The management of irrigation systems: How to evoke trust and avoid prisoner's dilemma

World Development, Vol. 16, No. 4, pp. 489-500, 1988. 0305-750></88 S3.00 + 0.00

Printed in Great Britain. 0 1988 Pergamon Press plc

The Management of Irrigation Systems:

How to Evoke Trust and Avoid Prisoners’ Dilemma

ROBERT WADE* The World Bank, Washington, DC

and Institute of Development Studies at the University of Sussex

Summary. -The performance of canal irrigation systems depends partly on trust by farmers in the good faith and abilities of irrigation oflicials. In India, this trust is typically lacking. There is instead a Prisoners’ Dilemma-type “syndrome of anarchy” under the canals. in which farmers lack the confidence that if they refrain from taking water out of turn they will get water on time. and officials lack the confidence that if they work conscientiously to deliver the water on time, farmers will refrain from rule-breaking. In East Asia this syndrome of irrigation anarchy scarcely exists. The reasons for the difference are partly to do with agro-climatic conditions. But they are also related to policy-relevant variables of physical design and organizational arrangements. Within the constraints of agro-climatic conditions, features of East Asian design and organization might usefully be transferred to Indian conditions; including break-point reservoirs, irrigation scheduling and monitoring, separation of construction organization from operations and maintenance organization, and catchment-based operating agencies with their own revenue base.

There is an element of trust in every transaction; typically, one object of value changes hands before the other one does, and there is confidence that the countervalue will in fact be given up. It is not adequate to argue that there are enforcement mechanisms, such as police and courts; these are themselves services that are bought and sold, and it has to be asked why they will in fact do what rhey have contracted to do. (Arrow, 1973, p. 24)

1. INTRODUCTION

Canal managers, faced with irrigators who steal water or break structures, typically respond by demanding more powerful deterrents. That response fails, however, because it misses the deeper problem of which the law-breaking is a symptom, and because in any case the law as a mechanical barrier can only be effective when a tiny minority of the population are likely to break it. Most of the observance of rules has to be more voluntary, because the cost of enforcement when large numbers of the population comply involun- tarily (through a calculus of evasion and punishment) is likely to be prohibitively high. What are

the conditions in which a group of people will voluntarily subscribe to a rule of restrained access to irrigation water?’

My context is surface irrigation systems where public officials operate the higher levels of the distribution network and farmers take over below a certain point. Here the question becomes: what are the conditions in which trust by farmers in the good faith and abilities of irrigation officials can be sustained? Trust here refers to the quality of confidence in a relationship which permits one party to act (e.g., invest resources) before knowing that the other will behave as promised.

*Based on paper given to the conference on “Water and Water Policy in World Food Supplies,” Texas A&M University, 26-30 May, 1985. The views expressed here are my own and must not be attributed to the two organizations to which I am affiliated. Acknowledgements: Ron Dore, Harald Fredetiksen. Henry Hart, international Irriaation Management Institute, Gilbert Levine, Arthur-Maass, Ronid Ng, Herve Plusquellec. Richard Reidinger, David Seckler, Gabriel Tibor.

489

490 WORLD DEVELOPMENT

2. THE NEED FOR AUTHORITY

Whenever water is scarce in relation to demand, there is a rationing problem. Even in those rare canal systems where water is sold in a market’ and where price is thus the principal means of rationing, an organization to administer and sanction the rationing is needed. In the more usual situation where water is not priced volu- metrically, the “load” on the rationing organization is heavier, and increasingly so as water becomes scarcer in relation to demand. More and more farmers will want to take water they are not entitled to.

The situation from the viewpoint of an individual irrigator (or small group of irrigators) might be modeled as a Prisoners’ Dilemma game.’ Each individual has a clear preference ordering of options: the first preference is for everyone else to abide by a rule of restrained access while the individual enjoys unrestrained access - while he “free rides”; the second preference is for everyone, including himself, to follow the rule; the third preference is for no one to follow the rule - for everyone to grab what he can; and the worst outcome is for everyone else to grab while he follows the rule. Given this order of prefer- ences, the stable outcome is likely to be the third- ranked alternative, unrestrained access. From the more socially desirable second-ranked alternative (restrained access by all), each individual has an incentive to cheat and go for his own first preference (restrained access by all except him), with the result that the aggregate situation deteriorates to the outcome of the third-ranked preference. In other words, mutual rulebound restraint is not necessarily a stable equilibrium, because from that position each individual can be immediately better off by cheating if the others do not also immediately cheat, and if the others do cheat, his own cheating at least ensures he does not end up at his own least preferred outcome. But the third-ranked preference implies that unless water is very abundant, many farmers will get little or unreliable water supply, because their own supply is a function of the unrestrained access of those higher up the distributory. They will in turn struggle to obtain better supplies by stealing water, breaking the gates, breaching the banks, bribing the officials, even murder; SO

compounding the overall problem. Many theorists of Prisoners’ Dilemma have

concluded that the socially desirable outcome of restrained access by all can be sustained only by an external authority imposing penalties against rule violation. If so, the conclusion is a counsel of despair. In the reality of most Third World countries, legal mechanisms and the authority of

government are simply not powerful enough to make a sufficient1 plausible threat across myriad micro situations. 1

But the conclusion is too strong. Prisoners’ Dilemma is useful in drawing attention to the point that just because a set of individuals has a joint interest in a certain outcome, this does not mean the outcome will be forthcoming. Beyond this point, however, Prisoners’ Dilemma is often misleading for the irrigation case. It makes good sense to suppose that in many situations individual irrigators will restrain their water rule- breaking ifthey are confident that others will also refrain, and if they are confident that they will still get as much water as they are fairly entitled to (even if not as much as they would like). They will more likely refrain from cheating if they are confident that by doing so they will not be the “sucker.” Where people are so motivated by an “I’ll restrain if you restrain” calculation. then institutions (such as an Irrigation Department) which convince them that these expectations are justified can promote voluntary compliance with the rules. The reason is that the rational individual will not free ride regardless of what others do, as the Prisoners’ Dilemma game assumes. He will calculate the likelihood that his own free riding will encourage others to do so and how much their free riding will reduce his benefits. It must be remembered that the individual’s first preference is to free ride while others do not free ride, so that he continues to get the collective benefit which their restraint produces. If that collective benefit might not be forthcoming because his free riding encourages others to do the same, it may be rational for him to comply with the rule. Free riding remains a possibility, but not an imperative (Runge, 1984; Kimber, 1981).

This argument is hypothetical. But there is a growing body of evidence - admittedly about contexts rather far from what we are discussing here - which suggests that ideas about “fair- ness” are important factors in individuals’ choices about whether to free ride, and that the strong version of the free rider hypothesis (that individuals will voluntarily contribute very little of their resources towards the provision of a public good) is implausible (Runge, 1984; Marwell and Ames, 1981; Margolis. 1982).

Or consider the converse. Henry Hart identi- fies a “syndrome of anarchy” at work under Indian canals, which is, he suggests, a major factor in their under-performance (1978). The syndrome of anarchy grows out of and reinforces a basic lack of confidence, of trust in good faith, on both sides of the irrigator/irrigation official divide. The farmers lack the confidence that if

MANAGEMENT OF IRRIGATION SYSTEMS 491

they from water out turn (re- from stealing breaking the structures,

bribing the offtcials) they will nonetheless get water on time. The officials lack the confidence that if they do work conscientiously to get the water delivered on time, farmers will refrain from rule-breaking. It is a “syndrome” in that the behavior of each party to the relationship tends to confirm the negative expectations held by the other. Each is the other’s headache. The syndrome may operate between individual farmers within a given village. who behave anarchistically towards each other over water matters; or, given the capacity of some villages to organize canal- water distribution within the village (Wade, 1986; 1988), it may operate only between village units, each village unit struggling to get more water than it is entitled to against other village units.

Breaking this syndrome has to be done primarily from the government side, by means of a sustained demonstration of the ability to deliver reliable and expected amounts of water if the farmers do not interfere. Our question then is: how can public officials assure farmers that if they restrain their taking of irrigation water, they will get the expected amounts?s

The answer depends on the authority which farmers grant to the irrigation staff. Political theorists generally stress two elements as precon- ditions of authority. First, that the person given decision-making power should have some competence which makes such a division of labor sensible; second, that he should evoke trust in his good intentions - that his use of power should be seen not to be predatory and self-interested, but concerned with the welfare of the larger whole of which both he and his subordinates or clients are part.

We can use this argument to guide a comparison of Indian and East Asian canals. In East Asia (meaning Japan, South Korea, and Taiwan) the syndrome of irrigation anarchy is much less developed than in India, and we may be able to learn something from East Asia about how to reduce it in India (and wherever else it is strong). But we must first separate out those reasons which are inherent in differences of climate and topography from those which are more amenable to policy. We begin by considering agro-climatic differences between South Asia and East Asia which influence authority and anarchy in irrigation relations; then go on to consider differences in physical design and in organizational arrangements; and end with some lessons for Indian irrigation.

3. ENVIRONMENTAL CONDITIONS AFFECTING AUTHORITY

(a) Rainfall and evaporation

Rainfall and evaporation (more precisely, pre- cipitation and potential evapotranspiration) are the two basic climatic variables which determine the nature of water control needed for efficient agriculture. Different configurations of these variables imply greater or lesser difficulty in maintaining appropriate soil moisture conditions. Consider the two configurations shown in Figure 1. Figure la is a stylized picture of the climatic pattern characteristic of much of South Asia, while Figure lb shows the same for East Asia. Notice three differences. Evaporation in South Asia is higher annually and in almost all months than in East Asia. Average rainfall is lower. And seasonal rainfall is much more concentrated

South Asia East Asia

JUM we An A”. au

Figure 1. Climatic patterns of South and East Asia.


(Vaidyanathan, 1985). So the irrigation requirement (excess of evaporation over rainfall) is larger in South Asia over the year and in most months.

(b) Topography

Most South Asian irrigation takes place on wide flat plains. Most East Asian irrigation takes place on narrow plains and alleys, and on land of relatively steep slope.

(c) Implications for physical design

(i) Size In Indian conditions, with rainfall more limited

in amount and more concentrated in time than in East Asia, local water sources are less reliable. Most surface irrigation water must be brought in from rivers that have their sources in areas of heavy rainfall or snowmelt or from reservoirs fed from such rivers. The large expanses of flat land make it possible for a single source to irrigate a very large area. In East Asia, local water sources can be more reliable, and for topographical reasons, the size of system must be relatively small. Hence a characteristic difference in size and number of water sources per system: Indian canals are large, with a single water source at one end; East Asian canals are small, with several sources. The typical Indian canal has a command area of 50,000 to 100.000 hectares (Wade, 1982b; 1985). Few canal systems in South Korea or Japan command more than 10,000 hectares, while Taiwan has a few of more than 50,000 hectares but none above 100,000 hectares.

(ii) Relative water supply Water is normally much scarcer in Indian

systems, in the sense that the supply available at the head of each system is much smaller per 1,000 irrigable hectares than in East Asia. In other words, “relative water supply” is less. meaning that the ratio of the amount of water available at the head of the system over the amount needed to meet the irrigation requirement for the actual (or potential) irrigated area is lower.

(iii) Intermediate storage East Asian canal systems normally consist of

linked series of small reservoirs and canals. The small reservoirs may each have their own catchment, but they may also be used as “on-line” storage in which to place flows from other reservoirs or rivers until needed. Indian canals, for

reasons of rainfall and the difficulties of local reservoirs in flat topography, typically have no intermediate storage. The water may take a week or more to travel unimpeded from head to tail.

(d) Implications for water control and authority

The amount of water control needed (at system level rather than farm level) to irrigate effec- tively is much higher in Indian conditions. The lower is relative water supply, the more control over the water is needed to irrigate to a given standard of adequacy (Levine, 1977). The less intermediate storage, the more difficult it is to reach that level of water control, in particular because lags in the adjustment of supply to demand are longer than where water can be brought to an area from a nearby intermediate storage. And the flatter the topography, the more essential it is that water in the canal is main- tained at full capacity level in order for it to command the expected area; whereas on steeper ground less water control is needed because even if water level falls in the canal the discharge may still be able to irrigate the land. These conditions mean that a larger amount of authority is required in relations between irrigators and canal staff than in East Asia.

4. PHYSICAL DESIGN FEATURES AFFECTING AUTHORITY

Both the need for trust in authority and the likelihood of it being forthcoming are related to the physical structures. The bigger the system. the more difficult it is for individual farmers to understand how the system operates and the constraints under which canal staff work. The more difficult it is for farmers to understand. the more crucial is the trust element (as opposed to the competence element) in the acceptance of authority. And the more that active act of faith is needed, so the more those given authority must be seen to be using their power in a benevolent and not in a predatory way, in good faith for the common good and not merely for personal aggrandizement.

From the viewpoint ot a farmer under the typical Indian canal, the rest of the system outside his immediate locality is a mystery. It is too large to comprehend as an entity. He simply has to take the word of the irrigation staff as to what the overall water supply situation is. He and his village neighbors have no stock of water which they can call their own. Their “own” water begins


at the outlet; but plenty more water may be available in the distributory on the other side of the outlet. They can go at night with a spade, take water against the rules, and leave little trace behind.

Moreover, the design of the typical Indian canal itself necessitates that the irrigators must have a high degree of trust in the irrigation officials if they are to refrain from grabbing. The typical Indian system is nor the kind found in the Northwest, with its ungated outlets and rota- tional delivery rule in the main system, although the Northwestern or warabandi system is best- known in the literature (Malhotra, 1982; Reid- inger, 1974). The typical design uses (more or less) continuous flow in the main system and gafed outlets. with few cross-regulators. Here water levels in the main system vary greatly, and to maintain constant discharges through the out- , lets it is essential that an irrigation’official run around raising and lowering his gates. But the raising and lowering of one gate will typically set up changes in upstream levels, requiring further gate changes there. In theory, the system con- tains a higll level of water control capacity, despite the lack of cross-regulators to control level in the main system. In practice, the irrigation staff rarely show the required devotion to duty and technical skills needed to carry out the required routines, and the unadjusted gates discharge fluctuating amounts. But farmers usually want constant, not fluctuating discharges and take whatever defensive action is needed to smooth out the fluctuations-perhaps by adjust- ing the gates themselves in line with their local interest. The failures of the irrigation staff force the farmers to compensate in ways which make the overall situation worse (Wade, 1982~).

By contrast, the smaller and more buffered canals of East Asia can be better comprehended by farmers. The lower irrigation requirement and higher relative water supply also mean that farmers need to have less trust in officials because the risks of water stress are lower.

In short, the design of most Indian canals requires that farmers have more trust in irrigation officials than is needed in East Asia; while it also makes that trust more difficult to create and sustain.

5. ORGANIZATIONAL FEATURES AFFECTING AUTHORITY

Several organizational features of Indian canals limit the farmers’ judgment of the competence of staff and check their trust in the staffs good intentions.

(a) Relative bureaucratic supply

The sheer density of the irrigation staff is dramatically lower under Indian canals. The ‘*relative bureaucratic supply,” or irrigation staff per 100 irrigated hectares, is of the order of 0.3-0.5 in many Indian canals, and around 2.5 in South Korean canals, or five to eight times greater.6

(b) Organizational structure

Three basic structural features of Indian Irriga- tion Departments help to keep competence and trust at low levels in India.

First, the Irrigation Departments are primarily construction agencies. This has the following consequences. Few of the professional staff on a particular canal have much prior experience of operation and maintenance (O&M). (If O&M posts account for, say, an eighth of the total professional posts and if there is no career specializa- tion, any one individual can expect to spend only an eighth of his career on O&M.) There is little or no training for O&M staff.’ Professional repu- tation is anchored in construction. The O&M budget is given little attention, because it is only a small part of the overall Irrigation Department budget. Engineers tend to behave while doing O&M work in the top-down hierarchical control mode appropriate for construction but inappro- priate for O&M.

Second, the budget of the Irrigation Depart- ment is unrelated to the collection of water rates. Water rates are normally collected by a separate department; and in those few Indian states where rates are collected by staff attached to the Irriga- tion Department. the department’s budget is still entirely unrelated to what it collects. Lacking a revenue base of its own, it must depend on the government budget. With no link between farmers’ water payments and their own income and professional activities, the staff need not concern themselves with the prosperity of agriculture in their area, or feel a sense of accountability to the irrigators whose money they are spending.

Third, coordination between the Irrigation Department’s decisions and those of other agencies which provide important inputs into irrigated agriculture is almost non-existent. As a result, irrigation water might be released into the canal at the start of the season in ignorance of whether farmers are ready to use it - yet delays in the disbursements of credit may have hindered farmers from buying the inputs needed to make prompt use of the water. The lines of authority


run parallel. from the top of each department in the state capital right down to the lowest rank of field staff; at each level. the staff of each department go their own way.

The three East Asian countries use a very different form of organization, which is a catchment-based parastatal agency. In Taiwan it is called an Irrigation Association; in South Korea, a Farmland Improvement Association; and in Japan, a Land Improvement District.s In South Korea and Taiwan the term “association” misleadingly suggests democratic control by farmers’ representatives, whereas in fact there is no such mechanism of accountability. (In Taiwan the irrigators’ council of each association - which in any case had fairly circumscribed powers - was abolished between 1975 and 1982, and the corresponding body in Korea’s Farmland Improvement Associations was abolished soon after Park Chung Hee took power in the early 1960s.) However, each parastatal has consider- able autonomy within its geographical jurisdiction.

The primary function of each irrigation agency (IA) is canal O&M, not construction. Construc- tion is carried out by a separate, narional-level organization which undertakes the design and supervision of construction of all sizable irrigation structures in the country. Second, the IA has its own revenue base: it collects water rates from farmers (so much per irrigated hectare), and from this revenue meets most of its operatin costs for staff salaries, routine maintenance. etc. So the staff have a direct sense of dependence on the prosperity of farmers under their own system. Third, South Korea’s IAs provide some agricultural extension and land development; indeed, the same people who operate the canals provide the extension. Here coordination between the water supply function and the extension function is built into the organization. The ordinary extension service, organized in provincially-based parastatals. also provides extension in irrigated areas. In Taiwan, the IAs do not provide extension - that is done by separate Farmers’ Asso- ciations. The jurisdiction of the IAs and Farmers’ Associations are not congruent; but the key point is that a stable set of officials is engaged in the same small area, some providing O&M, others providing extension, and informal coordination develops between them.

In none of the three countries are the IAs involved in input supply or marketing, or in other-than-minor construction. A higher level (provincial and/or national) agency monitors the performance of each IA against various kinds of financial and physical targets.

The East Asian form of Irrigation Association

is thus quite different to the Indian pattern of a centralized Irrigation Department (at state level), with a Chief Engineer General at the top of the pyramid in the state capital, a hierarchy of regional and project offices stretching out beneath him; with the same department attend- ing to construction, investigation, design, and O&M across the entire jurisdiction of the state (which may have a population of 40 million or more); dependent for its revenue on budget grants; and with closer ties to other public works departments than to the agricultural departments.

Organizational structure is only one deter- minant of how organizations perform; the same structure can be activated in different ways. But one can see just from the structure how trust between farmers and irrigation staff and within the irrigation hierarchy itself is easier to maintain in the East Asian type of organization.“’

(c) Procedures

Several operational and organizational procedures used in East Asia also help.

Taiwanese and Japanese systems have clear rules of water delivery (who gets water. when, how much) against which actual deliveries can be assessed. These rules may be only loosely observed at times of year when water supply is abundant, but kick in strongly as water becomes scarce. The explicit water delivery schedule helps farmers assess staff competence.

The contractual nature of the employment relation between the IA and individual staff members is obscured by, or replaced by a sense of common membership in a corporate entity which has objectives that can be shared by all members. A variety of methods are used to build up this corporate sense within the IA. In Korea these include competitions between the field sta- tions (in quality of canal maintenance, collection of water charges, volleyball, even the making of farmyard manure, intended to spur farmers to do the same); once-a-week joint exercise sessions for all the staff: and weekly planning meetings for all field station chiefs. Most importantly. all the staff (other than the patrollers) are paid on rising seniority scales, so that pay is not closely tied to hierarchical rank; all the staff expect to have life- time employment; and the lower ranks in each IA identify more closely with their IA than with lower ranks in other IAs, which helps to avoid the ingrained conflictualism which is found between lower and higher staff in Indian Irriga- tion Departments and which erodes a sense of common purpose and pride in work. Promotions


are much influenced by each person’s “promotion score,” which weighs length of service, supervisor’s judgment of performance (diligence, willingness to take responsibility, loyalty), and also test scores in the regular in-service training courses arranged by a national agency. The latter procedure gives individuals a clearly structured incentive to undertake in-service training (Wade, 1982b. p. 113).

In Korea, staff involvement with farmers is promoted by such devices as a monthly news- letter sent to all irrigators; and more importantly by procedures relating to the appointment and supervision of the “patrollers.” The patroller bicycles around a 10&150 hectare jurisdiction twice a day opening and shutting gates and performing minor maintenance. He is at the bottom of the irrigation hierarchy. At the same time, he must be a farmer with land within the jurisdiction which he irrigates, so that he experiences irrigation problems at first hand. He is nominated each year by the headmen of the villages within his jurisdiction. If the headmen are unsatisfied with his performance, they nominate someone else. In this way the lack of “farmer participation” in irrigation system management (noted above) is partly offset by internalizing farmers at the bottom of the formal management hierarchy itself, in the role of patroller.

Most staff of the IAs. even (male) clerks, are responsible for supervising one patroller in addition to their other duties. At least once every two to three days they are expected to go to their patroller’s beat and make sure that he is on the job. They are provided with use of small motor- cycles for the purpose. They are expected and inclined to stop and chat with farmers whom they pass along the canals. The intensity of this local contact helps further to make up for the paucity of more formal channels of communication between farmers and staff, which is an expression of the authoritarian character of the South Korean political regime (Wade 1982b; 1983).

Recruitment and promotion procedures ensure that the senior-level staff are natives of the area in which they work. Recruitment to these positions is generally by internal promotion (with some exceptions for technically-specialized positions, such as head of land development). Virtually all the senior management will have spent many years on the project before they make it to their position - except the President, who is appointed directly by the Ministry of Agri- culture and may be an Agriculture Ministry official near retirement. Hence the eyes of the irrigation staff are kept firmly on the locality. and identification between their interests and those of farmers is further encouraged.

Local affiliation of the staff is important because it gives both sides - staff and farmers - a set of shared experiences. This directly assists a sense of mutual obligation between them; and also provides a basis for a shared set of beliefs according to which the existing order is fair and just, and every betrayal is perverse and unjust - including betrayal of the irrigation agency’s rules. This is a much more cost-effective method of avoiding free-rider problems than relying on a calculus of punishment.

A higher level bureau within the Ministry of Agriculture lays down general constitutional principles and regulations about, for example, salary scales, staffing densities, ratio of administrative to maintenance expenses, water rate collection, and so on. The same bureau monitors the performance of the IAs against these and other criteria (such as rice production, irrigated area). In addition, each IA can expect to have a team arrive from the ministry at least once a year, to make an inspection which may last several days and includes not only the financial accounts but also the state of the physical structures. Prizes are awarded to IAs for good performance.

In most respects the organizational procedures of Indian irrigation are very different. For example, explicit water delivery schedules are lacking outside the Northwest (Punjab and Haryana). Officers are normally rotated in and out of any one post every 18-24 months, and have no identification with the area of their responsibility (Wade, 1982~). Other than at field- staff level there is no stable set of people asso- ciated with a particular canal. Transport facilities are poor and distances are large, so officers normally confine their irrigation-related travel to the most accessible parts of the command area, not being seen in the less accessible parts for months on end (e.g., Hart, 1978, p. A-126).

Such conditions are almost guaranteed to create a situation in which the legitimacy of the public officials who allocate water is always close to being questioned, whether by their own subordinates or by the farmers. The farmers response is not only a willingness to steal water and break the structures, but also to compensate for the weakness of personal trust by offering bribes to protect their local group from the officials’ neglect or abuse of power. The bribe payments and their distribution up the bureaucratic and political hierarchies are governed by well- defined institutional arrangements based on the ease of transferring officials in and out of posts. This system of corruption has far-reaching consequences for canal performance (Wade, 1982~; 1985b).


The difference between India and East Asia in the length of time that senior officials of a project have worked on the project is striking in light of a rule of thumb formulated by Fred Hotis. former irrigation advisor at the World Bank: the single best indicator worldwide of the effectiveness of project management, he suggests, is the number of years the senior staff have been in position (Hotis, 1981). Of course. the direction of causal- ity is debatable; senior staff may want to stay longer on well-managed projects. And low rate of personnel transfer is unlikely to be a sufficient condition of effectiveness; where construction dominates, assignment to O&M rather than construction may be such a stigma that senior O&M officers languish in post for years, ignored by the rest of the organization.

6. IMPLICATIONS

Large, water-short systems are a given in Indian conditions. Within that constraint. performance might be improved by changes in both the organizational and physical design. The guiding principle of these changes should be both to reduce the need for farmers to trust the staff and to increase the likelihood that the needed trust will be forthcoming. Changes which meet this criterion have a chance of breaking the damaging “syndrome of anarchy.”

The basic way to reduce the need for trust is to reduce the farmers’ dependence on the allocation decisions of officials. For this. two conditions are especially helpful: first, an “inventory” of water, so as to buffer fluctuations in water supply and water demand (thereby avoiding the necessity for changes in local demand to be transmitted to the source itself, with subsequent flow-time delays in adjustment of local supply); second, a clear hand-over point where the officials’ jurisdiction ends and the farmers’ jurisdiction begins. One design feature which is promising from this point of view is the “on-line” or “break-point” reservoir. The break-point reservoir does not have to be between the outlet and the farmers’ fields: it could be at a higher level of the system, serving more than one outlet or village. The important points are that it be at a low enough level for farmers to see the stock of water which is “theirs,” for water to be able to reach all the fields within a few hours, and for farmers to have a large (legitimate) hand in how it is allocated. However, the break-point reservoir also has major drawbacks: water losses are often very high (efficiency low); land acquisition is commonly difficult and costly; siltation may be high; health hazards arise; and the storage capacity

- therefore the ability to buffer - is typically small.

It is interesting that in Taiwan, the same Japanese engineers who put break-point reservoirs into the Taoyuan system during the colonial period did not put them into most of the other systems on the island, partly because the flatter slopes of most of the other systems made break- point reservoirs much more expensive. (In any case the other systems had more water sources within the command area than the Taoyuan, making break-point reservoirs fed only from the canal less necessary.) In practice, more use of within-canal storage can substitute for a break- point reservoir in terms of the “inventory” condition; though within-canal storage does not provide for such a clear hand-over point between officials’ and farmers’ jurisdictions. One wants to aim at a physical design which will help an organizational distinction between the task of water conveyance, which is properly the concern of experts in hydraulics, and that of irrigation, which should be the concern primarily of agricultural experts. The design should permit the former agency to deliver plugs of water according to simple rules (Levine and Coward, 1986).

Simple operating procedures help to make the system more transparent; the more transparent the system and its operating principles, the more easily farmers can judge the competence of the irrigation staff and so, by hypothesis, the less needed is trust in their good faith for the legitimacy of their allocation decisions to be respected.” But also, the simpler and more transparent the operating procedures the easier it is for the officials to be competent at performing them, and so the more farmers will acknowledge their legitimacy. Complex procedures, which require canal staff constantly to adjust gates, may impose an administrative load which the organization is unable in practice to meet. On the other hand, simple rules have a cost: when water demand is fluctuating below the outlet. a simple rule like “constant discharge to each outlet” results in water wastage (in the absence of intermediate storage).”

Whether the rules are simple or complex, a system of monitoring and accountability is needed. It helps both to reduce the need for trust and to increase the likelihood that enough trust will be forthcoming. It should comprise: (a) a clear definition of what the farmers are to get - how much water, when, and what flow rate; (b) a measurement of what the farmers (at each outlet or distributory offtake) have received, with the measurement given to the farmers on a regular (e.g., weekly) basis; (c) an internal water audit carried out at higher levels of the irrigation


hierarchy to verify the accuracy of the lower level measurements: and (d) an external sanctioning mechanism which allows farmers who do not believe the official measurements to bring the matter to public attention.

If the irrigation agency is structured in the IA rather than the Irrigation Department form, this competence-trust-legitimation dynamic can be greatly strengthened. For where the water supply agency is structured along IA lines, it is more likely that the staff will be - and will be seen to be - competent and trustworthy, because of their (a) O&M orientation, (b) sense of dependence on farmers’ prosperity, (c) embeddedness in local social relationships, supplemented by (d) the sorts of promotion incentives and (e) agency scrutiny from above, described earlier.

Of course, it is easier to introduce more inventory capacity and monitoring procedures than it is to switch a centralized Irrigation Department towards an IA form. (Such a switch has been brought about in very few countries, Morocco being a partial example.) To a degree, the break- point reservoir could be a substitute for that more difficult organizational change, since the centralized Irrigation Department could continue to run the main system, as now, but only as far down as the break-point. If the break-point came relatively high up the system, one could imagine an IA, covering the whole command area or major portions of it, taking over responsibility for water allocation from the break-point reservoirs; or if the break-point came lower, farmers might do it themselves in less formally organized ways. The need for a monitoring and accountability mechanism would remain, wherever the break-point.

Valuable but less far-reaching changes in the same direction could be, and in a few places in India, are being made. In the state of Maha- rashtra, for example. responsibility for the operation and maintenance of tertiary distributories has been taken from the Irrigation Department and given to new project-specific Command Area Development Agencies, which combine a staff of irrigation engineers with agricultural specialists. But the Command Area Development Agency is still vulnerable to one of the main problems, which is rapid transfer of officials in and out of post, with trust-eroding consequences.

Another promising experiment in India involves the sale of bulk amounts of water to a tertiary distributory - to all the farmers under that distributory as a unit. The farmers themselves then organize the distribution of water and the collection of the fee. The experiment, in the state of Gujarat, is still on a very small scale; the area of the tertiary is only about 400 acres. and

the crop is sugar cane - a high-value, low-risk crop. Extension of the idea to bigger units and other crops will be more difficult. However, the experiment warrants attention, particularly because it seems to provide a method of getting some of the benefits of the break-point reservoir but without the costs (in land removed from production, for example).

Similarly. the type of canal design used in France and Morocco - which combines hydro- mechanical automatic gates plus constant discharge modules - looks promising from a trust perspective. primarily because it reduces the dependence of farmers on the irrigation staff and simplifies what the irrigation staff have to do. But there are other problems with this technology, especially its cost. dependence on reliable supply at the head. rigorous design and construction standards, etc.” Again there are trade-offs.

7. CONCLUSIONS

The question of how to stop farmers from stealing or bribing more water than they are entitled to translates into the question of how to enhance the legitimate authority of the irrigation staff. Holding things like the degree of water scarcity constant. legitimacy depends on farmers’ judgment of how competent the staff are and on how much they trust their good intentions. Given that competence is often difficult to judge (especially when the system is big and the farmers’ knowledge of it limited to the locality), trust becomes the crucial factor. If the “syndrome of anarchy” under the canals is to be weakened, farmers must be confident that if they restrain their own individual or small group access to water, they will still get enough water; or at least that if they do not get enough there are good reasons beyond the control of the staff and the shortage is being shared fairly. The amount of needed trust can be reduced by features of the physical design - such as punctuating the hydraulic system at some point within the ken of farmers. Features of the organizational structure can make it more likely that actual trust will equal needed trust, by making it more likely that trust-evoking behavior by the irrigation staff will be forthcoming: an organizational separation between the O&M function and the construction function; a revenue base for the O&M organization, so as to link the amount collected in water rates with the organization’s budget; and O&M functions performed by geographically circumscribed parastatals, so as to strengthen impersonal trust with the more personal trust of enduring social relationships. Operational and


organizational procedures matter as much as organizational structure, however: especially clear rules of water delivery and an arrangement for.comparing actual against expected deliveries, the results going to farmers as well as up the hierarchy; tying of promotion to in-service training; parametric controls on irrigation agencies from above, with wide latitude within those para- meters; and so on.”

To deepen this argument it should be situated in an explicit typology of irrigation systems. Use- ful typologies are sadly lacking in irrigation studies. But the foregoing suggests one which uses both a “physical” and a “social’* variable as the basis of classification, perhaps “relative water supply” and “relative bureaucratic supply.” In that case, we have a typology of the following kind (Figure 2). Systems of type C are those with abundant water supply per area irrigated and

I _----,----_--

I I I

I.5

Figure 2. Typology

Relative water supply

of canal irrigation systems.

high density of irrigation staff. East Asian systems tend to be of this type. Systems of type A are those with scarce water supply and low density of irrigation staff. Indian systems tend to be of this type. Systems B and D are in between.

We could think of “bureaucratic stress” and the “syndrome of anarchy” as being a function of the product of relative water supply times and relative bureaucratic supply, in which case we have Figure 3. Type A systems are more likely than type C systems to show bureaucratic stress, low legitimacy of staff in the eyes of farmers, Prisoners’ Dilemma-type behavior, and the syndrome of irrigation anarchy. The foregoing features of physical design, organizational structure, and operational and organizational procedures may then be brought in as modifiers within these broad types. (And of course legitimacy is affected by variables which are not irrigation- specific, such as cultural beliefs about inequality, authoritarian character of the state.) At the least, however, this typology helps to guard against an easy optimism that transfer of organizational arrangements from East Asia to South Asia would permit South Asian irrigation systems to run as well as East Asian: because it shows that the Indian systems are much more highly stressed than the East Asian ones in both physical and social dimensions, which means that attempts by farmers to break the rules are bound to be more determined and frequent.

Figure 3. Bureaucratic stress, syndrome of anarchy.

NOTES

1. The same question can be asked for other common property resources, such as grazing and trees. See Wade (1986; 1987a; 1988).

2. As in the Alicante canals of southern Spain: Maass and Anderson (1978).

3. The parable of the Prisoners’ Dilemma is well- known. Two suspects are being separately interrogated about a crime they jointly committed. They know that if they both stay silent they will receive a light prison sentence. If one stays silent while the other confesses the first will receive a long prison sentence while the

other goes free. If both confess they both receive a medium prison sentence. Each person can choose only once -which means that if one chooses to stay silent while the second confesses the first cannot then confess upon learning of his sentence. This is what creates the dilemma. Their joint interest is for both not to confess (to “cooperate”). But the outcome is that both confess. From either’s viewpoint, staying silent while the other confesses would give the worst outcome. and confess- ing at least ensures that this outcome is avoided while also opening the possibility that the confessor will go free. Each player thinks he will be better off not cooperating regardless of what the other does, because


each cannot be certain that the other will cooperate. See Rapoport and Chammah (1965); Wade (1987a); Lipton (1985).

4. There are exceptions. See Wade 1982b, Chap. 5; also 1983, on South Korea.

5. Irrigation engineers in India, talking about how they try to manage water allocation, commonly stress the need to create trust: “You have to build up affec- tion,” explained one. Another remarked that “Water management is 25% water and 75% soothing.” The Indiaiide of my remarks is based on detailed tieid work carried out between 1976 and 1982, which included lengthy periods of residence in a canal command area as well as trips to several others.

6. The figures are based on a comparison of a South Korean canal and a South Indian canal (Wade, 1982a. p. 25; 1982b). The lower figure for the Indian system is based on the first season irrigated area only. to match the one crop a year of the Korean system. The higher figure is based on gross irrigated area.

7. Only in the 1980s has this surprising state of affairs begun to change, and then because of pressure from foreign aid donors (Sundar. 1984. p. 22).

8. My statements on South Korea are based on three months of field research on irrigation organization in 1979 (Wade, 1982b); for Taiwan, on short visits in 1979 and 1983 and on Moore (1983). The East Asian form has a family resemblance to that used in Italy (Wade, 1979).

9. The South Korea IA I studied in detail covered about 80% of its recurrent costs from water charges (Wade, 19826, p. 42).

10. In the interests of brevity, I fudge the distinction between trust in relations between farmers and the agency, and trust between hierarchical levels of the agency.

11. The argument for simplicity of operating procedures means that the demand to which water supply is to be adjusted has to be considered at a relatively high level of aggregation. rather than at the level of the

individual needs of each person or micro-location served by the system. One reason is limited information-processing capability of most bureaucracies (and the unreliability of computer technology in most rural Third World conditions). This applies even without the second reason. which greatly reinforces the point. Where high margins of discretionary control over water allocation are available to irrigation officials and where water is scarce. corruption is likely; which may reach the point where irrigation officials have an incentive to act to reduce farmers’ confidence in the reliability of water supply, because this enhances the protection money the officials are able to extract from particular groups of farmers in return for increasing their confidence about their own supplies (Wade, 1982~).

12. A rule of intermediate complexity might require a constant discharge during each 24-hour period but adjustment of the discharge at the end of the period, the adjustment being based on farmers’ demand or the agency’s calculations or some mixture thereof.

13. But cost comparison between this type of technology and the more conventional type are exceedingly difficult to make. The gates and modules are said to be about twice as expensive as conventional ones; but if, as I have heard said, this adds only another 5% to 10% to total cost compared to the costs of the same canal with conventional technology, this might be offset by the extra benefits of more reliable water supply.

14. What I am uncertain about is the degree to which improvements can be expected in management effectiveness by improvements in monitoring, without also making changes in organizational structure (Leonard, 1987). A noted Indian irrigation specialist leaves no doubt about his view:

Management is based on the premise that things can be done better. which in turn means that one wants better performance. In a sociopolitical situation where what is legitimate is what one can get away with, can there be any concern about public system performance? And if there is no desire to manage, what can management techniques do? “In the land of nudists, what can a washerman do?” (Sundar. 1984. p. 22)

REFERENCES

Arrow, K., Information and Economic Behavior Leonard, D., “The political realities of African (Stockholm: Federation of Swedish Industries, management,” World Development, Vol. 15, No. 7 1973). (1987). pp. 899-910.

Hart, H., “Anarchy, paternalism. or collective respon- Levine, G.. “Management components in irrigation sibility under the canals?” Economic and Political system design and operation.” Agricultural Admini- Week/y (Bombay), Vol. 13, Nos. 51 & 52 (1978). strarion. Vol. 4, No. 1 (January 1977).

Hotis, F., “World Bank irrigation experience,” ODI Levine. G., and W. Coward. “Irrigation water distribu- Irrigation Management Network, Paper 9d (London: tion: Implications for design and operation.” Overseas Development Institute. April 1984). AGREP Working Paper No. 125. Agriculture and

Kimber, R., “Collective action and the fallacy of the Rural Development Department (Washington, DC: ~~;;&fallacy,‘* World Politics, Vol. 33 (1981). pp. The World Bank, 1986).

Lipton, M., “Prisoners’ Dilemma and Coase’s The-


orem: A case for democracy in less developed Italy.” Economic und Political Week& (Bombay), countries’?” in R. Matthews (Ed.). Economics and Vol. 14. Nos. 51-52 (19-22 December 1979). Review

of Agriculture.

University Press, 1982).

Democracy (London: Macmillan.’ 1985). Maass, A., and R. Anderson, And the Desert Shall

Marwell. G.. and R. Ames, “Economists free ride.

Rejoice: Conflict, Growth and Justice in Arid Environments (Cambridge, MA: MIT Press, 1978).

Malhotra, S., The Warabandi System and la Infra- structure. Publication No. 157 (New Delhi: Central Board of Irrigation and Power, 1982).

Margolis, H., Selfihness, Altruism and Rationality: A Theory of Social Choice (New York: Cambridge

Wade, R. H. “Employment. water control and irrigation institutions: South India and South Korea,” Working Paper, Asian Regional Employment Pro- gramme. ILO. and IDS Discussion Paper No. 182 (Institute of Development Studies, University of Sussex, 1982a).

Wade, R. H., “The system of political and administrative corruption: Canal irrination in South India,”

Wade. R. H.. Irrigation and Agricultural Politics in South Korea (Boulder, CO: Westview Press, 1982b).

does anyone else?: Experiments in the provision of Journal oi Development &dies, Vol. 18. No. 3 public goods IV,” Journal of Public Economics Vol. (1982~). 15 (1981). pp. 295-310. Wade, R. H.. “South Korea’s agricultural develop-

Moore, M..

Reidinger, R.. “Institutional rationing of canal water in

sity of Sussex, 1983).

“‘Irrigation management in Taiwan.”

.

northern India: Conflict between traditional oatterns

Rapoport. A.. and A. Chammah, Prisoner’s Dilemma

Mimeo (Institute of Develooment Studies. Univer-

(Ann Arbor: University of Michigan Press, 1965). (1985a). _

Wade, R. H.. “On the sociology of irrigation statistics:

ment: The myth of the passive state.*’ Pacific View-

Wade. R. H.. “The market for oublic office: Whv the

point, Vol. 2-t. No. 1 (May 1983).

How do we know the truth about canal performance?” Agricultural Administration, Vol. 19, No. 2

and modern needs,” Economic Developm>nt and Cultural Change, Vol. 23, No. 1 (1974).

Runge, C. F.. “Institutions and the free rider: The assurance problem in collective action,” The Journal of Politics. Vol. 46 (1984). pp. 154-181.

Sundar. A.. “Modern techniques for management of irrigation systems: What can they do in the absence of commitment to manage?” Wamana (Bangalore: Indian Institute of Management. July 1984).

Vaidyanathan. A.. “Water control institutions and agriculture: A comparative perspective,” The Indian Economic Review, Vol. 20, No. 1 (January-June 1984).

Wade, R. H.. “Collective responsibility in the construction and operation of irrigation canals: The case of

Indian state is not better at hevelopment.” iorld Development, Vol. 13. No. 4 (April 19856).

Wade, R. H., “Common property resource management in South Indian villages,” Common Property Resource Management. National Academy of Sciences (Washington, DC: National Academy Press, 1986).

Wade, R. H., “The management of common property resources: Collective action as an alternative to pri- vatisation or state regulation”, Cambridge Journal of Economics, Vol. II, No. 2 (July 1987).

Wade, R. H., Village Republics: Economic Conditions of Collective Action in South India (Cambridge: Cambridge University Press, 1988).

Documents

The management of irrigation systems: How to evoke trust and avoid prisoner's dilemma