The benefits to society from job creation will not be typically taken into account when firms take investment and management decisions – they are what economists call a positive externality. Standard economics tells us that in such situations, there is role for ‘Pigovian’ taxes or subsidies to correct the mis- pricing of good jobs. One could see development finance institutions such as CDC as subsidising investment, to create more jobs than the market would deliver if left to its own devices. But patient public capital with the ability to tolerate lower financial returns than commercial investors, or wait longer to receive them, could also allow firms to choose modes of production and management that involve the creation of more, higher-quality jobs.
Some of the biggest names in economics now see job creation as a global priority. Professor Dani Rodrik of Harvard University says that until recently industrial policy would have concerned itself with factors such as knowledge spillovers and export competitiveness, and left jobs to other areas of government policy (such as education). He now believes that the creation of good jobs should be foremost amongst policy objectives, and that the labour market is the greatest market failure of them all.11 Rodrik and Sabel (2019) argue that good jobs are a source of positive externalities for society, analogous the better known example of environmental externalities: “We do not view this simply as a problem of inequality and exclusion, but also as a problem of gross economic inefficiency – a case of operating deep inside the production possibility frontier. That is because a shortage of good jobs is associated with a significant range of public ills.”
Creating better jobs
This section introduces a simple “accounting” framework for evaluating the impact of job creation on an economy’s overall employment. Despite its simplicity, this framework is deeply rooted in current state-of-the art labour economics.
After introducing the framework, we will use it to discuss the consequences of a lack of job opportunities, and to guide us through other effects of job creation, which are perhaps less apparent at first sight. In the next section, we extend the analysis to job quality.
1.1 A simple framework for understanding job creation
The evolution of total employment Nt over time can be written as the sum of employment in the previous period Nt-1 and net job creation, or the difference between gross job creation and destruction JCt – JDt:18
Nt= Nt-1 + JCt – JDt, (1)
Using (1), we can then express unemployment, or the number of people outside of formal employment, simply as Ut=Lt-Nt, where L is the labour force. In the context of developing economies, we would interpret U as composed of people in low productivity and precarious informal employment, rather than unemployment of the type that exists in rich countries with welfare states.
Importantly for our discussion, the labour force in developing countries is growing rapidly. We will capture that in our framework by a constant growth rate g, i.e. Lt=Lt-1(1+g). Putting all these features together, we can write the following expression for the number of people outside formal employment
Ut = Ut-1 + gLt-1 + JDt -JCt (2)
Once again, the above expression is very intuitive. For the number of people outside formal employment to decrease, job creation must not only exceed job destruction, but also exceed the speed at which the population is growing.
An important channel through which population growth reduces living standards operates through its effects on agriculture and the availability of land. More than 70% of the poor in sub-Saharan Africa live in rural areas and derive more than half of their livelihood from farming (Muyanga and Jayne, 2014).
Across Africa, the typical (median) farm is getting smaller as land is sub- divided, a problem compounded by deteriorating soil quality from intensive farming (Barrett et al., 2017). Increasing rural population density is associated with lower rural wages and higher food prices.20 The relationship between farm size and productivity is complicated, and studies often find the smaller more intensively farmed plots have higher yields, but at some point farms become too small. Currently, about 15 percent of the sub-Saharan population resides in areas where farms have already become too small, as judged by yields (Muyanga and Jayne, 2014).
Because rural employment is growing slowly, continued population growth in already densely populated rural areas leads to increased rural-urban youth migration. Unless urban employment opportunities can keep pace with rural population growth, the youth are at risk of falling into precarious informal employment there too, with incomes barely above subsistence. Jedwab and Vollrath (2014) have identified Malthusian dynamics within poor megacities (over 12 million inhabitants) where living standards are low and stagnant. These cities are experiencing internal population growth, in addition to migration from rural areas, and negative congestion effects are outweighing the positive agglomeration effects that have historically meant urbanisation has led to higher living standards.
This simple expression for the evolution of employment should not be interpreted as suggesting that gross job creation and destruction are independent variables, as if some firms create jobs and other firms destroy them, and there is no link between the two. Below we will use a simple framework to introduce some job displacement mechanisms, but for now a simple point can be made: supply-side constraints—the fact that there is only a finite number of workers—combined with the fact that gross job creation flows are large, implies the existence of mechanisms that connect job creation to worker reallocation and job destruction. To take the extreme case, in an economy at full employment it is impossible to create a job without another being destroyed elsewhere and creating one job may entail many workers switching jobs and reallocating across firms.
In economies with unemployment, the flow of gross jobs created each year is so large that unemployment would be quickly eliminated if it was not offset by job destruction.21 It is perhaps not widely appreciated how large gross flows are, in comparison to net changes in employment.
Few countries in Africa and South Asia produce reliable labour force data, making it hard to put a scale on the size of these flows in the countries that CDC invests in, but we know that across countries gross labour market flows dwarf the net changes in employment. In Ethiopia, a country with a growing economy and rapidly growing labour force, Shiferaw and Bedi (2013) find rates of job creation in manufacturing firms of 14% (above the OECD average), with lower job destruction rates at around 10%, so net employment in the sector grew on average at around 4% per year.
Scars of youth non-employment
Rapid population growth creates a workforce highly skewed towards the young. The median age in Africa is just 19.4 years and the share of individuals younger than 15 years is an amazing 43 percent (United Nations, 2018). This compares to Europe’s median age of 41.6 with 16 percent of the population younger than 15 years.
There is ample evidence from developed economies that youth unemployment has a long-term scarring effect (a lack of data prevents comparable research in low-income countries). For instance, Gregg and Tominey (2005) use survey data from the UK and find a large and persistent wage penalty from youth unemployment, after controlling for education, region and other characteristics, in the region of 13–21% lower wages at age 42.
These persistent effects are shown to stem not only from lower initial earnings, but also from the acceptance of lower quality jobs (accounting for almost half of the earnings losses). These results may not map exactly onto developing economies, where most individuals cannot afford to be unemployed, but the idea that having to accept low-quality jobs when young, because of a lack of better alternatives, may have persistent effects seems plausible in developing economies. We will return to informality and its consequences in the next section.
What this means for DFIs
The fact that labour markets exhibit a great deal of worker reallocation highlights the need for a more complex view of the effects of job creation. It tells us that the impact of an investment will extend beyond those individuals hired by the firm, many of whom might have come from a similar job and therefore experience a small change in their own quality of life. Impact evaluation exercises that confine themselves to studying the effect of job creation on workers hired by the firm a DFI has invested in will miss the full impact. Most impact evaluation methods look for an “effect of treatment on the treated” and it is much harder to gather evidence on the general equilibrium impact of individual investments at the market level. The presence of spillover effects also suggest that many more workers may sometimes benefit from an investment if it creates the opportunity for them to move up the job-quality ladder. That result emerges from formal models of labour markets in low income countries, such as Basu et al. (2018), which finds that increasing high wage employment reduces the number of workers stuck in involuntary low productivity self-employment.
Higher quality jobs can also be associated with more on the job learning and creation of transferable skills, and workers gaining experience then leaving to set up their own business can also be an important indirect effect of creating more jobs at top of the employment ladder.
What type of firm should a DFI with the objective of poverty reduction invest in? DFIs could seek out firms that will hire workers from the poorest and most marginalised communities directly, or they could invest in more productive firms that are likely to mostly hire relatively well-educated or experienced workers, but which will create these knock-on effects that could indirectly cause people to move from precarious self-employment on to the first rung of the jobs ladder. Investments at the bottom of the ladder, so to speak, may benefit fewer people if they do not initiate this process of reallocation.
Increasing the number of more productive firms offering better jobs seems necessary to transform poor low-productivity economies, but on the other hand we cannot be confident these positive spillovers will always reach the poorest.25 We know that whilst economic growth is associated with poverty reduction on average, it is by no means sufficient.26 Beyond a few generalities, such as that investment in the extractive industries has little impact on poverty, there is little evidence to guide DFIs on this question.27 It is difficult to trace the spillovers from individual investments, empirically. But we can say for sure that the question of whether DFI investments are reducing poverty is not always answered by looking to see whether the firms that they invest in are hiring poor people.
DFIs face a similar dilemma when it comes to job quality: should they invest in the expansion of firms that offer higher quality jobs, or target firms that offer poor quality jobs (which are more likely to employ poor people) and try to raise their quality? The answer will depend on the strength of positive spillovers, and the ability of DFIs to work with firms to raise standards. Because DFIs might have more impact by working to improve conditions offered by low quality employers, evaluating the performance of DFIs with respect to job quality will require more than looking across DFIs’ portfolios and counting the proportion of jobs that are rated as decent.
We have seen how job creation can spur large worker reallocation. But we cannot assume that ‘knock-on effects’ will always reach down and lift the least productive. Labour markets can be segmented, either geographically or by skill level, and these spillover effects may peter out. There is relatively little empirical evidence on the circumstances that determine the strength of “trickle-down” effects. Research that estimates growth-poverty elasticities shows how the relationships between investment and poverty reduction in aggregate vary across space and time. A study by the WTO (2009) found that economic growth arising from openness to trade could not be relied upon to automatically pull workers out of the informal sector. However, the fact remains that the informal sector does tend to shrink as countries grow. La Porta & Shleifer (2014) argue the process of development amounts to growth in the formal sector leading to the decline of the informal sector in relative and eventually absolute terms. They show few informal firms convert to formality, but more generally they disappear because they cannot compete with the much more productive formal firms.
These aggregate results tell us something about what to expect on average, but they do not give much insight into when and why job creation in one place will initiate chains of reallocation that eventually benefit the worst-off sections of society. The evidence from more microeconomic studies is mixed. Again, data availability constrains which questions it is possible to answer, so most relevant studies are in advanced economies. Hornbeck & Moretti (2018) use very rich data from the USA to show that local productivity growth in manufacturing reduces local inequality, as it raises earnings of local less-skilled workers more than the earnings of local more-skilled workers. However, it is the local housing market that really determines who benefits: landlords tend to capture rising wage from renters. On the other hand, Lee and Rodriguez-Pose (2016) find no such effect of high-tech growth on poverty in US cities.
“The fundamental impulse that keeps the capital engine in motion comes from the new consumers’ goods, the new methods of production and transportation, the new markets… [The process] incessantly revolutionizes the economic system from within, incessantly destroying the old one, incessantly creating a new one. This process of Creative Destruction is the essential fact of capitalism.”
The previous section finished on a somewhat sombre note that formal sector job creation may be offset by job destruction, implying that it will not always pull people out of precarious informal employment. In this section we move away from analysing the transitions that workers make to looking at the Schumpeterian process of creative destruction at the firm level and how this process affects overall productivity in the economy. There are unproductive firms in both the informal and formal sectors that need replacing.
Research on developed economies has brought about several “stylized facts” which are associated with the process of creative destruction (see Bartelsman, Haltiwanger and Scarpetta, 2004). These include:
It is not obvious, however, whether these facts also remain to hold in developing economies where firms typically have a harder time expanding, even if they manage to survive market pressures. For instance, Aterido et al. (2019) show that the South African economy is extremely sclerotic, dominated by large firms lacking in dynamism. Rijkers et al. (2014) show that small firms in Tunisia stagnate and grow considerably less than their counterparts in developed economies. More importantly still, there is only a very weak correlation between productivity, profitability and job creation. This suggests that the productivity-enhancing process of creative destruction may be obstructed by certain factors. Hsieh and Klenow (2009) suggest that a lack of dynamism and consequent inefficient allocation are related to constraints on obtaining financing, Bloom and van Reenen (2007) relate them to management practices, poor legal institutions and the high prevalence of family-owned businesses. As a result, compared to developed economies, small firms are more plentiful and the primary source of job creation in poor countries.
Going back to our Figure 4, we can immediately see that even in the “bad-case” scenario when the lowest productivity firm does not manage to refill its vacant job, the economy is still better off. And the reason for this is that the composition of existing jobs (recall that in this case the number of jobs has not changed, and net job creation is therefore zero) has shifted towards more productive firms.
This consideration would provide a motivation for supporting job creation at high-productivity firms, but we must ask the question of who benefits. Again, we cannot simply assume that such investments will create inclusive growth. It is an incontrovertible fact that in some countries the benefits of investment and growth have flowed to the better-off, with most workers suffering from stagnant real incomes over protracted periods. The introduction of new labour-saving capital investments or automation may tip the scales towards job destruction and the net overall effect may be negative. That said, existing evidence suggests mainly that the introduction of process innovations has a positive overall effect on employment (see e.g. Baffour et al., 2016 for evidence on Ghana).
Thus far we have shown how DFIs can contribute to creative destruction by investing in more productive firms, setting off a chain of events that ends with the least productive firms exiting, or workers ceasing the least productive activities (informal self-employment).
Replacing unproductive units with more productive ones is certainly a contribution to economic transformation, but some investments may also have more transformation effects with impacts that ripple out across the economy.
If we think of an economy as a production network, we could think of a non- transformative investment as affecting a small part of the network – a single chain of worker reallocation, or one firm entering and another exiting. At the other extreme, an investment that meaningfully reduces transportation costs across the economy, for example, could cause the entire network to rearrange itself, many firms to enter and others exit, and new connections to be made.
Paddy Carter cdcgroup.com/insight