In the Wake of the Crisis: Leading Economists Reassess Economic Policy

17. Do We Need to Rethink Growth Policies?

International Monetary Fund
Published Date:
March 2012
  • ShareShare
Show Summary Details

Dani Rodrik

The current economic crisis has taught us new things, but it does not require a complete rethinking of what we know about growth. The main new thing is that the context in which we are going to think about growth policies might be different. The context arises partly from the difficulties that the advanced countries are going to be facing with the debt overhang and possibly lower growth. What does that do to the growth prospects of the developing countries?

As we go forward, there are doubts about the system for cross-border financial flows, and there is a systemic worry about whether we are moving toward a world without a leader or without leadership where it will be difficult to sustain global cooperation.

One issue that is overlooked in discussions on growth is that the fundamental force that is driving growth in developing countries is convergence. There is a large gap between potential output (provided by the technology levels that already exist in the advanced countries) and actual output (based on the technology that developing countries currently have), and this gap drives development.

Another issue that is overlooked in discussions on growth is that this potential must be achieved rather than take place automatically. In other words, convergence is conditional. It is not automatic. It is going to depend on the things that economies do and get right.

The point about growth depending on convergence is important because it puts the focus on the supply side of things—on what countries do to absorb those technologies—and it downplays the demand side. Much of the discussion about growth is about whether developing countries can grow rapidly when the advanced countries are unable to. But regardless of how rapidly the rich countries grow, the convergence gap is still there. In fact, it is bigger than it has been in a long time.

To a first order of approximation, over the medium term the growth rate of the rich countries is largely irrelevant to the question of how much growth can occur in the developing and emerging markets. In other words, how rapidly the frontier is moving is of second-order importance relative to the gap between where the frontier is and where the poor countries are.

The bad news is that since convergence is conditional, it depends on policymakers’ having a good handle on what the right policies in the developing world are. Here there have been a succession of various “consensuses,” and now the consensus is perhaps best represented by The Growth Report: Strategies for Sustained Growth and Inclusive Development, which Michael Spence put together for the Commission on Growth and Development (December 2010). It is a consensus about pragmatism and about search and about context specificity and appropriate policies. It moves the discussion away from a list of specific dos and don’ts that were economists’ focus a decade or so ago.

Prior to the crisis, developing countries were growing rapidly, so one question is, “What does that rapid growth tell us about what is likely to happen in the future, and is that growth sustainable?” Much of the growth of Latin America and Sub-Saharan Africa prior to the crisis was misleading because they had experienced a long period of lagging behind and were making up for lost ground. On the other hand, that also means that the convergence gap between those parts of the world and the advanced countries is actually wider than at any time since the 1970s (figure 17.1).

Figure 17.1.A wider convergence gap (for most): Incomes as a ratio of high-income countries.

In figure 17.1, consider the ratio of per capita income in Latin America as a share of per capita income in the rich part of the world. The last decade in Latin America shows the process of convergence, but still the convergence gap between the average income levels in Latin America and in rich countries is wider now than what it has been since the 1970s. The ultimate dynamic, the potential for growth and catch-up and convergence, is larger now than at any time before.

That is also true for Sub-Saharan Africa, which experienced rapid growth in the last decade or so. But as you can see from the figure, relative to the frontier it is far below where the continent was earlier. Only Asia experienced sustained convergence in this period, so Asia is where the convergence gap has diminished.

What is convergence conditional on? What are its prerequisites? Here I develop the half of the argument that is relevant for this discussion. The economic component of convergence is a process of ongoing structural transformation. A lot of the growth in the developing world takes place through the creation of new industries at higher productivity levels and through a transfer of resources within those economies from the lower-productivity activities to the higher-productivity activities.

Tradables, particularly modern tradable industries, play an important role globally, and therefore safeguarding the health of modern tradable economic activities, monitoring the exchange rate, and enacting policies that promote tradables become important. From this perspective, a certain amount of financial deglobalization may not be bad news for developing and emerging markets because it allows them to maintain exchange rates that might be more competitive than would be the case otherwise.

The other thing that has been learned about the process of convergence is that sometimes we get the short-term economics right but we also need to get the medium-term politics right. The medium-term politics is really the building up of institutions of conflict management. Economies are consistently buffeted by a variety of internal and external shocks, and the ability to handle those shocks and bring about the macroeconomic and policy adjustments that those shocks require is a key determinant of whether growth spurts fizzle out. The key is whether domestic politics allows the appropriate responses to the shocks, and it is this that determines whether countries are able to engineer a succession of growth accelerations or experience only short-term growth that soon fades.

As economists, we have more to say on structural transformation, so I want to spend more time on it. An old concept in development economics is dualism, and this remains a key feature in the developing world. Developing countries have a mix of high-productivity and low-productivity activities, with large gaps in productivity levels across these activities. Consider the relationship between a measure of dispersion of labor productivity across different sectors of an economy and the level of development of that economy (figure 17.2). As the average level of labor productivity in an economy rises, intersectoral productivity gaps tend to shrink, approaching those of the rich economies. In the poor and middle-income economies, there are large gaps in labor productivity.1

Figure 17.2.Large productivity gaps in developing countries: The relationship between intersectoral productivity gaps and income levels.

The key implication of the structural transformation imperative from a policy perspective is that while the composition of output may be of second-order importance in a rich country, it is of first-order importance for economic performance and economic growth in a developing country. It is crucial for developing countries to achieve the right mix of economic activities.

Figure 17.3 compares agriculture’s labor productivity to productivity in the rest of the economy and shows what happens to this ratio over the course of development. There is a universal U-shaped relationship: the relative productivity of agriculture first falls and then rises. First, at very low levels of development, new industries need to arise. When a country is starting from a very low level of development, everybody works in agriculture, and there is no industry, so agricultural productivity is the same as productivity in the rest of the economy. Growth begins when new industries start developing, and therefore the relative productivity of agriculture starts to fall.

Figure 17.3.Growth from both new activities and ongoing structural change: The relationship between economywide labor productivity (horizontal axis) and the ratio of agricultural productivity to nonagricultural productivity (vertical axis).

This is the first dynamic, but over time, people shift from agriculture into the more modern parts of the economy. This drives the process of convergence within the economy, and the relative productivity of agriculture starts to catch up with that of the rest of the economy. Both dynamics are needed—new industries and a process of ongoing transformation (ongoing movement of labor and other resources from the old to the new). This is not much different from Sir Arthur Lewis’s model of dualism, where there is a quasi-automatic movement of workers from traditional industries to modern industries.

One of the most surprising things that I have seen in the last few decades is that in large parts of the world today, structural transformation is taking place in reverse. People are moving from high-productivity activities to low-productivity activities and not the other way around. Figure 17.4 shows the decomposition of overall labor-productivity growth in different parts of the world across different sectors. The light gray bar shows labor productivity growth on average within individual sectors. The dark gray bar shows how the reallocation of labor across different sectors has affected economywide labor-productivity growth—the structural transformation component.

Figure 17.4.Growth-enhancing structural transformation is not automatic: The decomposition of productivity growth by country group, 1990 to 2005.

In figure 17.4, Asia behaves in the way that all developing countries would be expected to behave. In China, India, and Thailand, labor is moving from agriculture and other low-productivity activities to high-productivity activities. But for the same period, growth-reducing structural change is taking place in Africa and Latin America. One explanation is that manufacturing is shrinking and informality is expanding. This is a part of the picture that we have missed by looking, for example, at how successful export-oriented manufacturing has become in Latin America. We have forgotten to ask what happens to the workers who are displaced from these firms that become more productive by rationalizing production, upgrading technology, and substituting capital for labor. These workers end up not in more productive activities but in less productive activities.

In the high-income countries, there are not large gaps in productivity, and therefore the intersectoral component is not large in that group of countries. Reversing this perverse outcome is critical if emerging markets in Latin America and Africa are going to generate ongoing sustained growth based on desirable structural transformation rather than based on high commodity prices or short-term capital-inflow-driven growth spurts (figure 17.5).

Figure 17.5.Conditional help from a large reservoir of excess labor: The association between the initial labor share in agriculture and the contribution of structural change to growth.

A couple of things seem to lie behind helping to drive this distinction between countries that are experiencing the right kind of structural change versus countries that are not. A country’s initial comparative advantage matters a lot. A country that starts out with a comparative advantage in natural resources and in primary products and that globalizes based on that comparative advantage specializes in activities that cannot absorb a lot of labor. Those kinds of growth models are not generating a lot of employment in the more productive parts of the economy. In other words, integrating into a world economy with a comparative advantage in natural resources is not conducive to the kind of structural change that drives long-term sustained growth (figure 17.6).

Figure 17.6.The bad news of a comparative advantage in primary products: The partial association between the share of primary products in exports and the contribution of structural change to growth.

But there are indications that the natural resources “disadvantage” can be offset with various policies. One policy is, once again, the exchange rate. Countries that have competitive exchange rates have more desirable structural change. Overvaluation is the enemy of growth-increasing structural change, and undervaluation is a help to achieving it. A second policy worth considering is labor-market policy. Countries with more flexible labor markets appear to be better at promoting structural growth and at increasing structural change than other countries are (figure 17.7).

Figure 17.7.Help from policy: The partial association between an index of currency undervaluation and the contribution of structural change to growth.

In conclusion, let me connect this perspective on growth with some of the systemic issues that I started with. First, the discussion should not be about which countries will be the growth engine of the world and whether developing-country growth will be adversely affected by low growth in the rich countries. This is a demand-driven, short-term perspective on growth and it is not the right way to think about medium- to long-term growth in the developing countries. The drivers of growth need to be analyzed from the supply side, and from the perspective of the mechanisms of engines of growth within the developing countries. That engine is structural change within the developing world, a process that is fraught with difficulties and requires policy support.

From the standpoint of multilateral institutions, the policy environment for developing countries needs to leave room for pursuing the appropriate structural transformation policies and for pursuing policies with the kinds of instruments that have the least adverse spillovers on the rest of the world. Developing countries need to have access to the policies that will help them change the composition of output and do it with the least amount of external spillovers.

The greatest failure here has been China and global imbalances. China has changed its composition of output through undervaluation policies that have spilled over into its external surplus. In principle, you can achieve the transformation of output that you need not through exchange-rate macropolicies but through sectoral microeconomic policies. That is, you can achieve transformation through trade and industrial policies that have direct effects on the structure of output and production (i.e., policies that alter relative prices) without altering the relationship between income and expenditure or the external balance. The international environment has pushed China to pursue the wrong kinds of policies. It has prevented China from using its trade and industrial policies because there is much greater discipline in the World Trade Organization on trade and industrial policies than on exchange rates.

Instead we need to move to a world where we have much greater discipline on currency policies because these create first-order spillovers across countries. Correspondingly, we also should have less discipline over trade and industrial policies so that we can reconcile the need for structural transformation in the developing world with the need to minimize macroeconomic imbalances.


These and the subsequent results on structural transformation I discuss below are taken from my work with Maggie McMillan. See Margaret S. McMillan and Dani Rodrik, “Globalization, Structural Change, and Economic Growth,” National Bureau of Economic Research Working Paper 17143, National Bureau of Economic Research, Cambridge, MA, June 2011.

    Other Resources Citing This Publication