How much are farmers losing from inefficient irrigation?

By Agha Ali Akram

How inefficient is Pakistan’s water irrigation system? If you were to use traditional methods to measure that problem, you would find no issues with water allocation. But in recent research I conducted with Robert Mendelsohn,[1] we used more precise measures and found that farmers lose 13 percent of potential net revenue because of inefficiency in irrigation water allocation.[2] This has important policy implications for maximizing the value of Pakistan’s canal irrigation networks.

Context: Pakistan’s canal system

The bulk of the canal system in present-day Pakistan was originally built by the British colonial administration and land irrigated by these irrigation canals was originally allocated in the late 19th Century.[3] The canal system has a three-level design – primary, secondary, and tertiary canals (Figure 1). A primary canal is a major trunk from which secondary canals are branched. Tertiary canals branch off from secondary canals and are the level at which farmers withdraw water into their farmland.

Figure 1: Simplified schematic of the Hakra Branch Canal
Canal Schematic v2

A simplified schematic of the Hakra Branch Canal, with its three levels of canals: primary, secondary, and tertiary.

The system of water allocation at the farm level is called warabandi, which literally translates as “turns” (wahr) which are fixed (bandi). Each farmer gets a fixed time each cycle where they can open the gate to their farm and tap the water flowing in the tertiary canal.[4]

The canal in our study – the Hakra Branch Canal – is part of the Indus Basin Irrigation System (IBIS) in Pakistan. IBIS is a continuous-flow, fixed-rotation system with two major multi-purpose storage reservoirs, 45 major irrigation canal commands, and over 120,000 watercourses delivering water to farms.[5]

The difficulty of knowing the allocative inefficiency of irrigation water

Water allocation models throughout the world offer a common insight on what constitutes efficiency: If the value generated by an additional unit of water consumed is equal for all users within a basin, society can maximize the value of its water supply (in other words, the marginal value of water must be equated across users). [6] This basic insight applies to water allocations across farmers within a canal irrigation system as well.

For a long time, we did not know if Pakistan’s canal system lived up to this model because of poor data: Traditional surveys of agricultural production omit important farmer water consumption data. While traditional surveys often tell us how many irrigations farmers make, the amount of water per irrigation is often assumed to be the same across farmers or is measured crudely. For example, rather than measure it, farmers are asked the perceived depth of each irrigation they apply on their farms. This is a problematic measurement, partly because fields are not perfectly level so that depth would depend on where the farmer made their estimate of applied depth. And while there is every reason to believe that both the number of turns and the length of each turn may be the same along the canal, it is not clear whether the flow is the same at each farm-gate. Thus, it is very likely that these traditional measures are not accurate.

What we did differently

Through better measurement, we provide a more accurate calculation of water use within canals, allowing us to rigorously demonstrate the inefficiency of irrigation water allocation and quantify the loss. Specifically, we coupled in-season water measurement with a post-season production survey.

We collected actual physical measurements of surface water from a sample of farms spread across the Hakra Branch Canal. The canal water discharge was measured using a flow meter and standard stream measurement protocols during the 2012 summer growing season. GPS tracking captured the precise location of the measurements made along the canal system, making it possible to calculate the distance from the head of the canal system to the measurement sites along with expected conveyance losses. These in-season water measurements were followed by a full agricultural production survey of the farmers at each canal location once the summer growing season was over.

Our results

Armed with better measurements of water delivered to farmers, we could determine the efficiency of water allocation by testing whether the value of each additional unit of water consumed was equal across farmers.

We found that farmers who use more water (located at the head and middle of the canal) receive less value for each additional unit of water they consume (Figure 2). The marginal value of water is much higher for farmers at the tail of the canal.  This suggests that the allocation of water is inefficient. If some water could be reallocated from head to tail farmers, water would move from lower to higher value use and the canal would generate more net revenue.

Figure 2: Estimated marginal farmer net revenues
Estimated farmer marginal net revenues (y-axis, in Pakistani rupees per acre) as a function of water delivered (x-axis, in acre-inches). The dots-and-dash line represents average water delivery to farmers closer to the source of the canal (the head segment), solid line represents average water delivery to those further along the canal (the middle segment) and dashed line represents average water delivery to farmers at the end of the canal (the tail segment).

Using these estimates, we calculate how much farmers would gain financially if water allocation was optimized, meaning surface water supply for all farmers was made equal. We found a potential 13 percent increase in farmer revenues from reallocation.

For comparison, we used traditional, less accurate measurements from the farmer agricultural production survey and found no difference in water allocations within the canal. We also found that traditional measures of water consumed could not predict net revenues for our sample of farmers. If relied on, traditional measures of water use lead us to incorrectly conclude that water allocation within the canal network is efficient, that there is no allocative problem.

Policy implications

That there is a substantial 13 percent welfare gain from reallocation makes a strong case for moving toward some kind of water trading system. Pakistan’s irrigation sector has, in the past, seen attempts at introducing an irrigation water market, but with little success. However, this doesn’t mean we shouldn’t keep trying. As I argue in other work[7], a water market is physically feasible in Pakistan’s irrigation system.

Moving toward a market-based trading system will require political will and a lot of time. There is institutional inertia and those responsible for the large and complex canal infrastructure cannot be expected to move toward a market-based trading system overnight. But certainly, small steps toward this goal can and must be taken – a lot of value is being lost to misallocation.

The first step toward a water market would likely entail better measurement. Irrigation water is incredibly valuable (as we show), yet the existing canal measurement system is unable to specify the amount that a farmer receives (like a utility that has no sense of what each customer is consuming). The institutional machinery for measurement work exists – every length of a secondary canal has an officer assigned to it. Moreover, the physical hardware to measure flow is relatively low-cost and easy to maintain. Therefore, a first step in moving toward better allocation could involve generating finer, farmer-level water data.

Agha Ali Akram is a visiting fellow at Yale University.

[1] Professor of Economics, Yale University.

[2] “Agricultural Water Allocation Efficiency in a Developing Country Canal Irrigation System”, (forthcoming) revised and re-submitted to Environment and Development Economics.

[3] Ali, 1988.

[4] Bandaragoda, 1998.

[5] Yu et al., 2013.

[6] Tsur, 1997; Chakravorty and Roumasset, 1991.

[7] Akram 2013


Ali, I. (1988), The Punjab under imperialism 1885-1947, Princeton, NJ: Princeton University Press.

Akram, Agha Ali (2013), “Is a surface-water market physically feasible in Pakistan’s Indus Basin Irrigation System?” Water International, Vol. 38, Issue 5.

Bandaragoda, D.J. (1998), ‘Design and practice of water allocation rules: lessons from warabandi in Pakistan’s Punjab’, IIMI Research Report 17, International Irrigation Management Institute, Colombo, Sri Lanka.

Booker J.F. and R.A.Young (1994), ‘Modeling Intrastate and Interstate Markets for Colorado River Water Resources’, Journal of Environmental Economics and Management 26: 66-87.

Chakravorty, U. and J. Roumasset (1991), ‘Efficient Spatial Allocation of Irrigation Water’, American Journal of Agricultural Economics, 73(1): 165-173.

Hurd, B.H., J.M. Calloway, J.B.Smith, and P.Kirshen (1999), ‘Economic Effects of Climate Change on US Water Resources’, in R. Mendelsohn and J. Neumann (eds.), The Impact of Climate Change on the United States Economy, Cambridge University Press, Cambridge, UK, pp. 133-177.

Hurd, B.H and M. Harrod (2001), ‘Water Resources; Economic Analysis’, in R. Mendelsohn (ed), Global Warming and the American Economy: A Regional Analysis, Edward Elgar Publishing, England pp. 106-131.

Lund, J.R., T. Zhu, S.K. Tanaka, and M.J. Jenkins (2006), ‘Water Resources Impact’, in J. Smith and R. Mendelsohn (eds.), The Impact of Climate Change on Regional Systems: A Comprehensive Analysis of California, Edward Elgar Publishing, Northampton, MA, pp.165-187.

Tsur Y. (1997), The Economics of Conjunctive Ground and Surface Water Irrigation Systems: Basic Principles and Empirical Evidence from Southern California, in Parker D. and Tsur Y. (eds.), Decentralization and Coordination of Water Resource Management, Kluwer Academic Publishers, Boston, pp. 339–361.

Yu, W., Y. C. E. Yang, A. Savitsky, D. Alford, C. Brown, J. Wescoat, D. Debowicz, and S. Robinson (2013), The Indus Basin of Pakistan: The Impacts of Climate Risks on Water and Agriculture. Washington, DC: World Bank.


  1. I believe informal water exchanges takes place between farmers. What additional value would a formal water market generate. I would wonder if your calculation has discounted these informal exchanges.

    Infact, the issue is in-equity, which is created due to influence (political patronage) and collusion. Outlet tampering has reached at alarming scale in Sindh.

    At the moment all the users are unaware, who gets how much per irrigated acre from the canal system. Start from the intake from the Indus (via barrages) for all provinces, canal in a province, distributary on a canal and so on down to watercourses.

    First and for most this needs to be corrected that requires political and administrative will. I agree a decision support system needs to be installed on the system.


    • Thanks for the comments Riaz. Let me try to address them.

      Water exchanges: Yes, I believe turns are exchanged by farmers within the same outlet command. So, they can exchange when their farm gates are open. But I don’t think they actually buy and sell water.

      Outlet tampering: Right, this is a good point. Our identification strategy relied on thinking of this as a natural experiment where the amount of water a farmer gets during the growing season from the canal system is exogenous i.e. as good as randomly set. Think of it like rainfall or river discharge – something people can’t choose. Moreover, qualitatively, we also observed enforcement and corrections of any tampering in the canal in our study.

      User awareness: Agreed. I don’t think farmers are aware of discharge. The provincial irrigation department in Punjab does measure primary and secondary canal segments, and upload these measurements on their website. (Do you know if this is this the case for Sindh?) But there is no discharge measurement at the tertiary canal level, and farmers do not engage in any kind of measurement, so you are absolutely right.

      Political will: The magic ingredient! I think that better farmer level water delivery data will help.


  2. If water availability is exogenous (determined by where a farm is long the channel) then how can we speak of inefficiency in the allocation of water? Can endowments/distributions be ‘efficient’ or ‘inefficient’?


    • Thank you for this comment Haris. Hopefully I respond having understood your question. To my mind, the full amount of water a farmer gets during the season is exogenous since it is not a choice and is determined by melt in the Himalayas. My sense was that if I estimate the relation between net revenue and water delivered and find that it is non-linear then this suggests inefficiency in the allocation because marginal net revenue is not equal across farmers. If instead it is linear then marginal net revenue is equal across farmers, which is efficient.

      For allocation to be efficient, ideally the irrigation system (supply) responds to demand. So farmers should be able to buy more water if they need or sell some.

      Does this get at your question?

      By the way, the full paper is up – see here:


  3. To me the problem is the control over the timing and quantity of water. Our current supply based system is not to the best benefit of farmer. That’s why there are about one million tube-wells in Punjab. But surface irrigation is highly inefficient and we have to switch modern ways of irrigation like drip, bubblier and sprinkler etc. The World Bank funded PIPIP in Punjab and SIAPEP in Sindh are supporting this. The uptake is slow by is picking up – 50,000 acres by now in Punjab and 10,000 acres in Sindh.

    (sorry I saw your discussion just now so responded late)


    • Agreed: the surface irrigation system produces inefficient results precisely because farmers aren’t able to buy (=control) how much water they get.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s