Decentralisation and spatial (mis)allocation of irrigation water

Canal water – an important source of irrigation in India – often tends to be unequally distributed among farmers within a geographical area. Based on a survey in the state of Odisha, following the decentralisation of the management of canal irrigation systems, this article finds that the policy reform has positive impacts – but the details of the institutional design matter.

Canals are an important source of irrigation in India. A longstanding concern with distribution of water in canals is that, since a canal system covers a large geographic area, water distribution is highly unequal across farmers within the system. Several ethnographic fieldwork studies (Bromley et al. 1980, Chambers 1988, Wade 1982) and empirical research (Jacoby and Mansuri 2020) have highlighted the fact that farmers located at the head of the canal have the propensity to extract disproportionate amounts of water as they get the first access to it. This results in farmers located at the tail of the canal receiving too little. Since canal irrigation systems are typically operated by a central authority who allocates water at the source and have limited capacity for monitoring and enforcement across the entire system, the extent of spatial misallocation may be large.

In this context, in our recent research (Das and Dutta 2023), we examine the effect of decentralising the management of canal irrigation in the Indian state of Odisha on the distribution of water across farmers and their agricultural performance. We hypothesise that decentralisation would result in improvement in water distribution, increasing water availability to farmers located farther away from the canal.

Background and institutional details

The state government in Odisha initiated the decentralisation process by implementing the Orissa Pani Panchayat Act in 2002.¹ Following the enactment, the Water Resources Department of the Government of Odisha began forming locally elected canal management bodies, known as “Pani Panchayats” (PPs) in various parts of the state in a gradual manner. PPs are primarily responsible for water management and equitable distribution of water amongst the farmers within an area. They coordinate with farmers about water allocation and perform other duties such as cleaning canal embankments, and maintenance and construction of field channels to facilitate better flow of water to all farmers. The major source of funds for the PPs are through grants received from the state government. As of 2014, there are approximately 25,000 PPs across 30 districts in Odisha covering all types of irrigation systems.²

Each PP covers 300-600 hectares of command area of an irrigation project. It is divided into a number of jurisdictions which are referred to as outlet command areas or “chaks”. Each PP on average has about 15 chaks. Each chak has a “Chak Committee” composed of three elected members. One member is elected each from the upper, middle, and lower reaches of the chak. A representative, called the chak leader, in each of these committees is chosen by rotation to serve as a member in the executive committee of the PP.

The executive committee of the PP makes decisions about the operation and maintenance of the relevant part of the irrigation system that falls under their jurisdiction. The elected members of PP have a term of six years. All the farmers are members of the general body of the PP.

Survey and methodology

To collect data on agricultural outcomes, including water allocation, we conducted a survey of farmers and elected members of PPs in Odisha during April-June 2019. The survey covered 80 PPs, pertaining to canal irrigation across eight districts covering ten irrigation divisions. Figure 1 shows the surveyed districts in red.

Figure 1. Survey districts in Odisha

We had two sets of questionnaires in the survey – one for the elected members of PPs and the other for the beneficiaries or farmers. In each PP, two chaks were randomly selected and in each of these, nine randomly chosen beneficiaries were interviewed. In all, we interviewed 1,423 farmers and 562 elected PP members.

The data on water supply is self-reported by the farmers. This is because the Water Resources Department does not collect data on water allocation using water meters. The Department releases water in the canals in tranches. In a season, the department releases several such tranches of water. Each tranche of water ensures supply for about 7-10 days in the field. However, if there is overextraction by farmers at the head of the canal, some farmers, especially those located at the tail, may not receive water in a given tranche. The number of tranches of water received by a farmer in a season, therefore, can vary. We use the number of tranches of water a farmer receives in the primary agricultural season as our main variable capturing water allocation. We also ask the farmers about water availability, output and price 10 years prior to the survey. This allows us to create a ‘panel data’ (of these variables over time) at the level of farmers.³

Most parts of the state were ‘treated’ (subjected to intervention) with decentralisation at the time of our survey. However, due to the differential timing of the constitution of PPs, there is variation in the number of terms that PPs had experienced. Hence, we use the variation in the degree of treatment or the length of exposure to the decentralised institution as our treatment variable. The length of exposure to the new institution is relevant in our context because the transition from a centrally managed system maintained only by government officials to having locally elected councils responsible for canal management was stark. Therefore, it took some time for the farmers, and especially the engineers of the Irrigation Department, who lacked experience in conducting elections, to be familiar with the new institution and iron out issues with its implementation.

Major findings

Using survey data and comparing the farmers in the same PPs over time, we find that in absence of PPs, distant farmers receive less tranches of water than those located closer to the canal. However, in areas that experienced a larger number of PP terms, this negative relationship reduces significantly. Moreover, we get the same result in a more stringent ‘empirical specification’ where we compare the same farmer over time. Consistent with this robust result, we find that the distant farmers generate lower revenue (conditional on plot size) in absence of PPs. However, with longer exposure to decentralisation, revenue improves more for those farmers. Notably, the result is driven completely by price and not the quantity produced. This suggests that the constrained supply of water to the distant farmers negatively affects the quality of rice they produce, but it has minimal effect on the quantity of production. We also find that distant farmers are significantly more likely to buy land with each additional PP term being completed. This indicates that better agricultural performance, owing to the decentralisation of water management, may have led to greater profitability and savings, which allowed those farmers to increase their landholding.

Our results are more pronounced in PPs where land inequality is high. This could be because the ability of farmers to overextract water under centralised management of irrigation may depend on the extent of land inequality among them. For example, it could be that in PPs with more equal landholding, it is easier to ensure cooperation among farmers leading to less overextraction under centralised management. In villages with high land inequality, ensuring cooperation is likely harder. Therefore, decentralised management could potentially have a larger impact in those areas. Additionally, we find that our main result is driven by farmers who interact frequently with elected PP members. This suggests that it may be easier for dissatisfied farmers to complain to PP members about misallocation of water, since the PP members are locally elected and hence are likely to be more approachable than government officials. PP members are also more likely to be aware of the local context.⁴

Related literature

Our work relates to the literature on decentralisation of natural resource management. Baland et al. (2010) examined the effect of constituting local forest management bodies in India (known analogously as “Van Panchayats”⁵) as opposed to forests being managed by government bureaucrats, on forest conservation. They find positive results of local management. Somanathan et al. (2009) find that forests in the state of Uttarakhand, when managed by village councils, are seven times more cost effective as compared to those managed centrally by the state government. On the other hand, Jacoby et al. (2021) find that decentralisation of canal irrigation led to greater corruption and water theft in the Punjab province in Pakistan, resulting in the worsening of spatial allocation of water.

Prominent work by Elinor Ostrom (Ostrom 1990, Ostrom and Gardner 1993, Ostrom et al. 1994) has demonstrated that community institutions that locally manage common pool resources can be effective in the sustainable usage of natural resources. Bardhan (2000) has examined community management of irrigation projects in southern India to analyse the determinants of better cooperation in their management. We contribute to this literature by analysing a context where local management is facilitated through formal decentralisation and is implemented at scale across a large state in India.

Policy takeaway

Our results are in contrast with the findings of Jacoby et al. (2021), who find negative effects of irrigation decentralisation. An important reason for this differential finding may be that the decentralisation institutions differed in important ways across the two contexts. Jacoby et al. (2021) write that, in the context of Pakistan, the canal outlet-level chairman was selected using show-of-hands among the local landlords, which resulted in the locally powerful landlords getting selected as chairmen in most cases. Additionally, even though the local council was selected from among the outlet chairmen through secret ballot, the council members in many cases did not face effective electoral accountability due to legal challenges in conducting future elections. Both of these factors ensured that locally powerful farmers remained in control of the local canal management. This resulted in exacerbation of the overextraction problem post-decentralisation. In contrast, in our context of decentralisation in Odisha, local council members were always elected through secret ballot, and re-elections, though delayed in some cases, were not stalled anywhere. Importantly, as part of the election design, farmers from upper, middle and lower reaches of the outlet command area formed separate electorates, ensuring representation from the different reaches of the canal. Hence, the power distribution among local farmers in this context was more even, especially with respect to distance from canal. Therefore, the effectiveness of irrigation decentralisation may depend on the details of the institution.

Our overall results highlight that appropriately designed institutional reforms are an important way to address the inefficient and inequitable distribution of irrigation water in agriculture. Given that canal irrigation covers a significant share of irrigated land in India, the potential gains from such reforms, if extended to the rest of the country, are presumably large.

Notes:

The Orissa Pani Panchayat Act allows farmers to participate in the management of irrigation systems in the state of Odisha. Pani is a Hindi word for water.
The canal irrigation projects are classified into three categories based on their command area, that is, the area covered by one irrigation project. Projects covering land area between 40 and 2,000 hectares are called minor projects, the ones between 2,000 and 6,000 hectares are medium, and above 6,000 hectares are major irrigation projects.
The price data reported by farmers matches well with the wholesale market prices for the corresponding districts in those years, validating our data.
We further examine whether elected representatives invest more in local canal infrastructure (primarily in the form of constructing field channels) to facilitate better flow of water to distant farmers. However, we find weak evidence of this.
Van is a Hindi word for forest.