The agricultural productivity gap: Informality matters

There are various explanations in the literature for the observed productivity gap between agriculture and other sectors, in developing countries. Using Indian data, this article questions the standard view of the two-sector productivity gap. It shows that while the productivity gap between the farm sector and informal non-farm sector is negligible, a worker in the formal non-farm sector is 3-4 times more productive than a worker in the farm sector.

As countries develop, the share of agriculture in labour and output declines declines – a stylised fact known as ‘structural transformation’. An associated feature of this process is that agriculture’s share in labour remains above its share in output, for all except the high-income countries.

A case in point is India. Agriculture’s share in GDP (gross domestic product) today is only 15%, while it still accounts for as much as 45% of the workforce. This implies that a worker outside of agriculture is about five times more productive than a worker in agriculture. It would seem, therefore, that ‘too much’ labour is locked into agriculture.

The productivity gap is not unique to India. McMillan and Rodrik (2011) leveraged this feature to demonstrate the potential of shifting labour from low-productivity to high-productivity sectors to substantially raise aggregate productivity and growth in developing countries. The implication is that countries ought to sort out the frictions and barriers that come in the way of structural transformation.

However, another strand of the literature questions the interpretation that labour is misallocated across sectors. The alternative explanation is that labour sorts itself into high and low productivity sectors (Young 2013, Herrendorf and Schoellman 2018, Alvarez 2020, Hamory et al. 2020). This implies that there are no large gains from reallocating labour from the farm to non-farm sector. Consistent with this view, these studies find only modest wage gains to those who switch occupation from one sector to another.

On the other hand, there are studies that find large returns to migration across sectors (Beegle, De Weerdt and Dercon 2011, Bryan, Chowdhury, and Mobarak 2014, Imbert and Papp 2020). Recent surveys of this literature point to a middle ground and assess a role for both sorting and labour mobility frictions in accounting for the APG (Lagakos 2020, Donovan and Schoellman 2021).

In our research (Jat and Ramaswami 2024), we address the question of whether the observed APG is driven by high-productivity large firms that are numerically small but economically substantial. At the same time, we know that non-farm sectors in developing countries are numerically dominated by small enterprises. We pursue the implications of such heterogeneity in the non-farm sector for the APG in India by utilising the distinction between formal and informal segments.

Extent of informality

Internationally, there is no one single definition of the informal sector. Typically, the informal sector is thought to consist of enterprises that are not registered with the government, that keep no records, pay little or no taxes, offer no employment benefits, and do not follow government regulations such as on labour and work safety (La Porta and Shleifer 2008). For 2016, the International Labour Organization (ILO) estimated that informal employment accounted for 73% of non-agricultural employment in low-income countries, 59% in middle-income countries, and 17% in high-income countries (Bonnet, Van and Chen 2019). Based on World Bank Surveys, La Porta and Shleifer (2008, 2014) estimate that informal firms may account for 35% of GDP in low-income countries. They also report large productivity differences between formal and informal firms.

In the Indian context, informal enterprises accounted for 43% of non-farm GDP in 2017 (Murthy 2019). In the same year, 68% of all non-farm employment was informal (Nagaraj and Kapoor 2022, Murthy 2019). This is strikingly similar to the disparity between agriculture’s share of employment and its share of GDP. At first glance, it would, therefore, seem that APG could depend on whether the farm sector is compared with the formal non-farm segment or with the informal non-farm segment. These impressions therefore merit a deeper investigation of the APG.

Methods

One approach is to compare wages across sectors. To account for self-selection of workers, identification of wage gaps rely on individual-level panel data that captures migration across sectors (for example, Herrendorf and Schoellman 2018, Alvarez 2020, Hamory et al. 2020). Such data are not available for India; even if it were, its coverage would be incomplete as much of the workforce in the informal sector is self-employed and does not report wage data. Indeed, La Porta and Shleifer (2008) use the percentage of the non-agricultural labour force that is self-employed as an indicator of informality. In India, the self-employed accounted for 43% of male employment and 51% of female employment in the informal economy in 2004 (National Commission for Enterprises in the Unorganised Sector, 2008).

Instead, we follow the approach of Gollin, Lagakos and Waugh (2014) by comparing the average output per worker across agriculture, the informal non-farm sector, and the formal non-farm sector. Like them, we too adjust these comparisons for differences in human capital and in days worked. Agricultural workers have lower human capital. They also work less – presumably because of seasonality. However, unlike them, we also adjust for differences in the labour share of value added. The formal segment consists of larger production units (by employment) and are typically associated with greater access to credit and greater use of capital. It would, therefore, be unwise to assume that labour shares in value added are equal across sectors.

We use the KLEMS (capital, labour, energy, materials, and services) dataset available from the Reserve Bank of India (RBI) to obtain the value added per worker in agriculture and in 24 non-farm sub-sectors. We use the National Sample Survey Organisation (NSSO) employment surveys to figure out the proportion of informal employment in these sectors. Our analysis is conducted for the years 1999-2000, 2004-05 and 2011-12. It turns out that every non-farm sub-sector contains a formal and an informal component; however, the proportions vary across these sub-sectors and we can identify the sub-sectors that are primarily ‘formal’ and those that are primarily ‘informal’. In a second approach, we use the same data to non-parametrically estimate, by sub-sector, the relation between informality and the APG.

Findings

Table 1 presents the productivity gap of agriculture relative to the formal and informal non-farm sectors. The gap is presented as a ratio of the formal (informal) non-farm sector productivity relative to agriculture. Values greater than one indicate a sector is more productive than agriculture. We find a negligible productivity gap between the farm sector and the informal non-farm sector, but a worker in the formal non-farm sector is between 3 and 4 times more productive than a worker in the farm sector.

Table 1. The agricultural productivity gap

Sector	1999-2000	2004-05	2011-12
Primarily informal non-farm sector	1.35	1.58	1.09
Primarily formal non-farm sector	3.33	4.37	4.10

Source: Authors’ computations, based on data from KLEMS, Indian Human Development Surveys, and Employment surveys of NSSO.

The ratios in Table 1 are computed after correcting for sectoral differences in effective labour input (human capital and hours of work) and sectoral differences in the share of labour in value added. Without these corrections, the productivity gaps are much higher – the formal sector is 15-20 times more productive, and the informal sector is 2.5-3 times more productive than the agricultural sector. In the comparison with the formal sector, all of these corrections matter – as agriculture is more labour-intensive, its workers possess much lower human capital, and they work fewer days per year. In the comparison with the informal sector, differences in the labour share of value added and in human capital are small. The largest difference is in the number of days worked.

To obtain the statistical significance of our finding, we regress the productivity gap of each of 24 non-farm sectors in the KLEMS database on that sector’s proportion of employment that is informal. The non-parametric regression is estimated using data pooled from 1999-2000, 2004-05 and 2011-12. Figure 1 illustrates this relationship.

Figure 1. Agricultural productivity gap and employment proportion in informal sector

Source: Authors’ computations based on data from KLEMS, Indian Human Development Surveys, and Employment surveys of NSSO.

Note: The figure displays a 95% confidence interval (CI). A 95% CI means that, if you were to repeat the experiment over and over with new samples, 95% of the time the calculated CI would contain the true effect.

The estimated function is downward sloping – the APG declines as the proportion of informal employment increases. The null hypothesis of perfect mobility between agriculture and the non-farm sectors is rejected whenever the percentage of informal employment is less than 83%. On the other hand, the hypothesis is not rejected for sectors where the proportion of employment that is informal is greater than 83%. Depending on the year, these sub-sectors account for 40-50% of all non-farm employment.

Conclusions

Our findings question the standard view of the two-sector APG. The informal non-farm sector is not much more productive than the agricultural sector. A worker in the formal non-farm sector is substantially more productive than an agricultural worker or a worker in the informal non-farm sector.

The small or negligible productivity gap between agriculture and the informal non-farm sector is consistent with free labour mobility between these sectors. It should be noted that much of the uncorrected productivity gap between them arises because of greater hours of employment in the informal sector. Therefore, even if the corrected productivity gap is low, the seasonality of agricultural activity means there is an employment gap and that may drive seasonal migration.

While our findings are specific to India, it may have wider applicability because of the substantial presence of informal segments in many low-income countries (Bonnet, Van and Chen 2019). Since small, relatively unproductive unincorporated enterprises are characteristic of the typical developing country (La Porta and Shleifer 2014), the findings here suggest that similar results may obtain for other countries too. Our analysis supports the view that the dualism in developing economies is primarily between its formal and all of its informal components – including agriculture (La Porta and Shleifer 2014).