The real and perceived accountability failures of top-down bureaucracies have generated tremendous momentum worldwide for decentralization. Policymakers have focused on moving the responsibility of key public services to elected local governments, whose proximity to the public is seen as a more direct route of accountability.
In contrast to this favorable policy impetus, existing theoretical research posits a more nuanced, and generally less rosy, picture of the potential gains from decentralization. The literature also suggests that the interests of local elites may dominate the policy choices and allocation decisions of elected local leaders, making power asymmetries, both economic and political, important. However, empirical support for either of these positions remains thin.
This lack of evidence is due in large part to the implementation of decentralization efforts, which rarely provide the opportunity for a careful assessment of the shift from a central, top-down bureaucracy to local management. Objective data on corruption and rent-seeking associated with governance failures are also hard to come by.
In a recent paper, my co-authors and I take advantage of a partial governance reform in the world’s largest canal irrigation system in Pakistan’s Indus Basin Watershed, providing some evidence on how shifting from centralized bureaucratic management to local governance impacts the distribution of power.
Irrigation bureaucracies manage most large gravity flow systems. Market-based water allocation is challenging from both a social and technical perspective, and more so in contexts where many small farmers rely on agriculture as a mainstay. Instead, water is allocated on a proportionality principal; in this case, the “warabandi” or turn-taking system set up during British colonial rule. In such quota-based systems, users have a strong temptation to bribe local officials to “look the other way” as they use various means to illicitly enhance their water entitlement. Invariably, such water theft benefits farmers at the head of the channel, where water is first to arrive, at the expense of farmers at the tail.
The promise of greater equity in water allocation and greater user participation in the management and maintenance of the canal system were important drivers of the decentralization effort, which transferred several large canal sub-systems in the province of Punjab to elected farmers organizations (FOs) constituted at the canal level.
We assess the impact of this shift on the incidence and scale of water theft along canals and examine how power asymmetries affect access to water under decentralization.
The heart of the paper is a unique dataset comprised of daily water discharge from gauges installed at the head and tail of each channel on the entire system. This database provides a credible measure of water theft along a channel. It is also available from 2006 to 2014, when the devolution of irrigation management to FOs occurred. We augment this with information on land ownership from the Agricultural Censuses. Census villages were matched with canal-level data at the head and tail, allowing for a landownership-based measure of relative power along canals.
We assess the impact of decentralization in two ways. First, we compare the outcomes across distributaries with early and late FO formation. Second, we compare zones with FOs with geographically matched controls drawn from adjacent administrative zones where FO formation had not yet begun. Comparator, or control, channels are thus selected because they are within a geographical buffer of given distance around a particular FO channel.
We find that upstream water theft increased substantially in FO transferred canals, leading to a 9 percent increase in the value of land at the head of the canal—a large wealth redistribution. Moreover, power asymmetries amplified the impact of locational (head versus tail) advantage in FO canals—the decline is water availability at the tail was greater in more unequal FO channels, dominated by large landowners at the head.
The decentralization effort in the Indus Basin clearly did not deliver on its promise of a more equitable (and efficient) distribution of canal water. It would, however, be premature to throw the reform baby out with the corruption bathwater. A more hopeful message from this reform experience might be that successful decentralization in contexts where asymmetries of power are important will likely require better mechanisms for moderating its negative impacts. For canal systems, one solution could be to give tail-enders exclusive control over FOs. Regardless of the precise governance structure, it is also useful to remember that effective decentralization requires continual support from higher tiers of government in setting and enforcing the rules of the game.