Most of the advanced economies of the world have been deindustrializing for decades and have moved into a new ‘post-industrial’ phase of development. This post asks how they got there.

Although there are many idiosyncrasies in trends, the evidence shows that as rich countries developed, the relative contribution of agriculture to both total employment and GDP declined, while the contribution of services rose in both respects. In the middle, the relative importance of the manufacturing sector first expanded and then contracted. In other words, the data shows there was first a period of industrialization as production shifted from agriculture to manufacturing, followed by a period of deindustrialization as production shifted from manufacturing to services.

In this blog post we discuss this transition and the drivers of this process of ‘structural transformation’.

The analysis in what follows is based mainly on a considerable update of the data provided by Herrendorf, Rogerson and Valentinyi (2014).¹ You can read more about how we updated the data by consulting the 'Sources' tabs in the charts below.

Changes in employment

One way to study the process of ‘structural transformation’ across countries is to track how employment changed among sectors in the economy. The following chart shows this for ten of today’s rich countries. You can change the displayed country and see that the overall picture is very consistent.

Click to open interactive version

In all these countries the share of workers employed in agriculture has been going down, while the share of those employed in services has been going up; and the manufacturing sector first increases and then decreases in relative importance.

By clicking on the chart settings, you can plot the total number of workers by sector. This shows a different picture, since we can see that the labor force in these countries has been growing substantially. As we can see, despite the fact that the share of employment in agriculture tends to go down constantly in all countries, in many cases, the total number of workers in agriculture first increases and then decreases. The US is a clear example of this pattern: the total number of people employed in agriculture peaked more than a century ago around 1911, before beginning to decline.

Changes in economic output

A second way to study the structural transformation is to track how output changes across sectors in the economy. The following visualization shows this by plotting the evolution of the sectoral composition of GDP between 1800 and 2010 for the same rich countries included in the previous chart.

Click to open interactive version

We can notice here the broad pattern outlined before: as rich countries developed, the share of GDP in agriculture declined, while the share of GDP in services went up; and in the middle, the relative importance of the manufacturing sector first expanded and then contracted. Indeed, this broad pattern is followed by all countries in the sample. (An exception is South Korea, where manufacturing has stabilized at around 40% of GDP since the early 1990s.)

This chart also shows us that in early industrializers such as the UK, Belgium, the Netherlands and France, the manufacturing sector already constituted over 30% of GDP in the first half of the nineteenth century. The USA exceeded that threshold only by the end of the century, while the other ‘latecomer’ economies reached it at the beginning of the twentieth century. The only exception is again South Korea, which was a largely agrarian country only 50 years ago, before it began an extremely rapid industrialization process—a process that has come together with a 30-fold increase in average incomes.

When comparing the sectoral composition among countries, one must keep in mind that the way we measure these shares is important. The chart below shows value added in current prices, which means that the figures correspond to estimates of output value at the prevailing prices when output took place. The picture would however look different if we had chosen to show value added figures in constant prices (i.e. estimating the value of output at the prevailing prices of a fixed point in time). In the technical note at the end of this post we explain in more detail why this is. The key point to note is that in constant prices, the pattern of ‘deindustrialization’ is less marked.

The forces at play

The charts above show that there is an unambiguous trend of the agricultural share of GDP and employment to fall over the course of economic development. However, this does not necessarily mean that agricultural output falls in absolute terms.

When we look at data on agricultural production in Sweden, for example, we find that agricultural output did not fall as the country industrialized. Despite the process of industrialization, agricultural output increased up to 1931, after which it fell slightly but then stabilized for many decades.²

Click to open interactive version

Given that the share of agriculture in total employment is decreasing throughout this period, labor productivity in agriculture was actually growing. In fact, in absolute terms it continues to grow even after agricultural output reaches its peak. This is shown in the chart, which plots the ratio between output and employment in agriculture.

A point that we need to keep in mind here is that this is an absolute increase in productivity. In relative terms the picture differs, since other sectors also increase productivity, and in fact tend to do so at a faster pace. Indeed, if we look at the charts of employment and output shares, we can see that throughout the whole period for which we have data, the share of workers in agriculture exceeds the share of agriculture in GDP. This attests to the well-documented fact that in comparison to the other sectors (i.e. in relative terms), agriculture tends to be the least productive sector in most economies.³

When taken together, the evidence tells us that the story of structural transformation is not one of a stagnant agricultural sector overtaken by dynamic manufacturing and service sectors; instead, what matters are the differences in productivity growth between sectors.

It is then important to keep in mind that 'deindustrialization' means that there is a loss in the relative importance of this sector. In absolute terms the industrial output might well go up as a country 'deindustrializes'; just as agricultural output might go up in absolute terms during periods of 'industrialization'.

Alvarez-Cuadrado and Poschke (2011) show that at early stages of development, technological improvements in manufacturing are the greatest contributors to ‘pulling’ workers out of agriculture.⁴ At later stages, the decline in the share of consumption devoted to agricultural goods, combined with continued improvements in agricultural technology, become the predominant force in releasing workers from agriculture, further contributing to the pattern of ‘structural transformation’.

Esteban Ortiz-Ospina and Nicolas Lippolis (2017) - “Structural transformation: how did today’s rich countries become ‘deindustrialized’?” Published online at OurWorldinData.org. Retrieved from: 'https://ourworldindata.org/structural-transformation-and-deindustrialization-evidence-from-todays-rich-countries' [Online Resource]

@article{owid-structural-transformation-and-deindustrialization-evidence-from-todays-rich-countries,
    author = {Esteban Ortiz-Ospina and Nicolas Lippolis},
    title = {Structural transformation: how did today’s rich countries become ‘deindustrialized’?},
    journal = {Our World in Data},
    year = {2017},
    note = {https://ourworldindata.org/structural-transformation-and-deindustrialization-evidence-from-todays-rich-countries}
}

Reuse this work freely

All visualizations, data, and code produced by Our World in Data are completely open access under the Creative Commons BY license. You have the permission to use, distribute, and reproduce these in any medium, provided the source and authors are credited.

The data produced by third parties and made available by Our World in Data is subject to the license terms from the original third-party authors. We will always indicate the original source of the data in our documentation, so you should always check the license of any such third-party data before use and redistribution.

All of our charts can be embedded in any site.