic education necessary for some structural transformation, that failure will prove increasingly costly because the human capital frontier for the new structural transformation has probably shifted further away. Technology will increasingly favor skilled human capital, where the requisite skills will include adaptability and the ability to learn continually. One might argue that growth itself will be based less on comparative advantage and more on some absolute human capital attainment.
D. CLIMATE CHANGE-INDUCED AGRICULTURAL STRESS
A final factor impeding late convergence relates to agriculture. It is often forgotten that Lewisian structural transformation required the release of resources into the modern sector under conditions of rising agricultural productivity. Part of the reason was the need to produce enough food to a growing population. That was only possible if agricultural labor productivity grew rapidly enough.
For the poorest, these growth rates have even declined post-GFC. For example, Indian agricultural productivity growth has been stagnant, averaging roughly 3 percent over the last 30 years. A later chapter of this Survey shows that Indian agriculture is vulnerable to temperature increase and still heavily dependent on precipitation. The analysis there shows that if climate change raises temperatures and the variability of rainfall, farmer revenues could decline by up to 20 percent to 25 percent in non-irrigated areas. For the late convergers, agricultural productivity is critical not just for feeding people but for ensuring human capital accumulation in those who move from agriculture to the modern sectors. Agriculture could yet come back to haunt the structural transformation fortunes of the late convergers.