The State of Food and Agriculture 2022

Automation can affect agricultural production and decent employment opportunities in various ways. In crop production, it becomes possible to expand cultivated land or improve yield per hectare, which in turn increases production. In livestock production, automation can improve labour productivity and substantially reduce drudgery by enabling workers to milk or feed animals with minimum manual intervention. Similar reasoning applies to fisheries, aquaculture and forestry (see Chapter 2); in the case of forestry, improved worker safety is an important additional benefit driving automation. All these benefits can result in major increases in welfare. If automation involves large economies of scale, widespread adoption among larger producers can sometimes put smaller producers out of business and precipitate consolidation in the agriculture sector. As agricultural labour demand decreases and new technologies make some skill sets obsolete, automation can displace workers, especially the poorest, who may struggle to find employment elsewhere. Appropriate policies, legislation and investments must be in place to avoid, mitigate or address the negative social impacts, especially for the most vulnerable.

The following sections consider the impacts of agricultural automation on employment in agrifood systems in different contexts. This report analyses the impact of agricultural automation under the particular lens of decent rural employment, defined as work that provides a living income and reasonable working conditions. Box 19 describes the standards of decent employment to be used to evaluate the impact of agricultural automation technologies.

The employment impacts of agricultural automation are difficult to measure, because they typically involve changes across agricultural production activities, as well as upstream changes deriving from changing demand for inputs, and downstream changes affecting transport and logistics, processing, distribution, and retail. As agricultural transformation unfolds, people leave agriculture to seek higher-paid jobs, and the share of people employed in agriculture continues to decline, as described in Chapter 1 (see Figure 3). The process reshapes labour supply and demand in all agrifood systems, as it affects the production, processing and distribution of food and other agricultural outputs. When all the nodes in agrifood systems are changing more or less simultaneously, it is difficult, if not impossible, to ascribe social impacts – such as changes in decent employment and the implications for gender, youth and small-scale producers – to specific incidents of agricultural automation. Understanding the transformation of agrifood systems is a fundamental step towards grasping its social impacts, especially on employment. Note that this chapter does not include the potential indirect effects of automation adoption (e.g. increased demand for researchers and scientists to develop and improve technologies), nor does it consider the possible economy-wide implications, which may also have significant social repercussions. How the comprehensive set of possible final impacts plays out in reality remains an empirical question and will depend on the specific circumstances in different countries and societies.

Figure 7 helps to illustrate two main points. First, the possible effects of agricultural automation are multiple, and impacts on farm employment are likely to be diverse. Demand for low-skilled labour – whether family or hired labour – is likely to decline as many tasks are automated. Automation of some tasks may resolve problems of labour bottlenecks, allowing production to increase, by either horizontal expansion or intensification. Automation is likely to increase the demand for relatively skilled workers who complement the new technologies. Second, the overall impacts of agricultural automation on decent employment within agrifood systems are likely to be very different from the impacts on individual agricultural business sites. Automation could easily reduce low-paying seasonal employment on farms but increase higher-paying, less seasonal employment upstream and downstream. The question is whether the positive social impacts of the increase in higher-paying, less seasonal work compensate for the negative impacts of the decrease in availability of low-paying seasonal employment for workers, allowing the latter to find alternative employment.

Employment seasonality is a concern in agriculture around the globe. Crop and livestock production activities are inherently seasonal. This means that unemployment and underemployment tend to be high in some seasons, while there may be severe labour shortages in others. For an agricultural producer, not having access to labour at critical times (e.g. during crop harvest and livestock shearing) can have serious ramifications for farm operations and may lead to losses or discourage cultivation altogether. Automation that eases excess labour demands during some seasons could, in theory, maintain employment in other seasons. This raises important questions. Which cropping tasks, in which seasons, are easiest to automate, and do they coincide with the labour shortages farms face? Conversely, what are the impacts for the poorest, unskilled workers who find themselves without a job once businesses start to automate and their skills become obsolete? Which policies can ensure a more productive, efficient, sustainable and inclusive automation process?

For the most labour-intensive crops – primarily fruits and vegetables – tasks occurring in the most labour-lean seasons are often the hardest to automate because of the potential damage to plants or fruits caused by machinery. By way of illustration, it is worth taking a look at automation in the richest agricultural areas, where farm wages are relatively high and automation solutions most available. In California, United States of America, land preparation is universally mechanized, including ploughing, tilling and land levelling. Harvesting of crops for use in processing (e.g. tomatoes or wine grapes) is automated. Harvesting of fresh fruits and vegetables for final consumption, however, still relies on manual labour and is harder to automate, even though produce-picking robotics solutions are on the horizon, incentivized by a shortage of harvest workers and rapidly increasing wages.

These new employment opportunities are appropriate for many kinds of workers. Drivers, warehouse workers, machine operators and mechanics all require little formal education, but they experience differences in terms of pay, job security and the job skills required.^{29, 30} Such jobs can also be seasonal, especially in small processing firms, but they may be stable if offered by large commercial processing firms. In both cases they are less seasonal than field jobs in agriculture. The vast majority are filled by men.^{31, 32} Office workers, salespeople and specialists requiring more formal education, training and experience, are the highest paid and typically include a higher proportion of women employees.³³

Implications for small-scale and subsistence producers

Implications for labour demand depend on the type of work and production. Subsistence producers operate their production units using family labour. They are often poor, food-insecure and with limited access to markets and services.³⁴ In the Plurinational State of Bolivia, up to 83 percent of small-scale producers are poor, compared with a national poverty average of about 61 percent. In Ethiopia, where 30 percent of the population lives under the national poverty threshold, the poverty headcount ratio for small-scale producers is 48 percent. In Viet Nam, more than half of small-scale producers are poor, while in the country as a whole only about 20 percent of the population lives in poverty. In such conditions, the higher rate of poverty among those involved in agriculture is caused – at least in part – by low productivity rates as they survive on subsistence or quasi-subsistence farming. If these farms adopt automation, they can raise productivity and improve incomes and livelihoods by expanding production, potentially becoming a family commercial farm. For example, availability of tractors for small-scale family farms in Zambia allowed producers to more than double their incomes, primarily by cultivating more land and applying more inputs (mainly fertilizer), increasing yields by 25 percent.³⁵ Adopting automation can free up time, to be used for other activities such as education for children, and lead to long-term economic benefits for households. It can also allow household members to find work in non-farm activities, where available.

Agricultural automation can also give access to higher-value markets and allow agricultural households to sign contracts with supermarkets or foreign buyers, provided their produce is of consistent quality and quantity. Participation in such high-value markets can bring significant welfare gains to agricultural households. In Kenya, supermarket contracts with small-scale vegetable farmers increased farmers’ household incomes by more than 40 percent and led to the largest reductions in multidimensional measures of poverty for the poorest households.³⁶ Farm households supplying goods to supermarkets have also exhibited significantly higher consumption of calories, vitamin A, iron and zinc.³⁷

Even in other regions of Africa, where labour is relatively abundant and fertility rates high, there is evidence that lack of agricultural labour limits production. Thus, automation offers the possibility to improve production and household income. A study of farm-level data from four countries in Eastern and Southern Africa justifies the current efforts to mechanize agriculture in Africa, as labour and other sources of farm power appear to be major factors limiting agricultural productivity in the region.³⁸

Many of the potential benefits of agricultural automation are neither immediate nor automatic. Small-scale producers and subsistence farmers lack the managerial and technical skills to benefit from the opportunities of agricultural automation. They also need to update and modernize their business models to align with prevailing market requirements and standards. This highlights the importance of building capacities and putting in place effective rural advisory systems that can ensure timely access to information on technologies and markets (see Chapter 5).

Implications for medium- to large-scale commercial production

Commercial family production units are owned and operated by family labour, but may also use hired labour (e.g. hired field workers, labour supervisors, contractors). Automation can reduce demand for all three types of labour, but can also induce producers to expand their operation. If family commercial producers choose to expand towards corporate commercial agriculture, family labour will most likely be replaced by hired professionals, including farm managers, sales personnel, machine operators and mechanics. If, as is often the case, technology adoption is spurred by rising wages and scarce labour, agricultural automation will tend to increase labour productivity and wages, in which case automation might enhance welfare for both producers and hired workers. However, automation can also displace workers, especially the poorest and least skilled, who will be forced to seek jobs elsewhere, possibly putting downward pressure on wages for unskilled labour as their skill set makes it difficult to find other jobs (see Box 20). Another possibility is that subsistence farms exit agriculture entirely due to technology adoption by commercial farms – so-called farm consolidation. In these cases, policies, legislation and investments must be in place to ensure that subsistence and small-scale producers, as well as low-skilled workers, are not left behind, but rather are able to reap the benefits of agricultural automation. It may be necessary to provide targeted social protection and training during the transition.

Corporate commercial farms employ all types of labour, except family labour. These farms are the most advanced and are generally automated to a significant level. They often have the economies of scale and capital to invest in more robotics technologies that may reduce considerably on-farm labour demand – with potentially negative consequences for workers, in particular low-skilled workers – or change the type of labour needed on the farm. For example, with digital automation, a former tractor driver may supervise a swarm of autonomous crop machines or retrain to carry out repairs. However, robots are not typically economically viable for most farms, unless labour is scarce. By way of illustration, although robotic milking technologies have been in commercial use for many decades, few dairies in the United States of America have adopted them, as farm labour is still relatively inexpensive.⁴⁰ In contrast, they have been in commercial use in Western Europe since the 1990s.

In general, if automation technologies are adopted where there is no labour scarcity, but because they are made artificially cheap (e.g. due to government subsidies), there is a risk of displacing workers and generating unemployment. Labour displacement can be costly for farm workers; the overall impact will depend on whether they can move to new jobs generated upstream or downstream (see Figure 7). On the other hand, agricultural technology adoption spurred by rising wages and increasing competition for scarce labour is likely to increase both wages and overall productivity, benefiting both producers and hired workers.

Automation on farms in high-income countries or regions within countries could have negative impacts on migrant remittances to poorer countries and regions. If demand for unskilled migrant agricultural workers declines, this could increase unemployment levels in migrants’ home countries and regions, as well as reduce remittance flows.⁴¹ In Brazil, automation of coffee harvesting has significantly reduced the demand for unskilled labour – mostly internal migrants from poorer areas of the country – but increased the demand for skilled workers.⁴² This calls for immediate, inclusive social policies to help unskilled workers who lose their jobs so they can find employment elsewhere.

Automation often seems to occur in the context of diminishing farm labour and rising wages in migrant-sending areas. Box 21 provides an example of how agricultural automation in the United States of America is being driven by a growing scarcity of labour in the migrant-sending communities of Mexico. Another study in the United States of America found that the automation of greenhouses increased the gross revenue of horticultural businesses, allowing them to pay higher wages and retain migrant workers longer, while hiring fewer new skilled workers.⁴³