Chapter 3. Projecting Output, Employment, and Productivity Forward through 2020
- Louise Fox, Alun Thomas, and Cleary Haines
- Published Date:
- April 2017
This chapter uses the previously developed output and employment projections and analyzes what they mean for medium-term structural transformation in sub-Saharan Africa. The methodology for the projections is described in Fox and others 2013; the projected output and employment structures are briefly reviewed here.
The projected sector output distributions reflect a pattern of continued strong growth in all country groups, with a sharp decline in the agricultural output share for all except resource-rich countries (Figure 12). The resource-rich and lower-income countries are projected to maintain their recent growth patterns of above 6 percent per annum growth, with lower-middle-income countries not far behind. The industrial sector is projected to grow at 7 percent per annum in low- and low-middle-income countries. This shows some optimism about output transformation (for example, growth in sectors such as construction and agro-processing), but it also reflects resource discoveries coming online. In the resource-rich countries, the service sector is projected to grow very rapidly, with agriculture remaining fairly stable.1
Figure 12.Sub-Saharan Africa: Sectoral Share of GDP by Country Type, 2005–20
Source: IMF, African Department database 2013.
The projected employment distribution corresponding to the output growth pattern shows little change from the current structure (Figure 13 compared with Figure 4).2 With the majority of new jobs created in countries currently classified as low income (such as Democratic Republic of the Congo and Ethiopia), the agricultural sector remains important for creating employment. Stronger growth in other sectors could push this estimate down slightly, but it is unlikely that the labor force in agriculture will shrink over the next decade—young people seeking jobs will simply have no other option. If African agriculture realizes its potential, however, agricultural jobs could be more productive, higher-earning jobs.
Figure 13.Sub-Saharan Africa: Estimated Distribution of Employment by Country Type and Sector, 2020
Sources: Authors’ estimations.
A major factor in the slow-moving employment distribution is the very high growth rate of the labor force. Indeed, the share of industrial wage jobs in total employment rises only from 2.3 to 3.2 percent because the jobs are growing from such a small base relative to the projected increase in the labor force. Moreover, even though the industry growth rate is promising over the next decade among LICs, the growing importance of resource exports in this sector means that the share of industry wage employment in the total remains at less than 3 percent. The sector that is growing fastest among LICs is household enterprises, with the share projected to rise by 4 percentage points to 22 percent of employment. This expansion is mainly in the service sector (reflecting the output changes shown in Figure 12).
Taking the output and employment estimates, projected productivity changes are analyzed for low- and middle-income countries. The median estimate of labor productivity growth for LICs over 2010–20 is 2 percent per annum and varies between a low of 1 percent per annum for Benin, Mali, and Senegal to about 4 percent per annum for Rwanda and Sierra Leone. By historical standards for sub-Saharan Africa, this is very ambitious, although given the recent growth performance, it might be possible.
A major element of structural transformation is the movement of workers from low-productivity to more productive activities. The split in total growth in labor productivity between within-sector productivity movements and changes in employment composition can be measured by the Shapley decomposition. This method splits the change in labor productivity between the change in sectoral labor productivity at initial employment shares (the first term in the following formula) and the change in the employment shares at the levels of productivity at the end of the period (the second term):
where Δ Y is the change in aggregate labor productivity between the start and end point, Δ yi is the change in sector i productivity, and θ i is the employment share in sector i.
While some might view these employment projections as pessimistic, analysis of the employment structural change underlying the projections reveals that they are actually highly optimistic. Using the formula reveals that about 63 percent of the expected average productivity change is accounted for by with-in-sector productivity changes and 37 percent by movements across sectors (structural transformation in employment). Not only is this figure higher than the historical pattern in Africa, it is also considerably higher than that found by MR for east Asia between 1990 and 2005. They calculate that the between-sector productivity movements for Asia accounted for only 15 percent of the total productivity gain (estimated at a very high 3.9 per annum) over the 1990–2005 period. Similarly, Saccone and Valli (2009) have found that China’s development has had little to do with reallocation across sectors, while for India, the structural change component is 33 percent, similar to the estimates enumerated here for the sub-Saharan African countries.
Therefore, while there remains considerable leeway to raise productivity growth over the medium term in sub-Saharan Africa, analysis of the projections here shows that there are limits to how much can be achieved through labor reallocation, if that reallocation puts more employment in the service sector and the household enterprise sector, rather than in industry as in east Asia. According to Rodrik (2015), today industrializing countries can expect the share of employment in manufacturing to peak at about 15–18 percent of total employment. But no low- or low-middle-income sub-Saharan African countries have a manufacturing employment share anywhere near that; in most countries the share is below 5 percent. In addition to attracting the private investment needed to grow the manufacturing sector, sub-Saharan Africa must also strengthen efforts to raise within-sector productivity in order to continue a strong growth performance into the next decade.
How would relative employment and productivity change amid an environment of overall productivity gains and employment shifts? Overall, relative employment data points are located in the “transformation” quadrants—the lower left for agriculture and the upper right for the others (Figure 14). But most of the employment transformation shifts come from services again. The corresponding figure for the relative change in productivity shows that, except for outliers, the change in agricultural productivity hovers around zero while the change in productivity for industry and services is generally negative, reflecting high labor absorption (Figure 15). The industry outliers are mostly resource economies where an output surge in the mining sector generates little employment and raises sector average productivity.
Figure 14.Sub-Saharan Africa: Labor Productivity and Changes in Employment Shares: 2005–20
Sources: Fox and others 2013; and IMF staff estimates.
Figure 15.Labor Productivity and Changes in Relative Productivity Level, 2005–20
Sources: Fox and others 2013; and IMF staff estimates.
In sum, the fairly static transformation picture shown in Figures 12 and 13 actually reflects a very optimistic underlying dynamic. But this dynamic produces little net effect because over the past decade and a half, the population has been getting younger and the labor force has been growing rapidly. So, although strong output growth is projected to continue in many sub-Saharan African countries, the net effect will still leave most of the labor force in low-productivity employment.
These projections were prepared before the collapse of oil prices in 2014–15. Overall growth will most likely be lower in the resource-rich countries, reducing the growth of public sector wage employment in services, with knock-on effects on earnings in the household enterprise sector. The structure of employment was not expected to change much, however, so the employment projection is still valid.
As discussed in Fox and others 2013, an elasticity model was used to generate these employment projections.