Agricultural Administration 19 (1985) 63-19

On the Sociology of Irrigation: How Do We Know the Truth about Canal Performance?

Robert Wade*

Institute of Development Studies, University of Sussex, Great Britain and the World Bank, Washington, USA

(Received: 6 November 1984)

SUMMARY

The methods used to collect irrigation statistics in India give rise to very large margins of error. The unreliability of the data throws into question the utility of careful economic analyses of irrigation projects. More importantly, it means that in the case of large surface-flow schemes canal of$ciuls cannot manage the system in the absence of reliable information about the system’s output; they can only administer the inputs. This is an important constraint on betterper_formunce. To improve the information system attention has to be given not only to the hardware but also to the incentives which bear on those who collect the statistics.

INTRODUCTION

It is widely believed that the performance of large, publicly-operated canal systems in most parts of the arid and semi-arid tropics is poor,1T2 and that there are good prospects for improving the productivity of canal irrigated agriculture, by ‘institutional’, ‘managerial’ or ‘technical improvements in the water delivery system. 3,4 It is argued here, however, that the margin of error in the statistics used to measure canal

* Present address: Room N1038; The World Bank, 1818 H Street NW, Washington, DC 20433, USA.

The views expressed are the author’s and must not be taken as the views of the World Bank.

63 Agriculmral Administration 0309-586X/85/$03.30 0 Elsevier Applied Science Publishers Ltd, England, 1985. Printed in Great Britain

64 Robert Wade

‘performance’ is commonly so large that one must be wary of any statement about the actual performance, or capacity utilisation, of canal systems. Since, even with a good information system, large canal facilities (irrigating several hundred thousand acres a year) are amongst the most complex of public sector enterprises to manage, it may be presumed that the absence of reliable performance data is an important reason for ‘poor’ performance-if in fact their performance is poor. It is a matter of urgency that more reliable information about performance be obtained.

Statistics are socially generated and socially appraised-their use for monitoring and control affecting both content and accuracy. To understand why the statistics on canal performance are so unreliable one has to examine the incentives of those who collect the figures and those who make use of them. If the perverse incentives now operating are not also changed, improvements in the hardware components of the information system are likely to have little impact, Moreover, cost/benefit calculations for existing or planned irrigation projects will continue to be subject to huge and unnoticed uncertainties.

PERFORMANCE MEASURES

Irrigation is an input to agriculture, and it seems sensible to use an agricultural measure of canal performance-total production from the irrigated area, or production per unit of water. The problem, of course, is that agricultural production from a canal command area varies for reasons other than the quality of water supply through the canal. It may accordingly be more sensible to use a hydrological measure of performance, treating the reliable delivery of ‘adequate’ or ‘stipulated’ amounts of water throughout the command area as the ‘output’ of the canal system. (The distinction between ‘adequate’ and ‘stipulated’ refers to a basic difference in the principle of water allocation between South and Northwest India, discussed in Ref. 4, p. 182.) It would be possible to place soil probes over the command area to relay a flow of information about soil moisture levels over the agricultural cycle or, perhaps, to use aerial or satellite photography for the same purpose.

However, such methods would be expensive. In any case, one needs a simpler index of performance, so that performance of one canal can be compared with past performance of the same canal or with that of others. Hydrological ‘efficiency’ is one such measure: the ratio of water quantity

Reliability of irrigation statistics in India 65

which reaches the fields (or crop root zone), over the water quantity let in at the head of the canal. Alternatively, it is often more useful to use the inverse (which may be called ‘relative water supply’): the ratio of the amount of water let in at the head of the system divided by the amount which reaches the crop root zone.5 A relative water supply value of 3 means that in order to get a given amount of water to the crop root zone (over the whole command area), three times that quantity must be let in at the head of the system, the balance being lost en route.

To calculate ‘efficiency’ or ‘relative water supply’ one needs to know the area to which the water is applied. In practice, the most usual measure of irrigation performance is simply ‘area irrigated per season or per year (with no qualification according to how ‘adequately’ irrigated). Performance is commonly then judged simply by the rate of change of irrigated area. Alternatively, this may be elaborated, especially in planning documents, by comparing actual irrigated area against ‘potential’ irrigated area (the ‘capacity’ of the system); a higher percentage of capacity utilisation obviously indicates better performance than a lower percentage. This may be done for individual canal systems; alternatively, states may publish statistics showing, year by year, actual irrigated area state-wide, against total potential created.6

Finally, the most complex measure which is likely to receive any practical application is the ratio of water quantity let in at the head of a canal per season, over the irrigated area per season. This can be expressed as X thousand mill ion cubic feet of water (TMCs) per season per 100 000 acres. If the irrigated area is standardised for paddy (unmilled rice), converting non-paddy acres to paddy acre equivalents, the resulting coefficient can be used to compare different canals. The measure can be further improved if it is compared against some standard of ‘reasonable’ performance. One might, for example, calculate how much water would be required to offset ‘potential evapotranspiration’ (PET)* over an average season on 100000 acres of paddy; then, calculate how much irrigation water would be required by subtracting average rainfall from average PET; then, make some ‘reasonable’ allowance for conveyance losses from the head of the system to the field; and finally, arrive at a figure of Y TMCs per 100 000 acres of paddy, as the standard of ‘reasonable’ * Potential evapotranspiration (PET) is the water potentially evaporated from the leaves of a crop and from the land or water in which it is growing. When total water supply (rainfall, soil moisture, irrigation) is enough to satisfy PET, plant growth is at or near maximum, other inputs being held constant.’

66 Robert Wade

performance. By comparing X TMCs (actual performance) against Y TMCs (reasonable performance), one can estimate the extent to which water use is ‘excessive’.

CANAL PERFORMANCE IN ANDHRA PRADESH

Let us carry out the calculation given in the last paragraph for canals on the uplands of Andhra Pradesh (South India). The intention is only to obtain a rough idea of canal performance and to illustrate the method of calculation; thus, some heroically simplifying assumptions can be made. Average total potential evapotranspiration over the 150 day irrigation season between July and November is 640 m m , or 2.1 feet, for Hyderabad.7 To meet potential evapotranspiration, 2.1 acre-feet, or 0.09 mill ion cubic feet of water per acre are needed; or 9 TMCs/lOO 000 acres. Assume that land preparation and puddling are met by utilisable rainfall. Then, if we assume-to be on the safe side-that rainfall (on average a total of 618mm between July and November) contributes nothing towards meeting PET, the total irrigation water required per 100000 acres is 9TMCs, plus a ‘reasonable’ estimate for conveyance losses. The 1972 Report of the Irrigation Commission’ estimated the extent of water conveyance losses in the alluvial plains of North India as 62 %: in other words, only about 38 ok of water let in at the head of a canal reaches the crop root zone. If we take this figure as ‘reasonable’ (and obviously to take an actual average as the standard of reasonable losses is conservative), a reasonable standard of performance would be 23 TMCs per 100000 acres, If such conveyance losses are thought too high to be ‘reasonable’, the reasonable coefficient might be put at, say, 20 TMCs. This is in fact the figure which has been taken for planning the big new projects on the uplands (such as Pochampad), and it has been calculated with little more subtlety than the method used here.

By this standard, the existing upland canals of Andhra Pradesh are not doing well. For example, Nagajunasagar Left Bank Canal irrigates (according to the Irrigation Department’s figures) 387 000 acres in the first (rainy) season (June to December), almost all under paddy, and has been using about 121 TMCs to do so, or about 3 1 TMCs per 100 000 acres. Irrigation started here in the late 1960s. Or take the Tungabadhra High Level Canal, where irrigation began in the mid-1960s. Currently, it irrigates 54000 acres in the first season, and 74000 acres in the second

Reliability of’ irrigation statistics in India 61

(dry) season, with most of the first season area under paddy and most of the second season area under non-paddy. If we make the conventional, time-sanctioned assumption that one paddy acre uses as much water as two non-paddy acres and reduce the second season area to paddy acre equivalents, we get a total irrigated area per year of 91000 paddy equivalent acres. Total canal water taken in at the head of the canal, over both seasons, is about 26TMCs. The performance coefficient is thus 29TMCs per 100000 paddy acres. Indeed, none of the post- Independence upland canals are operating at less than about 28 TMCs per 100 000 acres, against a ‘reasonable’ standard of about 20TMCs. (The source of these figures is evidence collected by the Government of Andhra Pradesh’s One Man Commission for Irrigation Utilisation, operating throughout 1981 and 1982.)

In this sense, water use is ‘excessive’, and since more land could physically be brought under irrigation if more water were available, there is potential for reducing the excessive use in the existing irrigated area and bringing new land under irrigation with the same stock of water. Crude though this comparison of ‘reasonable’ against ‘actual’ performance is, one should note that not even this exercise is made as a matter ofcourse by any government agency (whether executive or legislative). This point will be returned to later.

The Irrigation Department itself prefers to use the measure of ‘actual in relation to potential’ irrigated area as an index of performance. By this standard, and using the Irrigation Department’s figures, Andhra Pradesh is doing remarkably well compared to other Indian states. According to a recent statement by the Irrigation M inister, the potential created by June 1978, was 2 783 000 hectares, and by June 1979, 2756000 hectares were being irrigated, a lag of only 0.97 % . The All-India figures, by comparison, are 24 767 000 hectares and 21977 000 hectares, respectively; a lag of actual behind potential area of 11.7 % .” There is no cause for complaint about Andhra Pradesh’s canals here.

UNCERTAINTIES

The problem, of course, is to know what if anything these figures mean. The main concern here is with the measurement which either on its own or in combination with some other figure enters into almost all performance measures: actual irrigated area. First, though, a brief word about the

68 Robert Wade

meaning of two other important figures: ‘quantity of water taken in at head of the canal’, and ‘potential irrigated area created’.

‘Potential created’

There is, in fact, considerable uncertainty in the minds of irrigation engineers working on canal construction projects about what the term ‘potential created’ means, or more precisely, about when they can declare a certain area to be part of ‘potential created. It might seem sensible to take the construction of the outlet as the criterion: if the land planned to be irrigated under a given outlet is, say, 200 acres, that 200 acres may be included in ‘potential created’ when the outlet is constructed. However, since it has been observed that farmers often do not construct field channels of their own accord within an outlet block, and that, partly for this reason, substantial areas of the block do not in fact receive any irrigation, there has been support in government circles for imposing a stricter definition of potential created: only when a field channel has been constructed, at government expense, up to each sub-block of 5 hectares can that 5 hectare sub-block be included in ‘potential created. (This is now embodied in the Andhra Pradesh Command Area Development Act 1982.) In practice, irrigation engineers in charge of construction, anxious to show that the work is progressing at speed, are often inclined to take the off-take to the minor distributory as the criterion: when the off-take is constructed, even if the minor distributory itself and the outlets from it are not constructed, then the whole area intended to be irrigated under the minor is declared as potential created. One says ‘are often inclined, because the point is that the definition used in practice varies from one engineer to another. Therefore, it is by no means clear what aggregate figures of ‘potential created mean.

Water quantity

As for ‘quantity of water taken in at the head of the canal’, compared to the figures of actual or potential irrigated area, figures on this variable are likely to be relatively accurate. However, it is to be noted, first, that figures of water quantities within any one canal command (e.g. figures on how much water is reaching a certain sub-division) are vulnerable to deliberate falsification, either directly, through false gauge readings, or indirectly

Reliability of irrigation statistics in India 69

through accurate readings of a deliberately falsified gauge. The reasons why a sub-divisional engineer may wish to falsify the water quantities received and let down to lower sub-divisions have been discussed at some length elsewhere, and will not be repeated here.l’

Secondly, and more important for the monitoring of whole canal systems, the recorded figures of quantities released into the canal system may tend to understate actualquantities. Even when a river or reservoir is fed from a watershed wholly within one state, each canal is given a quota for the amount of water it can draw. But the incentive on the managers of each canal is to take more than the quota, because this makes for an easier life-and they may well be able to sell the extra water illicit1y.r r When the canal is regulated by an agreement governing division of inter-state river water, the pressures to understate the actual quantity drawn are still greater, and may come not only from the engineers on the spot but also from the state government itself. The state government may, for example, wish to start a new project in an inter-state river basin. It may order the engineers of existing canals within that river basin to understate their actual water intake, so as to be able to say to the Planning Commission, the Central Water and Power Commission, and other authoritative bodies governing India’s water resources, that water for the new project can come out of the state’s existing quota. After the new project is completed and water is found to be chronically short, various sorts of pressure can be mounted to get more water to the state; moreover, the politicians who benefit electorally and financially from approval and construction of the new project may then not be around to take the consequences of water shortage.

Finally, some canals are designed so as to pick up rainfall run-off and drainage water en route, so that in these canals the total water in the canal is not simply a function of how much is released at the head of the canal. In the unusual event that comparisons between water quantity and irrigated area are made, a common error is not to include the additional water taken in en route, which makes the canal’s performance seem better than it is. One reason for the error is simply that, in the typical case, no data on how much water is normally added to that which enters at the head have been compiled. Even figures of rainfall collected by different departments within the same small locality can differ by non-trivial amounts. Rainfall figures from the Irrigation Department and Agriculture Department, for the same place (rain gauges 1 km apart at the same elevation), for the June to December period of 1978,1979 and 1980, show

IO Robert Wade

the Agriculture Department’s total to be less than the Irrigation Department’s by 7, 7 and 11 %, respectively. I do not know if there are systematic pressures to falsify the rainfall readings.

Area irrigated

We saw earlier that according to the measure of ‘actual’ in relation to ‘reasonable’ water use per 100 000 acres, Andhra Pradesh’s upland canals are not performing well. However, those calculations were made using the Irrigation Department’s figures of irrigated area. Yet, virtually all published statistics on irrigated area come (via the state’s Bureau of Economics and Statistics) from the Revenue Department. To be more specific: all of Andhra Pradesh’s official’statistics are elaborated and presented by the Bureau of Economics and Statistics, and the Bureau obtains its figures of irrigated area only from the Revenue Department.

Table 1 shows the irrigated area under several upland canals according to the Irrigation Department figures and the Revenue Department figures. The comparisons are for the same year and season(s) in each case, and it is important to emphasise that the figures are meant to measure exactly the same thing: both the Irrigation and Revenue Departments claim to record as ‘irrigated’ any land which receives at least one wetting from the canal (it is not the case that the Revenue Department records only irrigated cropped area, and the Irrigation Department the irrigated planted area). To eliminate one source of variation it is desirable to concentrate attention on the figures for both first and second season figures combined, because there is great variation in the criterion used by the data collectors on the ground for ‘first’ and ‘second’ season. In principle, land planted before 30 September is to be classed as first season, land planted after 30 September is second season. But this is not how farmers think of the distinction; they normally think of the second season as beginning in late December or January, and in practice, the data collectors on the ground, themselves mostly from a farming background, often use this criterion too. So one source of variation between the two departments’ figures season-by-season is different definitions of which to include in first season and which to include in second. Even with this source of variation eliminated, the variation between the Irrigation and Revenue Departments is large: on the three newish canals, the Revenue Department figure is about 30-50 % of that of the Irrigation Department; on the old canal, 70-75 %. The latter figure is especially noteworthy,

Reliability of irrigution statistics in India

TABLE 1

71

Irrigation Department and Revenue Department Figures on Actual Irrigated Area

Canal Date Season” Irrigated area (acres) Rev.1 Years Irrig. since

Irrigation Retienue (%I irrig. Department Department began

Nagajunasagar Right Bank Canal 198081

Pochampad 1980/8 1

Nagajunasagar Left Bank Canal (Kummum district only) 198182

1978179

MN Canal 1979180

I98018 1

197819

One village under

1

1979/80 MNC

, 1980/‘81

1+2

1+2

800 000

180000

370 000 46 12-15

59 000 33 9

1 38 000 18000 47 12-15

1 189000 2 140 000

total 329 000 1 191000 2 138 000

total 329 000 1 207 000 2 136 000

total 343 000

1 1303 2 594

total 1897 1 712 2 1648

total 2 360 1 1647 2 765

total 2412

159 000 86 000

245 000 151000

86 000 I 237000 1

147 000 88 000

235 000 1

849 ! 18 ~

867 ; 896 126

1022 1 1101 j

206 1307 J

74

lOO+ 72

69

46

50+ 43

54

’ Seasons: 1; first season; 2, second season.

because government officials, talking about the problem of irrigation statistics, commonly say that there is discrepancy only on new projects, which falls away with repeated data collection. This seems to be only a very small part of the truth: i.e. the discrepancy falls from very big, to merely big.

Ifwe use the Revenue Department figures for irrigated area, then canal performance in upland Andhra Pradesh is disastrous. Nagajunasagar

12 Robert Wade

Right Bank Canal, for example, uses some 207 TMCs of water a year; if only 370000 acres are irrigated, this means a water use coefficient of nearly 70 TMCs per 100 000 acres. Pochampad works out at roughly the same. If true, this is nearly flooding; you would need a lifeboat to move around.

THE ORGANISATION OF DATA COLLECTION

The Revenue Department’s figures come from its lowest-level employees, the karnams; there is one karnam per village (normally he is a permanent resident of the village, and even today (in practice, though not in law) is recruited by hereditary right, as long as he is moderately literate). The karnam is responsible for feeding the Revenue Department with a great mass of figures about the village, including figures on the irrigated area, by crop type. On the basis of his figures, land taxes and water charges are levied: each acre recorded as irrigated is thereby liable to pay the flat rate charge for water per acre (different for paddy and non-paddy). If land is being irrigated in violation of the crop zoning laws, it is liable to a heavy financial penalty, of some nine times the normal water rate. Not surprisingly, it is to the mutual advantage of karnam and irrigator to conceal some of the irrigator’s irrigated area, especially if it is out-of-zone, and in return the irrigator will pay the karnam some fraction of what he would otherwise have to pay the government. The land under paddy may be entered in the records as rainfed, or as fallow, or it may be entered as irrigated non-paddy (since paddy pays a water charge 50 % greater than non-paddy). Similarly, much of the irrigated sorghum area is classed as rainfed sorghum. For these reasons statistics on the irrigated area under different types ofcrops are to be treated with great caution. In the village I know best (see Table l), the actual irrigated area in the first season of 1980/81, as estimated by villagers and myself (also by the Irrigation Department’s Work Inspector, who benefited from our attempts at estimation) was about 1600 acres, almost all under paddy; the Revenue Department’s karnam, however, declared 1100 acres in total, of which 700 acres under paddy and 400 under non-paddy. In another area, observation on the ground showed about 2000 acres under irrigated sorghum; Revenue Department records for the same area in the same year showed 69 acres under irrigated sorghum.

Reliability of’ irrigation statistics in India 13

On top of the incentives to falsify, there are the inevitable enumeration errors made when, as is often the case, hundreds of figures have to be added up by each kamam, without any mechanical aids.

Of course, the kamams’ figures are meant to be checked by superior officers, taking random samples for field-level checks. But in practice the Revenue Inspectors are very busy at the time such checks would have to be made, and are quite unlikely to feel inclined to forsake the comfort and shade of the village to make spot checks on a karnam’s returns in the field. In each taluk office* there is a Taluk Statistical Officer, who is an employee of the Bureau of Economics and Statistics. He aggregates up the karnams’ returns on irrigated area and presents the tahsildar (head of the Revenue Department in each taluk) with village-by-village and aggregate totals. These figures then guide the work of water rate collection (made at the same time as collection of land taxes and other taxes). If higher levels of the Revenue Department need figures on irrigated area, these are the figures that are sent upwards, and these are the basis of the figures published by the Bureau on the state’s irrigated area.

The Irrigation Department’s figures are collected village-by-village by the field staff (bankers and Work Inspectors). They have less close a tie with the particular villages for which they are responsible than do the Revenue Department kamams. They have no measuring devices with which to measure irrigated area. In practice they arrive at their figures by one of several ways: by standing on the canal bank and surveying the surrounding land by eye, survey map in hand, estimating the area which is growing an irrigated crop within easily recognisable boundaries; or, by calculating how much water has (or should have) passed out of each outlet, and deriving the actual irrigated area from the area which should have been irrigated with that quantity of water (according to traditional rule-of-thumb coefficients); or, simply by using the karnam’s own figures. The figures carry no liability for water rates, so farmers do not mind what they are. On the other hand, the figures do provide the basis for calculating the maintenance grant, and the irrigation staff are interested to maximise the size of the grant. Moreover, the figures can be used to provide a check on the Revenue Department’s figures (for this reason they are commonly aggregated by village units, not by units of water control, so as to use the same units as the Revenue Department).

Nevertheless, while there are good reasons for supposing that the

* The taluk is a sub-division of a district, and may contain 100 000-200 000 people.

14 Robert Wade

Revenue Department’s figures understate and the Irrigation Depart- ment’s figures overstate irrigated area, some m icro data suggest that the reality may be more complex. Village-by-village, it is not always the case that the Revenue figure is less than the Irrigation figure. A study of 16 villages under the MN Canal showed that in a three-year run of Revenue and Irrigation figures, 15 % of the pairings had a Revenue figure greater than the Irrigation figure. Further, the study showed that some villages regularly show a large discrepancy between the two figures, while others regularly show a small discrepancy; and the Revenue figure is more likely to be greater in the small discrepancy villages than in the large discrepancy villages. It is to m icro data of this sort that one needs to look in order to investigate the nature and causes of the unreliability in irrigated area figures.

Is no attempt made at governmental level to reconcile the two sets of figures? Intermittently, a District Collector (head of government administration in the district) may try to do so, depending very much on the extent of his personal interest in irrigation questions. He may ask his tahsildars to supply him with their totals for a given canal system, and ask the Irrigation Department for the same. But one disincentive for such action is especially worth noting: the tahsildars may have to ask the Taluk Statistical Officer to prepare figures specifically for the purpose, because, surprisingly, the Taluk Statistical Officer normally collects no figures on the basis of canal system by canal system. Indeed, on the form which each karnam has to fill in, there is no place for the name of an irrigation facility to be stated; he has only to distinguish how much area is irrigated from ‘canals’, ‘tanks’, ‘wells’, etc., not which canals, which tanks. This reflects the fact that the purpose of the kamam’s returns is to provide the figures on liability to land and water charges, and for this purpose it does not matter precisely which source the irrigation water comes from. The problem is that these same returns are then also used for aggregate statistical purposes, but do not enable important statistical distinctions to be made-such as how much area is irrigated from each separate canal system. Of course, if only one canal runs through a taluk, as is in fact common, it will not be difficult to calculate the area irrigated from that canal in each taluk. By requesting this figure from each tahsildar, the Collector can find out the Revenue Department’s total irrigated area under that canal, and compare it with the Irrigation Department’s figure. The point to emphasise is that a special effort has to be made to get comparable aggregate figures.

If the discrepancy is found to be large, the Collector may order what is

Reliability of irrigation statistics in India 15

called a ‘joint azmoish’ to be undertaken. In principle this means that each Revenue Inspector (the karnams’ supervisor) has to carry out a re- inspection jointly with an irrigation Work Inspector or Supervisor, and arrive at some commonly agreed figure (if time remaining in the present irrigation season is insufficient, the joint inspection may be ordered for the following year). This, of course, is a great nuisance for all concerned, and the temptation to avoid it is great. In practice, the tahsildar may agree with the Irrigation Department on a joint figure without a joint inspection. So in the case of Nagajunasagar Left Bank Canal, Kummum District (see Table l), when an over-zealous Collector ordered a joint azmoish after the first season of 1981182, the joint figure was declared to be 22 000 acres (compared to the Irrigation Department’s original figure of 38 000 acres, and the Revenue Department’s original of 18 000 acres). But this was, in fact, wholly a ‘desk compromise’, agreed at the level of the concerned tahsildars and their Irrigation Department opposites; it is especially important to note that neither in this specific case nor in joint inspections generally, does the ‘joint figure’ alter the figures on which water rates are assessed. Whatever the joint figure, water rates are assessed on the basis of the kamams’ returns, and indeed, no attempt is made to disaggregate down the joint figure to new figures in the records of each village. The function of the joint figure is to placate the higher-ups. For some canals an arrangement exists which makes it possible to avoid awkward Revenue-Irrigation Department comparisons altogether. Some canals (or clusters of canals) have a special Revenue Department officer (Special Collector’) assigned to look after revenue-related business, in order to take the load off the ordinary Revenue Department staff. This officer may find it worthwhile regularly to prepare figures on irrigated area for his project (or projects)-whereas normally, as noted, the Revenue Department does not prepare figures project by project as a matter of course. One would expect that since he is a Revenue officer, he would base his project-by-project figures on Revenue Department figures-those sent on to him by the tahsildar and Tuluk Statistical Officer. It makes life much easier for all concerned, however, if he simply uses (without identifying their source) the Irrigation Department figures. Then, if the Collector demands Revenue Department project-wise figures to compare with the Irrigation Department’s figures, the Collector will be satisfied, and no joint azmoish will be needed. Nobody need know that the special officer’s aggregate figures do not tally with those he has got from the taluk revenue and statistical hierarchy.

There is no higher point above the district in the executive or legislative

76 Robert Wade

branches of government at which the discrepancies might in the normal course of events be revealed. In particular, it should be noted that in the documents on the Irrigation Department’s activities which go before the Legislative Assembly, the Department naturally uses its own figures of irrigated area and of potential created. As we have seen, using these figures the performance of Andhra Pradesh’s irrigation canals can be made to appear very good indeed. Not only might the irrigated area figure be inflated, but for this spec$c purpose the potential created may be deflated, so as further to reduce the gap. There is no way that the legislative body is likely to compare the various figures of potential created-some made to show the Irrigation Department’s construction work proceeding at speed, others to show only a small gap between potential created and area actually irrigated. In the same way, the Revenue Department’s special officer for a certain project can, as we have seen, break the line of aggregation up from lower units of the Revenue Department, for no one is going to find out, no one is going to ask where exactly his figures come from.

There is a diffuse awareness in government as well as academic circles that irrigation statistics are unreliable. Responding to it, the Bureau of Economics and Statistics has attempted to make a survey of paddy irrigated area with its own investigators, to obtain an independent estimate of how much area is being concealed. But the data, collected for most years since 1976, have not yet been analysed because of the pressure of other work. One senior statistical officer admitted he was not confident in the accuracy of the results, which seem on casual scrutiny to suggest only 224 % concealment; he had had reports that his field investigators (many of them graduates, doing exactly the same job, with no promotion prospects, for the past 10 to 20 years) were in fact taking the karnams’ figures, instead of undertaking the more laborious work of making their own.

CONCLUSIONS

The data presented here suggest that large discrepancies exist between the two possible sources of information on canal irrigated area (even under long-established canals). The argument made about the institutional incentives on those who collect and present the data suggests that the

Reliability of’ irrigation statistics in India II

errors tend to be systematic rather than random-Revenue Department’s being downwardly biased, Irrigation Department’s being upwardly biased. Clearly the question of the systematic or random nature of the errors needs to be further researched, in order to find out the extent to which the errors add up rather than cancel out and the implications of the errors for conclusions about the benefits of existing and new irrigation (conclusions which perhaps would be different than if one believed the reported data). It follows that much of the effort that goes into analysing, evaluating the performance of existing canal systems by economists and others may be misguided, if existing data on variables such as irrigated area and ‘potential created’ are used. It is said that statistics on area irrigated from minor irrigation projects (‘tanks’) are even more dubious than those for large systems. 6 As one senior statistical officer summed up the situation, ‘Anyone who tries to use these figures of irrigated area is, frankly, a damned fool’.

If the utility of careful economic analyses of irrigation is thrown into question by the unreliability of the data, so also is the possibility of better management of irrigation systems, and of better monitoring of the canal managers. Effective management systems depend on a feedback of information from the ‘outputs’ to the ‘inputs’ of the system. However, if data on outputs are not known with confidence, the attention of the managers is likely to remain concentrated on control over the ‘inputs’ -on, in the irrigation case, control over expenditure, over the use of petrol, over the placement of personnel, with little attention to where the water is going, how much is being used, and how much area is being irrigated and to what standard of adequacy. In effect, the systems cannot be ‘managed, only ‘administered’. r 2 This of course is a familiar problem for any public sector enterprise which does not depend on sales for its budget. It is especially pressing in irrigation, because (i) irrigation is so potent a means for accelerating the rate of agricultural growth, (ii) it is absorbing a large proportion of the states’ development budgets, and (iii) better information about the ‘outputs’ of canal systems could be provided without too much difficulty once the need was recognised.

Whatever improvements in the information system within the Irrigation Department were made, however, an independent monitoring organisation should be established, whose job should be limited to measuring the ‘outputs’ of each canal system.4’12 Its reports would have to be made public. In this way, one of the major weaknesses in the current canal control system could be reduced: the lack of any monitoring of the

78 Robert Wade

performance of each canal system outside the Irrigation Department, and in particular, the lack of such monitoring from the legislative arm of government on a routine, non-particularistic basis. Some practical suggestions as to how the monitoring function might be carried out can be derived from a study of how it is done on South Korean and Taiwanese canals.13

ACKNOWLEDGEMENT

I am grateful to Syed Hashim Ah, IAS, for help in collecting some of the figures on irrigation performance presented here, and for his unfailing encouragement of the research.

REFERENCES

1. Wade, R., Water to the fields: India’s changing strategy, South Asian Review, S(4) (July/October 1975). Reprinted in: Irrigation and Agricultural Development in Asia: Perspectives from the Social Sciences (Coward, E. W. (Ed.)). Cornell University Press, Ithaca, New York, 1980.

2. Wade, R. and Chambers, R., Managing the main system: canal irrigation’s blind spot. Economic and Political Weekly (Bombay), XV(39) (September 1980). Review of Agriculture.

3. Wade, R., Employment, water control and water supply institutions: South India and South Korea, Asian Employment Programme Working Papers, Asian Regional Team for Employment Promotion (ARTEP), ILO, Bangkok, also IDS Discussion Paper No. 182, 1982.

4. Wade, R., The World bank and India’s irrigation reform. Journal of Development Studies, M(2) (1982).

5. Levine, G., Management components in irrigation system design and operation, Agricultural Administration, 4( 1) (January 1977).

6. Bhaduri, A.,.A critical approach to irrigation statistics in India. Int. Sot. Sci. J., XXVIII(3) (1976).

7. C. W. Thornthwaite Associates, Average Climatic Water Balance Data of the Continents, Part II, Centerton, New Jersey, 1963.

8. Ministry of Irrigation and Power, India, Irrigation Commission, Report (4 vols), New Delhi, 1972.

9. Wamana, Report on Fifth Conference of State Ministers of Irrigation, Bangalore, November 1980. Indian Institute of Management, Bangalore, October 1981, Vol. l$ No. 4.

Reliability of irrigation statistics in India 79

10. Wade, R., The information problem ofsouth Indian irrigation canals, Water Supply and iManagement, 5 (1981).

il. Wade, R., The system of administrative and political corruption: canal irrigation in South India, Journal of Development Studies, lS(3) (1982).

12. Seckler, D., The new era of irrigation management in India, mimeo, Ford Foundation, New Delhi, January 198 1.

13. Wade, R., Irrigation and Agricultural Politics in South Korea, Westview Press, Boulder, Colorado, 1982.