DISCUSSION DRAFT
This internal working paper is prepared for
Staff Use Only. The views expressed are not
necessarily those of the World Bank.
IMPROVING IRRIGATED AGRICULTURE:
INSTITUTIONAL REFORM AND THE SMALL FARMER
AGREP Division Working Paper No. 34
Prepared by:
Daniel W. Bromley
University of Wisconsin
for
Economics & Policy Division
Agriculture & Rural Development Department
December 1979


PREFACE
Thi,s represents an early attempt to develop a conceptual model of
farmer behavior in irrigated agriculture for the purpose of then deducing
logical programs for improving the,economic situation of smaill-scale
irrigators. -I have relied on three major sources for ideas. The first
is my own experience in working with small-scale irrigators in the
Philidp;iqes,, and the Dominican Republic, and in talking with irrigation
administratc.'s in Thailand and Malaysia. Additionally, my own research
efforts concern irrigation problems in Pakistan were valuable.
The second source is the published literature listed at the end of
this report. The section entitled "Literature Cited" contains that
material directly referred to here; the section entitled "Additional
References" contains material surveyed during earlier work on irrigation.
The third source of ideas--and usually the most helpful--is the
personal interviews and discussions with a number of individuals knowledge-
able about irrigated agriculture in the developing countries. These
discussions have occured over the past 5 years in 7-8 developing countries,
and more recently (during the period May-July, 1979) involved researchers
and i.rrigation administrators in England, India, Malaysia and Thailand.
In this regard I am grateful to: Tony Barnett and John C. Harriss of the
School of Development Studies at the University of East Anglia, Norwich,
England; Anthony Bottrall of the Overseas Development Institute, London;
Vira Chankong, Pradit Nornongkol, Suruvuth Pradisthananda and Sacha
Sethaputra of Khon Kaen University, Thailand; Walt Coward of Cornell Univer-
sity; K. Gopal4krishnaya, Waheeduddin Khan, M.N. Kulkarni, and K.K. Singh


of the Administrative Staff College, Hyderabad, India; Yujiro Hayami, Tokyo
Metropolitan University; Don Parker, Indiana University; P.C. Sun, Kunio
Takase, and S.C. Hsieh of the Asian Development Bank, Manila; Frances and
David Korten of the Ford Foundation, Manila; Roberto Lenton, David Seckler,
and Jeff Romm of the Ford Foundation, New Delhi; Donald Taylor of the
Agricultural .Development Council, Kuala Lumpur; Sam Johnson of the Ford
Foundation, Bangkok;


1-1
CHAPTER 1. INTRODUCTION
Irrigation is a cultural adaptation to situations in which nature
provides rainfall in too meager quantities for cultivation of certain
valuable crops. The shortage may have a seasonal component, or it may be
too arid through the full year. Whatever the case, the manipulation 'f
surface and groundwater has become a dominant part of man's relentless
pursuit of enough to eat. In Table I-1 we see a tabulation of the
cultivated and irrigated land by region of the world in 1970. While
irrigated lands accounted for only 14 percent of the total cultivated
area, the majority of this land is devoted to high-valued and/or crucial
food crops. The regions of the world of interest in this report--Asia,
Latin America, and Africa--contained (in 1970) approximately 55 percent
of the world's cultivated lands, and approximately 80 percent of the
irrigated lands. Within these three regions Asia contained 60 percent of
the cultivated land, and 90 percent of the irrigated land. While irrigation
is locally important in several countries of Latin America and Africa,
it is.the dominant form of agriculture in most of the countries of Asia.
The interest in this report is with small farmers and irrigated
agriculture. We are not interested 4n-,small or large irrigation projects
or systems--our interest is with the farmers who depend on water within
those systems, especially small farmers. It is our intent to develop a
theoretical framework which allows us to analyze the current behavior of
small irrigators and to relate that behavior to the way in which their


1-2
TABLE 1-1, CULTIVATED AND IRRIGATED LAND BY COUNTRY AND REGION,1970.
Land (million ha)                  No. of persons
Irrigated   per irrigated
Country or Region  Cultivated    Irrigated   land (%)        hectare
Europe            145.0          12.3         8.5          37.6
EEC,                    52.4         5.4         10.2         46.6
Eastern Europe          46.0         2.7          5.8         38.9
U.S.S. n.         232.6          11.1         4.8          21.9
North America     236.1         16.2           6.9         14.0
U.S.                   192.3        15.4          8.2          13.3
C-anada.                43.8         0.4          0.8         60.0
Oceania            47.0           1.6         3.4          11.9
Australia &
New Zealand             45.4         1.6          3.5       . 9.4
Asia              463.0         145.7         31.5         14.1
Japan                    5.4         2.G         48.2         39.2
China                  111.2        76,0         68.3          10.0
India                  164.6        27.5         16.7         19.5
Indonesia               18.0         6.8         37.8         17.1
Latin Amrerica    118.9          10,4         8.8          27.2
Mdexico                 23.8         4.2         17.6         12.1
Brazil                  29.8         0.5          1.6        202.2
Africa            214.0           6.3         3.0          54.3
World         1,457.0        203.6         14.0         17.8
Source: Wortman and Cummings [1978].


1-3
irrigation supply system operates. We will argue that their behavior
in certain important respects is directly related to the way that an
irrigation system functions (or fails to function). This model wi-1-1
provide the conceptual rationale for programs to alter the operation of
irrigation systems. That is, we are not interested in improving irrigation
systems merely to make them more "efficient" in,their use of water. Our
interest here is one of working on an important element in the decision
structure of a small farmer; we seek to improve the operation of irrigation
systems because that will improve the operation of small farmers, because
that will provide more food for them and for their country, and because
that will provide them a better diet and an increased level of income.
The irrigation system becomes a proximate goal--an instrumental variable--
which is to be manipulated for the benefit of the farmer and the nation.
Any discussion of irrigated agriculture must commence with a discussion
of the nature of the relevant decision unit. In conventional economics
the concept of a finn denotes a decision unit over which fixed and variable
factors of production are manipulated for the benefit of the "owners" of
the firm. The presumption being that the owners are in full control of
the relevant variables such as fertilizer, water, seeds, labor use, planting
dates, haywest dates, sales decisions, and so on. This does not mean
that each firm must have absolute control over all variables. What it
does mean is that firms in similar industries have similar control over
the same general variables.
Our conventional agricultural assistance holds this picture of the
farm firm, even when we deal with tenant farmers who operate at the mercy


1-4
of a powerful landlosd. In these cases we are concerned with ways in
which that control exerted by the landlord influences the decision makin.g
of the tenant farmer. While there is an extensive theoretical literature
on landlord-tenant relations, we do not have anything comparable in
irrigated agriculture. Those writers who deal with irrigation are aware
of the special problems, but there has been little integration of this
irrigation literature with the conventioiiala'gricul:t'ur al' develop.m:nt
literature. The irrigation literature is largely concerned with illustrating
the poor water management practices in effect around the world, and with
suggesting organizational solutions to these problems. The agricultural
development literature generally treats irrigation water as just another
input along with labor and fertilizer. There is rarely any recognition
of the special nature of irrigated agriculture.
As will be demonstrated in Chapter II, irrigation introduces an
important interdependence among farmers on a system (project). This
physical linking renders traditional development assistance quite inappro-
priate at times. This interdependence comprises the essential difference
between irrigated and dry-land agriculture. When interdependent farmers
must operate in a world of institutional uncertainty we have yet another
complication in our conventional models of agricultural development.
This is discussed in Chapter III. Few treatments of small-farmer develop-
ment deal with institutional uncertainty as we will here. Given this
uncertainty--and the usual risks which attend agriculture--it is important
to understand the ways in which small farmers adjust their behavior. This
is the subject of Chapter IV.


1-5
In Chapter V we will offer a brief discussion of what development
will mean in the present context, with a direct link to institutional
uncertainty and the adaptive response of farmers. Chapter VI is devoted
to a discussion of irrigated agriculture in various regions of the world.
This discussion will highlight the,various performance problems of the
irrigation systems under consideration and the behavior of farmers.
The examples-were chosen because they illustrate important issues raised
in the earlier chapters.
In Chapter VII we will focus on implications of the foregoing for
improvements in irrigated agriculture for the small farmer. Throughout
our focus will be on programs to improve the economic position of small
farmers, whether they are located on small systems or on large systems.


II-1
CHAPTER II. IRRIGATION AND THE SMALL FARMER
In .this chapter we will be concerned with the economic environment
of the small irrigation farmer. The presentation will focus on the essential
economic fact of irrigation--that of the physical interdependence of
farmers linked by an irrigation canal and their joint use of the scarce
water.
A. A MODEL OF FARMER INTERDEPENDENCE
We will build upon the literature in externalities to develop a
model of physical interdependence among farmers situated on an irrigation
project.   This physical interdependence is in contrast to the normal type
of market interdependence we take for granted. In the usual notion of
interdependence, all farmers are linked together in certain input and
output markets by virtue of the fact that they all bid for the same factors
of production, or must sell their output in the same markets. If the supply
Throughout, the term "project" will refer to a group of farmers located
along a watercourse when all of them are served by one major outlet. That
is, if a large canal carries water from a river and along that canal there
ar.e 41major outlets to secondary canals, then we will refer to each of the
four groups as separate projects. Interdependence between or among projects
will thus refer to reliance of several of the groups of farmers on the
quantity of water in the major canal.


11-2
of seeds or fertilizers is less than perfectly elastic then one's purchases
reduce the quantity available for others--and this may drive up the price
that subsequent buyers must pay. Similarly in output markets the increased
production by several farmers will drive down the product price such
that other producers are made to suffer an income loss. Of course in our
theory we usually assume away these interdependencies. Or, if we do not
assume them away we label them pecuniary externalities. A pecuniary
externality is a price--or cost--manifested change in the economic environ-
ment of firm i because of the actions taken by other firms in the same
factor or product markets as i.
Pecuniary externalities are said to be of no economic interest--in a
policy sense--because they merely reflect the workings of the market, and
because they represent transfers of income instead of efficiency gains
or losses. In the above example of factor price changes, the induced
price increase arising from greater competition for a limited supply of
factors merely transfers income from buyers of factors to suppliers of
factors. These new rents to factor owners supposedly signal a greater
interest in the production of said factors and will--in theory--eventually
result in their increased availability. Once this occurs their price will
supppsedly-drops
On the marketing side, if more farmers grow wheat than previously
thereby lowering its price, the gainers are consumers of wheat. If
previous wheat growers suffer an income loss because of the new lower price
this is offset by the income gain of the consumers of wheat. The


I1-3
efficiency of the economic system has not changed, but the income released
from, its previous commitment to wheat may now be spent on other items.
But the sort of interdependence we have in mind here--and the
foundation of the model of farmer interdependence--is referred to as
technological externalities. The term to be used for the remainder of the
paper is physical interdependence, and it should be understood that the
interdependence refers to that created by the joint use of water resources
as factors of production. We will elaborate on this interdependence by
presenting a taxonomy of ways in which irrigators are physically linked.
The typical factors of production in irrigated agriculture are
land, seed, fertilizer, pest control, labor, capital, rainfall, temperature,
solar energy, and irrigation. Of these, rainfall, temperature,, and
solar energy are beyond the control of the farmer. Land is a fixed factor
in most planning horizons, leaving the farmer to control the nature and
the levels of seed, fertilizer, pest control, labor, capital, and
irrigation water. Of these factors only two--pest control and irrigation
water--are of a nature that physical interdependence among farmers will
arise.
The example of pest control is obvious; if farmer A sprays for leaf
hoppers then neighbors may reap some of the benefits feom these efforts.
But our interest here is in the physical interdependencies arising from
irrigation. There are two types of interdependence in irrigated agriculture,
those based on drainage of irrigation water, and those based on supply
considerations. First consider drainage.


I1-4
In those irrigated agriculture situations where supply is not subject
to farmer interdependence--say where water supply relative to the demand
is not scarce--there will be interdependencies arising from the fact that
tail water from farmer A will be carried to "downstream" farmers. The
impacts could be several. Such tail water might carry toxic compounds
and hence poison downstream drinking water. The tail water could be
saline, thereby ruining downstream water quality for agriculture. Or,
the tail water may simply not be wanted when it arrives and thus represent
flooding. Another aspect of this interdependence would arise when a
drainage ditch has limited capacity to remove water and its use -by farmer
A means that the tail water from farmer B's field cannot be carried away.
It is important to recognize the interdependencies which can arise
quite aside from water supply-issues. But it is the supply interdepen-
dencies which attract the most attention. There are four major categories
of supply interdependence: (1) individual pumps; (2) individual diversion;
(3) joint pumps; and (4) joint diversion. Each will be discussed in turn.
Individual Pumps
There is a general impression that when farmers have their own
pumps.the supply interdependencies disappear. This is only true when
the supply of water is adequate given the demands for its use; in this
sense then, we have a similar situation to surface water. It emphasizes
the point that interdependence among farmers for water is a function of
supply and demand rather than the fact that the water is below the ground
or above it.
When farmers have their own pumps there are three possible situations.
Farmers pump from groundwater aquifers, farmers pump directly from a


11-5
river, or farmers pump from a canal. When groundwater is being utilized
the Extent of the interdependence is the most difficult to discern.
Only with very precise monitoring of aquifer recharge and drawdown is it
possible to know whether or not there is true interdependence among
farmers pumping from the aquifer. This interdependence would show up in
terms of higher pumping costs as the water table recedes, and in periodic
requirements that wells be deepened.
When farmers pump directly from canals or rivers the nature and
extent of interdependence is more apparent. However, as before if water
supply is adequate vis-a-vis the demand placed on the water there will
be no problem.
Individual Diversion.
When individual farmers divert water directly from a river--a rare
event except in small-scale mountain valley agriculture--there i:s.an
opportunity for water scarcity among farmers.
The more common situation is whEre individual farmers divert water
from a canal--where."canal" here can mean a major watercourse, a secondary
canal, or a tertiary canal (ditch). When each farmer has a turnout from
the water source, the sort of interdependencies which arise are the type
that have Ceceiv,ed the bulk of the attention in the literature on
iFrigited agriculture. Here, interfarmer conflict over scarce supplies
is the'major problem.
Joint Pump
When we turn our attention to joint pumping arrangements we add an
additional, dimension for conflict among farmers arising frof their


11-6
interdependence. That is, not only do we have the types of interdependence
discussed immediately above, but we also have the interdependence which
arises from the jointness in supply at the pump set/stilling basin.
This would be the same regardless of whether the source is groundwater,
a river, or a watercourse through a project.
Joint Diversion
The final category of farmer interdependence concerns the joint
diversion of water from rivers or a canal. By joint diversion we mean
a situation in which water is diverted from its source and then shared
among a number of farmers. The most obvious example is found in*rice
culture where water moves from paddy to paddy over some considerable
distance. Here each farmer does not withdraw from a canal or ditch
but instead depends upon the movement of water from an "upstream" paddy
being farmed by another person.
Hence we find that the scarcity of water is a necessary condition
for farmer interdependence, and that such interdependence is the basis
of inter-farmer conflict over water. Our interest here is confined to
those situations in which conflict exists. For where water.is adequate
vis-a-vis the demand for water we do not have an irrigation problem. We
may need better -seeds, more -fertilizer, better pest management, improved
crop production practices, and the like. But is is not--by definition--
an irrigation problem.
How might one express a basic interdependence among farmers in more
formal terms? One way is through the specification of production functions
for two or more farmers. In what follows we will detail the case for


I1-7
two interdependent farmers; it should be kept in mind that this can be
generalized to n farmers.
Start by considering two farmers mutually reliant upon limited water
for irrigation. We can write the production function for farmer A as:
YA = Y (SA, FA ,    LA, WA  KA  TA
where  SA  is the seed varieties used
FA  is the application rate and timing of fertilizers
PA  is the pest management strategy of the farmer
LA  is the labor input
KA  is the capital input
WA  is the quantity and timing of irrigation water
TA  is the land base of the farm
Likewise for farmer B we have a production function given by:
Y = Y (SB, F , P , LBI W8, K, TB)                    (2)
where the symbols are as defined for farmer B.
However, this formulation fails to recognize the basic interdependence
which exists between the-two farmers over the scarce commodity water. For
we know that the water use by the "upstream" farmer (A in this case) has
important implications for water availability to farmer B. We would thus
need to reformul,ate the two production functions as:
YA = Y (SA, FA  PA, LA, WA, KA, TA)                 (3)
Y = Y (S , F , P , LB, h(WA) KB, T8)                  (4)
where WB has now been replaced by the expression h(WA
This indicates that water use by B is a function of water use by
farmer A (and all other "upstream" farmers). If we consider only the
two-farmer case, then it is possible to express water receipts for farmer B as:
W I        WA                             (5)
w   e   A    B     - W
where  W  + W     W4, and W4 represents the total available water.


I1-8
The vulnerability of the last farmer on a watercourse is emphasized
by writing that farner's water receipts as:
Wn       WA _8 W          n-1                      (6)
where there are n farmers on the watercourse.
It is readily apparent in equation (4) that h(W A) is not a variable
under the control of farmer B. This violates conventional economic theory
where a firm is defined by a set of economic variables:over which it has
exclusive control. In a sense then, farmer B (and all other farmers who
do not control the quantity or the timing of their water receipts) does not
fit the classical notion of a firm. Indeed, it goes beyond this simple case.
If we introduce the fact that seed varieties, fertilizer    applications and
the use of labor are functions of the expected water receipts for farmer B
we could write the production function as:
Y     B WA )    A   PB, LB, h(WA)K B, TB)               (7)
where SB = f(WA)
'and  FB = g(WA)
Before moving on to an elaboration of the above interdependence model
it warrants mention that water control comprehends quantity timeliness and
reliability of water receipts. The above formulation lends rigor to the
notion of quantity, but ignores timeliness and reliability. This omission
is necessitated by the desire to keep the formulation simple.
-The nature of the inter-farmer conflict over limited irrigation water
can be illustrated using the conventional Edgeworth diagram of Figure 11-1.
If we assune that both farmers are efficient, and that farmer A is the up-
stream farmer, then farmer B must take what water is left after A's use, and
also adjust other farm inputs accordingly; a failure to adjust other inputs
would mean that B is wasting seed and fertilizer when there is no water with
which to gain maximum advantage from these other inputs.


11-9
wo
150
10100
505
A                                                    A
FIGURE I1-1. EFFICIENT WATER ALLOCATION.


II-10
Under these assumptions we would find ourselves at point N in the
figure. Here, of the N of water available to A and B, farmer A is using
A                           B
WA.an.d farmer B must be content with WB. Assuming efficiency on the part
of both farmers, A would use IA of other inputs (some composite of seed,
B
fertilizer, pest control, and labor) while farmer B would use 10. The
absolute limit on water availabilty at  means that B can only get more
water if A gets less.2 In normal theory the locus of points L-M-N is
referred to as the contract curve, implying that once resource allocation
is such that productive efficiency has been attained--as evidenced by
tangency of the two farmer's isoquants--further movements can only come
about because of contracting between the two parties. When the inputs
move in conventional markets this notion of contracting is perhaps a
satisfactory one; after all one party might very easily approach the
other and seek to acquire some.of the latter's factors of production.
2 The-vertical dimension on the Edgeworth Box imposes an upper limit on
the quantity of other inputs available to farmer A and B. Unlike water,
this is not a realistic situation since A & B should be restrained only
by their respective budgets rather than by some absolute physical quantity.
However, in Figure 11-1 it is possible to imagine the vertical dimension as
representing the maximum quantity of other inputs which A & B would purchase
constrained by their fixed land base and by the aggregate quantity of water
available to them (W).  Hence the horizontal dimension of the box is given
by physicai availability*of water, while the vertical dimension is given by
the technological relationships among land, water and other inputs.


II-*11
However irrigation water in the developing countires is quite another
matter. Rarely is it traded in markets analogous to those for fertilizer
and seeds; indeed in most countries water markets are outlawed. The
standard practice is to allocate water--in a legal sense--on the basis 6f
land to be irrigated. The defacto allocation may be quite different from
the de jure allocation. The allocation depicted in Figure II-1 is de facto
rather than de jure, and hence it is more appropriate to refer to the locus
L-M-N as the conflict curve rather than the contract curve.
Reference to a conflict curve raises immediate questions about the
location of the two farmers along the locus L-M-N. In conventional analysis
relative prices between inputs will determine their mix in the production
process. But irrigation water is not priced on a per unit basis, and
because of the difficulties in accurately measuring water on irrigation
projects it is probably unreasonable to expect thatit could be so priced.
The defacto position along the conflict curve is--in simplest terms--a
function of the relative power of farmer A and 8.3 When power is intro-
duced into the-anaylsis the term conflict curve becomes more understandable.
For while we have difficulty treating power in a rigorous fashion in our
economic theory, there is little doubt that the exercise of power in
irrigated agriculture is pertinent to water allocation.
By power it is not necessary to have reference to brute strength or
massive land holdings--though both surely do not hinder those so endowed.
In irrigated agriculture "power" can be something as innocent as position
along the main canal or a-tributary. Recall from equation (6) that any
farmer's water reliability (quantity and timeliness) depends upon the
This has been emphasized in Bromley, Taylor and Parker [1980].


II-12
number of irrigators upstream from the farmer in question. They may not
exercise great power at any one moment, but the downstream farmer is
exposed to their exercise of discretion which comes from their more
favorable position vis-"a-vis the water source. In the language of John
R. Commons the .upstream farmer has liberty which the downstream farmer
does not have. What the downstream farmer has is xoosre to the actions
taken by the upstream farmer(s). If they do not hoard water then the
downstream farmer is not harmed. If they do take extra water then the
downstream farmer is harmed. The downstream farmer is vulnerable to the
water management whims of those upstream. We have elsewhere referred
to such a farmer as the least-advantaged farmer on a system [Bromley,
Taylor and Parker, 1980].
The foregoing discussion has focused on farmer interdependence,
and the notion that some farmers are at the mercy of their upstream
neighbors. By assuming efficiency in production on the "project" one was
left with a problem of fairness or equity or justice among (or between)
farmers. Economists are reluctant to say that farmer B should get more
water than at present since it will require that farmer A get less. With-
out knowing which farmer "deserves" more water--that is without revelation
of the-prevailing social welfare function--we are at an analytical cul-de-
sac.. This'makes it easy to dismiss the problem as one of politics and
outside the domain of economics. However, such a conclusion is a mere
artifact of the assumptions in the foregoing model. For it is rather
more plausible to argue that the type of water problem under discussion
creates important inefficiencies within agriculture. To locate ineffi-
ciencies is always to be assured of gaining the attention of an economist.


Consider Figure 11-2. Here we do not assume that both farmers are
happily located on the efficiency locus L-M-N (the conflict curVe).
Indeed we assume that they are not on it, and that they will likely never
attain it. Why might that be? To understand fully the inefficiencies
inherent,in the prevailing irrigated agriculture of the developing
countries we mfst remind ourselves of the basic fact of agricultural
production--it is spread over a considerable period of time, it is sequen-
tial, and with few exceptions there is "no going back." Add to this the
fact that each decision is made on the basis of certain assumptions about
the state of nature when the next decision must be made and the foundation
for an important problem has been laid.
When decisions are made about the amount of land to be cultivated
during the coming season one variable taken into account is water
availability. When decisions are made about which crop(s) to plant one
variable taken into account is water availability and the probable timing
of wa.ter receipts. When decisions are made about fertilizer purchases
one variable taken into account is water availability and the probable
timeliness of water receipts. Now, when this sort of planning is under-
taken by our upstream farmer (A) there is one set of expectations about
each of these important issues. When the same questions a-re pondered by
our downstream farmer (B) there is another set of expectations. We are
quite .safe in stating that the variance associated with the estimates for
farme.r B are several orders of magnitude greater than the variance associated
with A's estimates. We are also safe in asserting that the variance of
these estimates is an increasing function of the number of farmers upstream
from the particular farmer under discussion.


11-14
Without getting into the literature on risk aversion among traditional
farmers we can safely argue that such variance in the above critical
variables pertaining to water quantity and timeliness will have a chilling
effect on certain farming practices of the "downstream" farmers. This
has been*extensively documented ,in the literature.4 How might this
uncertainty show up in terms of economic efficiency?
The inefficiency arises because of the different degrees of control
over requisite inputs on the part of A and B. If farmer A is the upstream
farmer then we assume that he/she can plan the use of other inputs and
water on the basis of previous experience and cropping desires for the
coming season. Assume that this input combination would place farmer A
A
at point M in Figure 11-2. Assume that A indeed purchases IA of other
inputs during the cropPin.g :s.easbn. However, instead of using W of
p
irrigation water which is the correct amount given.the level of other
inputs being used, farmer A falls into the predictable pattern of taking
a little more than necessary, just "to be sure." Assume that farmer A
A
actually takes.WA of irrigation water.
Now what about farmer B? We can assume that history has taught B
to be cautious in the purchase of water-dependent inputs because of the
B
great unreliability of water receipts. Assume that B therefore uses 10
of other inputs during the course of the cropping season on the assumption
A               B
that'farmer A will use WA leaving only W, for farmer B. However, we see
that $ was overly cautious since A only took WA, leaving WB for B to use.
We can reasonably assume.that B will take what water is available.
4 This literature is reviewed extensively in Bromley, Taylor and Parker
[1977], and is discussed briefly in Bromley, Taylor and Parker [1980].


II-15
eL
I               I
5050
A                                W                   A
F
FIGURE 11-2. INEFFICIENT VIATER ALLOCATION AND AGRICULTURAL PRODUCTION


II-16
Thus we see that farmer A ends up at point K in Figure 11-.2, using
IA of other inputs, and WA of water. Farmer B ends up at point J in the
figure, using 1B of other inputs, and W of water. Notice that neither
farmer is on the efficiency locus L-M-N and that total production between
the two is less than it could be if they were at point H (180 from J and
K versus 200 from H).
While the example here has depicted a one-period situation, it is not
difficult to imagine that this sort of process occurs annually as the
interdependent farmers attempt to plan their own agricultural enterprise, and
anticipate the actions of other farmers with whom they are tied via
irrigation water. As indicated previously, given this type of interdepen-
dence and uncertainty, it seems reasonable to suppose that only by accident
would the two farmers be found on the efficiency locus. When the situation
is generalized to a large number of farmers on a watercourse the accidental
nature of productive efficiency is underscored.
The obvious question now is how can it be possible for this sort of
obvious disequilibrium to be sustained? For our theory tells us that a
stable equilibrium is one in which relative factor prices are brought into
equality with relative marginal rates of technical substitution between
inputs (in this case water and the composite of other inputs). We will'
argue in Chapter III that certain powerful interests are able to influence
iniStituitional arrangements such that small farmers are kept in a perpetual
state of bare subsistence. But there is another issue at work here and
it pertains to differential factor prices paid by farmer A and farmer B.
We assume that farmer A is not only more favorably situated on the
irrigation,canal, bt that A is a relatively influential farmer compared
to B. Hence not only is B exposed to the water-use whims of farmer A, but


11-17
also to the normal problems faced by subsistence farmers in countries
where*a dualistic agriculture is well entrenched. This dualistic agricul-
ture permits the "commercial" sector to enjoy certain economic and political
advantages over-the subsistence sector.
The disequilibrium in Figure 11-2 is stable because of a difference
in factor prices paid by the two agricultural sectors. In Figure 11-3 we
see the releyant price lines for farmer A and B relating the unit cost of
water with the unit cost of the composite "other inputs." The two relative
prices are given by:
A pB
W       PW                            (7)
P APB
I        I
We can assume that the unit cost of water as between the two farmers
is equal--and quite low. This would be the case in the vast majority of
the developing countries where water is sold in large "blocks" depending
upon the land 'area to be irrigated. With equation (7) and the condition
that
PA   PB                              (8)
W    W
then it follows immediately that
P I P                                                       (9)


II-18
B
B
PS
I
.                                     A
P
A.A
P
A-
--FIGURE II-3. A SUSTAINABLE DISEQUILIBRIUM IN IRRIGATED AGRICULTURE


11-19
In point of fact, we should probably not assume that PA   PB.  While
w ..s.w
the fee paid by A and B to the water agency may indeed be equal, the
situation will normally be one wherein the additional costs to the down-
stream farmer (B) are higher than for the .upstream farmer (A). We refer
to these extra costs as transaction costs. Such costs would include the
extra time and effort necessary to insure delivery to the turnout, the
extra "policing" of the ditch for illegal turnouts, water stealing, and
poor maintenance, plus the likely need to devote more time to ditch main-
tenance. In sum, the downstream farmer usually pays more for an equal
quantity of water when costs are understood to include both cash and non-
cash elements. Moreover, to the extent that there is a discrepancy between
the amount of water paid for and the amount actually received by the down-
stream farmer, the per-unit cost difference between upstream and downstream
farmer is further exacerbated.
The recognition of different 'factors prices between the "commercial"
sector (typified here by farmer A) and the subsistence sector (farmer B)
is rather commonplace among economists. Grabowski cites evidence from
work by Barbara Tuchman in Mexico, Kieth Griffin in Indonesia, Joshi and
Rao in India, and Robert Wade and Gunnar Myrdal [Grabowski, 1979]. A
recent study by Berry and Cline indicates that borrowed capital for
Indian farmers with holdings of less than 2 hectares is 17.3 percent per
annume for those with between 2 and 6 hectares the per annum rate is 13.8
percent, for those between 6 and 10 hectares the perannum rate is 12.2 percent,
and for those with over 10 hectares the rate drops to 11.8 percent per
annum [1979].


II-20
In a rather careful test of Indian data Surjit S. Bhalla (in an
appendix to the Berry-Cline study) finds that the unit rental cost of
land decreases by Rs. 5 - Rs. 10 for each additional acre rented. His
findings on capital markets are cited immediately above. In 1970-71
government, cooperatives and the commercial banks handled 30 percent of
the total lending, compared to 19 percent in 1961-62, and only 7 percent
in 1951-52. The private money-lenders' share declined over the same
period from 75 percent to 50 percent; the remainder of credit is supplied
by friends and relatives, and by landlords. In 1974 the interest rates
charged by commercial banks was 8.4 percent, it was 8.8 percent for
government organizations, and 8.9 percent for cooperatives. In contrast
the rate for money lenders was 22.8 percent. Whereas the organizational
lenders had fairly equal rates for the various farm sizes, money lenders
charged 22.5 percent for those farmers with less than 2 hectares, 20.9
percent for those with between 2 and 6 hectares, 23.3 percent for those
with between 6 and 10 hectares, and only 16.3 percent for those with over
10 hectares. Finally, the imperfections in the labor market were found
to be pervasive and greater than in either capital or land [Berry and
Cline  Appendix A by S. S. Bhalla].
B. SUMMARY
Irrigated agriculture is characterized by physical interdependence
which links farmers via the watercourse--the physical structure which
brings them one of their most valuable inputs also ties them inextricably
to their neighbors. One's control over water availability is known to be
a function of the number of other irrigators located along the watercourse,


11-21
and given differential power and economic influence we find the small
farmer on a system with a mixture of large and small farmers in double
jeopardy. That is, not only must this farmer contend with the usual
problems of,bei-ng on the edge of subsistence, but it is also necessary
to contend with the exigencies of highly uncertain water receipts. On
watercourses where all farmers are of similar political and economic power
those located some distance from the head of the system still find them-
selves vulnerable to the water schedule of those upstream.
This physical interdependence--a technological externality--introduces
serious inefficiencies into irrigated agriculture, not to mention pervasive
equity problems. When there is a mixture of powerful and subservient
farmers on a watercourse the inefficiencies can be stable because of
different factors prices paid by the two classes of farmers. When this
socioeconomic mixture.is not present it is still possible to have
inefficiencies because of uncertain water receipts, but it would not be
classified as a "stable disequilibrium" by the economist. This of course
is not to say that it would not persist.
One implication for project planning relates to the issue of homo-
geneity among farmers on a water course. In areas where there is a degree
of socioec6nomic difference among farmers to be served by a new watercourse
we simply compound those differences over the long run if the engineering
works and the institutional arrangements over water allocation are not de-
signed very carefully. Through careful engineering it might be possible
to group rather similar classes of farmers on a watercourse. The danger
here, however, is that one merely shifts the locus of conflict from the


11-22
watercourse to the main canal serving several watercourses. However, if
the irrigation bureaucracy is more powerful at the level of the main canal
than at the watercourse--and this seems to be the general pattern--then
we may have improved the situation.
Where the problem is one of improving existing irrigation systems we
have less flexibility in rearranging physical facilities to meet socio-
economic realities. In these cases the only solution is to make up for
heterogeneity along a water course with a more forceful institutional
arrangement. More will be said on this below.


III-1
CHAPTER III. INSTITUTIONAL UNCERTAINTY AND THE SMALL FARMER
An institution is a socially sanctioned set of ordered relationships
among people which defines an individual's rights, duties, obligations,
and exposures vis-a-vis others. A rental contract is an institution, so
is a job contract, an agreement among apartment dwellers over stereo noise
after 10 p.m., and speed limits. All of these "rules" define rights,
duties, obligations, and exposure. If I exceed the speed limit I am ex-
posed to the force of law, if I play loud music at midnight I am exposed
to the sanctions of my community of apartment dwellers. Likewise, I have
the right to be free of stereo noise after 10 p.m.
But if institutions define rights, duties, obligations, and exposure
then it follows that prices are also institutions.- If I hire a laborer
and the agreed-upon price is $3.45 per hour then this defines our rights
and duties in this particular transaction. If a farmer buys fertilizer
at a given price then that price becomes an institution for the transaction
under consideration. For something to be an institution it is not
necessary that it be codified, or even written down. What is vital is
that a society be able to operate with the presumption and-expectation
that codified rules will be enforced, and that non-codified rules will be
honor-ed. This simple fact introduces predictability into social inter-
action and produces a measure of order out of chaos.
Predtctability of rights, duties, obligations, and exposure are
necessary conditions for a dynamic society. Uncertainty over institutional
arrangements is a primary cause of rural stagnation and the persistence of
a subsistence sector in agriculture. Those economists inclined to


111-2
celebrate private property rights have this in mind when they advocate
ubiquitous private property so that atomistic agents might optimize by
having knowledge that they will be able to "reap where they have sown."
The fallacy in their logic, of course, is to confuse predictability with
excludability. Farmers in the Swiss Alps do not need to exclude all of
their neighbors from the summer common pasture to have predictability over
the availability of grass; there is no need to carve up the commons into
individual fenced plots when a social arrangement exists which introduces
predictability into human interaction of an economic nature.
The institutional economist interested in development would be
inclined to look at technology and institutions as equally important
factors in economic change. The conventional economist would be more
inclined to view technology as the engine of economic change, and'institu-
tions as mere constraints on the adoption of new technology.
Irrigated agriculture is the epitome of this preoccupation with
technological solutions to economic development. The presumption has
always been that the construction of dams, canals, and control structures
was sufficient to create a bounteous agriculture. In point of fact, the
weight of evidence runs to the contrary, with the vast majority of irriga-
tion systems around the world plagued by excessive water loss, maldis-
tribution of water, poorly maintained drainage facilities, and water theft.
'The concept of rural. stagnation will here be used to describe a
situation in which the bulk of farmers in the developing countries adhere
to traditional agricultural practices. The conventional wisdom in
economic development--indeed the raison d'etre of the scientific revolution
in agriculture--is that farmers need to have access to the improved
technologies and to foreswear their traditional practices. The basis of
this approach to economic development is the quality of the productive


111-3
factors at the disposal of the cultivator--the seeds, the fert-flizer, the
land, the water control, the pest control, and so on. The argument goes:
give'the farmer better factors of production and agriculture will produce
a surplus of both food and income; the former to feed the rural and urban
masses--if not -to export--and the latter to feed the insatiable demands
for investment funds for thc rest of the economy. Agriculture is the
engine of economic development broadly conceived, and that engine runs
best on high-quality "fuel" (the factors of production). We refer to
this as the factor-quality hypothesis.
There is a second hypothesis which places greater emphasis on the
nature of the economic environment in which the subsistence farmer operates.
This hypothesis would suggest that rural stagnation arises for two
related reasons--the inability of subsistence farmers to: (1) accumulate
an economic surplus; and (2) plan on accumulating a surplus. It is the
presence of this surplus which cushions the perils of change, and which
holds the reward for abandoning traditional practices. At first thought
it appears that this is nothing more than a problem in "agricultural
policyn broadly defined. That is, there is a need to adjust relative
factor and product prices so that subsistence farmers might accumulate a
surplus which could then be used for economic development. However, the
issue runs'deeper than that, and is not so readily solved. We refer to
this as the institutional-uncertainty hypothesis.
This difference of opinion about the conditions which will lead to
agricultural development are quite central to any discussion of irrigation
and the small farmer. If the factor-quality hypothesis is correct then
the way to deal with small-scale irrigation is simply to build better
C


111-4
water transportation and control facilities and leave the farmins to
their own devices. On the other hand if institutional aspects dominate,
then'better irrigation facilities are insufficient to enhance small-scal.e-
irrigated agriculture. In the following, each of the two hypotheses is
discussed. Then, some conclusions are drawn which will form the basis
for several subsequent chapters.
A. THE FACTOR-QUALITY HYPOTHESIS
Farming. babed. wholly upon the kinds of factors of production
that have been used by farmers for generations can be called
traditional agriculture. A country dependent upon traditional
agriculture is inevitably poor, and because it is poor it spends
much of its income for food [Schultz, p. 4].
Theodore Schultz is perhaps the most articulate spokesman of the
school which blames rural stagnation on low-quality factors. The
Schultzian peasant is small, poor, tradition-bound, but efficient. That
is, given the resources at the-disposal of the farmer, given the prevailing
institutional environment, and given the objectives of the farm enter-
prise it is impossible to reallocate factors of'production and improve
upon the prevailing situation. Schultz states:
The presumption is that when farmers are limited to traditional
factors of production they reach a point after which they can
make little or no contribution to economic growth because
there are few significant inefficiencies in the allocation
of factors, the removal of which would increase current
production, and because investment made to increase the stock
of traditional factors would be a costly source of economic
'growth [1964, p. 24].
Hence the Schultzian farmer is situated on a production possibility
frontier. From this Schultz draws the conclusion that subsistence agriculture
cannot be an engine for economic growth since there are "few significant
inefficiencies in the allocation of factors." As we know, Schultz uses
this model to advocate the investment in human capital and in modern
(non-traditional) agricultural inputs. He would say that rural stagnation
C


III-5
arises because no one is investing in new "income streams." To transform
traditional agriculture it is necessary to invest in the modern inputs
which create these new income streams.
Indeed the legacy of agricultural assistance is consistent with this
notion. We have seen decades of investment in new seeds, fertilizer
plants, pest control, farmer training, and the like. We cannot say how
great the transformation has been, because we do not have an experiment
in which we can hold some other things constant. We of course know that
some farmers in some countries have indeed made impressive strides in
terms of increased production and increased incomes. We also know that
there are still millions of subsistence farmers barely able to make a living.
C,


111-6
B. THE INSTITUTIONAL-UNCERTAINTY HYPOTHESIS
The Schultzian world is one of wise and efficient farmers doing the
best they can with a poor resource endowment, and with inadequate purchased
factors. It is a world of rural stagnation brought about by efficiency
and the lack of good investment opportunities. But it is a curious world,
for much of the blame for stagnation rests with the fact that farmers are
already efficient. While it is true that there are instances of develop-
ment arising out of the elimination of inefficiencies, it seems safer to
conclude that economic development is more commonly induced by the
accumulation of a surplus by entrepreneurs--such surplus then forms the
wherewithall for new investment, and a cushion for risk taking. Few
economists place the elimination of inefficiency at the center of their
model of economic growth and development--indeed some would probably argue
that the issue of efficiency or inefficiency is irrelevant to growth.
There are surely instances of very efficient economies continuing to
enjoy rather rapid growth rates.
What then can be the cause of the absence of change in subsistence
agriculture? The symptom is the absence of change, and the.proximate
cause -is the lack of an economic surplus with which to facilitate that
change. Bvt the logically prior cause is an institutional structure
which does not permit the accumulation of an economic surplus within the
subsistence sector. Not only does a surplus ease the psychic and
economic costs of change, but it provides the means whereby new equipment
and seeds might be purchased, not to mention fertilizer, pest management
aids, and,the like. And, in a macroeconomic sense, the existence of a
C


111-7
reliable economic surplus in rural areas will show up in increased
aggregate savings which can then be utilized for either rural or non-rural
investments. Indeed, one of the persistent problems in the developing
countries is the absence of significant savings from 60-75 percent of
the population engaged in agriculture.
Institutional uncertainty exists because of the pervasiveness of
what Myrdal calls the "soft state." To quote from Myrdal:
When we characterize the countries of South Asia as "soft
states" we mean that, throughout the region, national govern-
ments require extraordinarily little of their citizens. Even
those obligations that do exist are inadequately enforced.
This low level of social discipline is one of the most funda-
mental differences between the South Asian countries today
and the Western countries at the beginning of their
development [p. 182].
The "soft state" is not confined to South Asia, and its reference
in the above quote arises simply because Myrdal was writing about poverty
in th-at part of the world. But the existence of a "soft state" is
virtually universal in the developing countries--even those nations we
would normally think of as military dictatorships. For there is an
important difference between the nature of the government (dictatorship,
democracy), and the social discipline in daily economic life. A "soft,
state" is here defined as one in which: (1) formal institutional arrangements
are merely "suggestive;" and (2) these rules are changed at will by a
subset of the polity without much regard for those who will lose thereby.
The second element of the model of rural stagnation flows immediately
from the preceding, and pertains to the exercise of power by a small
number of individuals in the agricultural sector or the economy at large.


III-8
This power-elite has been identified in some instances as "landlords,"
and-in others as merely "urban interests." But the truth would carry us
beyond these generic terms to a group of individuals which controls
agricultural land, fertilizers, labor, seeds, water, pesticides, and a
variety of important elements on the marketing side of agriculture.
primarily agricultural price and credit policies.
To bring these elements together then we have a hypothesis which
holds that rural stagnation is caused by the inability of traditional
agriculture to generate a sustainable economic surplus. This ephemeral
surplus is the result of a power-elite manipulating institutional
arrangements in order that the economic environment of subsistence farmers
be just sufficient to keep them in production, yet not sufficiently
propitious to encourage experimentation.
With subsistence agriculture there is enough "slack" that extreme
fine-tuning is not required. That is, the economic opportunities for
subsistence farmers are limited--their opportunity wage is probably lower
than even the   meager return on their agricultural labor. Given this,
those who seek to hold economic rents to a minimum in the subsistence
agricultural sector need not worry excessively about "over doing it."
After all,.it is. recognized that reduced rents will show up as lower food
c6nsumption by the subsistence sector. There is some degree of labor
O           immobility in subsistence agriculture and workers do not respond to
reduced rents by instantaneous adjustments out of the sector as our
conventional theory would predict.


111-9
The presence of a "soft state" and a power-elite will virtually
ensure that the subsistence sector will bear the brunt of the incessant
manipulations by the fortunate to further enhance their economic position.
The issue of risk aversion among small farmers is usually discussed in
terms of the inability of the farmer to survive one bad year with the
more fickle modern varieties. But there is another aspect which rarely
receives the attention it deserves. This involves such things as
uncertainty over the price and availability of requisite fertilizers and
insecticides, and the uncertainty of product price at harvest time. The
advantage of subsistence agriculture is that the small farmer is-minimally
dependent upon the outside world; a world he suspects of wanting to keep
him subservient. By turning to cash crops or production for exports he
increases his dependence upon -input markets as well as product markets.
It is pervasive institutional uncertainty which causes rural
stagnation; an inability to know what the rules of the game will be, and
thus an inability to become a dynamic economic agent. The farmer is sure
that if the crop promises to be a good one prices will be low. If farm
income goes up too much the landlord will notice and modify the rental
arrangement, and agricultural policy will surely change next year under
pressure from urban consumers. The farmer is confident that any surplus
which appears in the subsistence sector will quickly be transferred to the
urban sector, to the agricultural inputs sector, or to the product market
sector.
To appreciate institutuional uncertainty it is necessary to draw a
distinction between risk and uncertainty. The farmer knows that every
4-5 years the monsoon comes too late to be of much benefit. The farmer
C


11II-10
knows that every so often the leaf hoppers arrive in such numbers that
much-of the crop is lost. The farmer knows that rust can be expected to
show up ,every 6-7 years. These are risks, and they are conceptually
equivale.' to the calculations we make whenever we cross a street, sit
down in an airplane, or drive to the beach. But uncertainty is, by
definition, the absence of any empirical basis for anticipating random
events. Much of the economics literature has blurred the distinction
between risk and uncertainty, since uncertainty does not lend itself
well to the probabilistic optimizing formulations currently fashionable.
But institutional uncertainty is just that--the absence of any
empirical basis for knowing what the price of wheat will be 7 months hence,
for knowing whether or not there will be any irrigation water when the
crop is under stress, for knowing whether or not there will be fertilizer
available when the optimal date for application arrives. These are risks,
to be sure, but they are not risk--they are uncertainty. How does the
subsistence farmer deal with institutional uncertainty? In a sense we
find a transformation taking place. That is, institutional uncertainty
is transformed into "certainty." The farmer simply assumes that water
will not be available, that fertilizer will not be available (or not
affordable), thqt a bounteous harvest will bring low prices, and so on.
Given this "certainty" there is adherence to traditional methods, there
is a reluctance to experiment, and--most importantly for rural stagnation--
there is no economic surplus.
C. TWO COMPETING HYPOTHESES
The causal chain of the two competing hypotheses can be depicted
as follows:


III-11
The
Factor-
Quality,
Hypothesis
low factor quality                                            stagnation
efficient 5c         -       lack of profitable investments->and
)pove,tty
The
Institutional-
Uncertainty
Hypothesis                                                         stagnation
institutional uncertainty----yno eoanomic  pilus -------------. and
poverty
We can differentiate between the two hypotheses in a rather
simple and familiar fashion. The factor-quality hypothesis maintains that
the subsistence farmer possesses the ability to invest in non-traditional
technology but not the willingness to invest--such lack of resolve arising
from the absence of good investment opportunities. On the other hand,
the institutional-uncertainty hypothesis maintains that the subsistence.
farmer possesses the willingness to invest in non-traditional technology,
bu.t nQt the ability to invest--such lack of ability a.ising from the
absence of an investable surplus. The factor-quality hypothesis focuses
our attention on the individual farmer and the agricultural practices. That
is, how we induce the farmer to use certified seed, more and better fertilizer,
modern pest-control means, and other higher-quality inputs, Indeed, it is
remini'scent of the contests in the U.S. among dairy, wheat and corn farmers to


III-12
see who could produce the highest yields. By way of contrast, the
institutional-uncertainty hypothesis insists that factor quality in a
physical/biological sense is subordinate to the question of the stability
and predictability of institutional arrangements.
These preliminary thoughts on the competing hypotheses do not permit
a thoroughgoing investigation of how one might test these two alternative
explanations for rural stagnation. However, it is possible to offer a
few possibilities, tentative though they may be. The place to start would
seem to be on the matters of willingness versus ability to invest. Do we
find high rates of saving in the subsistence sector? Does the subsistence
sector have access to capital at rates comparable to those paid by the
commercial agricultural sector? When a surplus appears in the subsistence
sector is it used on farm investments? Are rental contracts in the
subsistence sector rather invariant with respect to lagged income? Is
there'a stable class of "innovators" in the subsistence sector who continue
to make incremental transformation in their agricultural practices?
If we find that these questions can be answered in the affirmative
then it would lend credence to the factor-quality hypothesis. On the
other hand, if saving in the subsistence sector is minimal, if farmers in
the subsistence sector must pay.higher rates for capital; if periodic
surpluses are allocated to consumption rather than to investment, if rental
,'         terms are adjusted after a particularly good year, and if there is no
continuous innovation by the bulk of the subsistence farmers then the
view that these farmers have the ability to invest but not the willingness
would be open to doubt. For the institutional-uncertainty hypothesis
would predict that the existence of a soft state keeps the ability of the
subsistence farmer to invest at a nominal level, in spite of any willing-
ness which might persist.


111-13
D. IMPLICATIONS FOR SMALL FARMERS
Small farmers in the developing countries find themselves in an econ-
omic environment which is largely inimical to the types of practices which
development planning traditionally advocates. The essence of "modernized"
agriculture is to become more involved in the various markets--for inputs,
outputs, information, and the like. By way of contrast, the essence of
subsistence agriculture is to operate with a minimal involvement in such
transactions. In a world of institutional uncertainty--and pervasive
powerful interests--the subsistence farmer has little incentive to break
out of the rather secure (but impoverished) existence of the past. By
"secure" I do not mean to imply that the farmer is happy and comfortable.
I 'mean, inste0d, that at least the subsistence farmer is not overly dependent
upon a host of external markets and agents over which there can be little
control. True, the farmer may be at the mercy of a landlord, but there is
little interest in also becoming dep?ient  upon the fertilizer dealer, the
tractor repairman, the chemical supplier, and so on. This disinterest stems
not from a rejection of the benefits which "modernization" might bring, but
rather from the costs that come from modernization. Those costs are the
increased dependence on a world which the subsistence farmer does not trust,
and ibich the farmer may only imperfectly understand.
When irrigation is introduced into the picture it becomes even more
complex. The interdependence discussed in Chapter II has already.formed
the small farmer into an economic environment where individual control is
reduced. The essence of developing irrigated agriculture in a way that it
helps--rather than hurts--the small farmer is to ensure that the institutional


111-14
arrangements governing water allocation and system maintenance do not
exacerbate the already unfavorable economic environment of the small farmer.
When new technology in the form of ditches and control structures enters
an area where irrigation has not been practiced it creates the opportunity
for some farmers to reap new income streams. When the institutional arrange-
merits are not well established at the time the new technology creates these
new income streams then a basis for conflict and division has been established.
That is, when' irrigation comes to an area some farmers will be able to
expropriate for themselves some of the income streams made possible by this
innovation. If this is allowed to continue for several crop seasons then
those fortunate few who were able to appropriate the new income will come
to think .pon their good fortune as "legitimate." Then it will become
difficult--if not impossible--to rectify the situation.
It is for this simple reason that the traditional approach to irrigated
agriculture has been s-uch a failure. It is not sufficient to construct
the engineering, works and "get the water flowing" and then to turn tothe
institutional arrangements governing water allocation and maintenance of
the system. The general conditions of system operation (allocation and
maintenance) must be defined before water moves through the'system. These
conditions need not be overly specific, but they must represent a "consti-
tution" in the following sense. Before any ditch is built it is important
to have general agreement about how water-ought to be allocated prior to
any one farmer knowing precisely where the watercourse will be located
vis-a-vis his/her farm. That is, general rules must be articulated behind
the Rawlsian "veil of ignorance." l  For in the absence of specific knowledge
1 This approach is spelled out in greater detail in Bromley, Taylor and Parker [1980].


III-15
about who will be "head end" farmers and who will be "tail end"' farmers
it is easy to imagine that all farmers will agree on a general principle
of equity in water receipts and in system maintenance.
Having such agreement then provides a basis for enforcement once the
system is in operation. The argument always advanced in opposition to this
formulation is that farmers will not even sit down to discuss irrigation
principles until they see the water flowing. While it may indeed be difficult
to exact an understanding that each farmer ought to have, say, 4 hours of
water every 5 days, it is not difficult to get general agreement on the sort
of "constitution" implied here.
For existing irrigation systems in which there is an interest in re-
habilitation the problem of institutional uncertainty is not so readily
solved. Here, where there are existing patterns of water allocation and
system maintenance it will be more difficult to bring about change. As
mentioned previously,.those well served by the existing arrangement will
consider it legitimate, and will fight any efforts toward reform. To the
extent ,that they are the more powerful members of the local community their
wishes well carry some weight. However difficult it may be, the interests
of small farmers are badly served by a world in which they-are at the mercy
of a number of economic agents. The goal of improving small-farmer irrigation
must recognize that the central issue is one of institutional uncertainty.
Once understood, the policy instruments will be more easily identified.
Institutional uncertainty is manifest at the farm level in several
important ways. Our interest here is in terms of how the small farmer in
general--and the small irrigation farmer in particular--adapts to this
uncertain environment. To that we now turn.


IV-1
CHAPTER IV. FARMING AS ADAPTIVE BEHAVIOR
In ,Chapter II we were concerned with the physical linking which
inevitably follows the joint use of a drainage ditch or a watercourse.
This interdependence, when accompanied by persistent unreliability of
water receipts, was seen to introduce productive inefficiencies into the
agricultural .sector. This inefficiency is sustained by two sets of
relative prices in agriculture--one for the commercial sector, and another
for the subsistence sector. In Chapter III our concern lay with institu-
tional uncertainty, and with the ways in which that uncertainty influenced
the farmer's choice process. We now turn to a more definitive treatment
of uncertainty within the context of adaptive behavior theory.1
The central element in the model of farmer interdependence in
Chapter II was the notion of farmer B;'s expectations about how much
water woul4.-be available after A had taken a portion, and the timeliness
of the deliveries of that residual. This represents learned behavior on
the part of B, for the situation has probably existed for as long as the
two have been linked via the watercourse; R has learned through experience
that A can be expected to take so much water and that there will thus be
a certain quantity available once A is satisfied.
This body of theory owes much of its development to Richard H. Day.
A recent book represents an application to economic development: Economic
Development as an Adaptive Process [Day and Singh, 1977].


IV-2
The essence of learned behavior is feedback--the receipt of informa-
tion or results on the basis of certain behavior in the past. We
assume that in the early days of the irrigation project B might have had
a presumption that all farmers would enjoy equal "access" to irrigation
water--where access comprehends both quantity and timing. However,
experience taught B that this was not to be. Early planning on the part
of B was found to be deficient because part of the economic environment
was beyond his/her control. Along with the weather, disease, and market
prices this new input (water) brought with it another element which
introduced yet additional randomness into daily economic planning. Not
only must the farmer respond to--and anticipate--the normal vagaries of
farming, but now the added vicissitudes introduced by one or several
upstream irrigators must be reckoned with.
It requires no great wisdom to see that farming is--above all else--
adaptive behavior. The central questions for dvelopment planning are,
therefore, adaptation to what? what is the nature of that adaptation?
how does adaptive behavior alter the chice set? how does adaptive behavior
alter the goal set? and how can development activities incorporate these
implications into the design of economic policy?
We have already talked about the forces to which the small farmer
must adapt--weather, crop diseases, uncertain supply of fertilizer, unknown
prices for the product, the upstream farmers who have prior access to
scarce irrigation water. We can depict this feedback process by imagining
two discrete time periods for the farmer: (1) the upcoming season, denoted
by t=1; and the sum of recent experiences into which the array of


IV-3
exogenous forces have been incorporated: t=(-l) + (-2) + ...+ (-n).
This is depicted in Figure IV-1.
The learned behavior of farmers can be thought of--in totality--as
constituting a-set of feasible management choices. The farmer knows,
within certain bounds, what will "work" and will not. Hence, the
adaptation we have in mind here should not be thought of as a series of
major adjustments. It is more correct to imagine them to be small
changes from the normal pattern in light of experience, hunches, prognoses,
extension-service advice, what the neighbors are thinking of doing, and
even astrological indicators. With this input the farmer is able to define
what we would call a feasible set or a feasible range of enterprise and
activity choices. Day and Singh draw attention to the important distinc-
tion between those choices which are perceived by the farmer to be
feasible, and those which are, in actuality, feasible. That it is important
for the set of perceived feasible choices to lie completely inside of the
set of actua.T  . feasible choices ban-be illustrated with Figure IV-2.
In panel (a) of Figure IV-2 the actually feasible set contains the totality
of the perceived feasible set. In panel (b) however we see that the
actually feasible set does not contain all of the set of choices perceived
by the farimer. 'This means that the farmer is liable to make a choice
with respect to one of the decision variables which is not possible. An
example might be the choice for planting 7.4 hectares but upon completion
of land preparation then discover that is is only'possible to buy enough seed for
3.6 hectares.. To go back to a previous example, the farmer may make an input pur-
chase (say, fertilizer) on the assumption that water will be available


tROGENOUS  FACTORS
prices, weather, diseases, pests
FARM                A+ -FARM  -FARM       FARM    .      FARM
t= -nt              t   -3      t= -2      t= -1           t1
FARMER  INTERDEPENDENCE
water availability, water timing, salinity
FIGURE IV-1.   FEEDBACK INFLUENCES ON FARM DECISIONS IN t     1.


(a)                                              (b)
ACTUALLY FEASIBLE SET                                    CTUALLY FEASIBLE SET
PERCEIVED FEASIBLE SET
cj                    PERCEIVED FEASIBLE SET
co
5-4
cn                                     tn
DECISION VARIABLE I                                   DECISION VARIABLE 1
FIGURE IV-2.   ACTUAL AND PERCEIVED FEASIBLE SETS.


IV-6
at the critical time, only to apply the fertilizer and then discover that
water is unavailable. The farmer avoids serious mistakes
by avoiding      situations in which the actually feasible set does not
contain the totality of the perceived feasible set.
Day and Singh argue that when farmers are faced with extreme
uncertainty they will respond by behaving cautiously; this is the risk-
aversion one 'sees in the literature. One aspect of cautious behavior is
to restrict one's choice set to ensure that the perceived feasible set is
totally contained in the actually feasible set. This would resemble
panel (a) of Figure IV-2. Having altered the choice set in such a manner,
the next question pertains to the nature of the goals of the farmer.
Much has been written on the nature of optimizing behavior, and we
will spend little time on that here. The difference between optimizing
and satisficing behavior, however, can be illustrated with the help of
Figure IV-3. Optimizing behavior would focus the decision process on
finding,the optimum point and adjusting the decision variables under one's
control so as to achieve this point; this is noted in the Figure. Satis-
ficing is being content with a larger target. In the Figure it is more
desirable to move to the northeast in terms of accomplishing the goals of
the decision maker; the farther to the northeast the better. However,
satisficing behavior would indicate that some subset of the feasible
region intersects the target region and any outcome in this intersection
is "good enough." This region is denoted in Figure IV-3 by the shading.
There is no great preoccupation with finding the best solution--there are
several which will do.


IV-7
TARGET
FEASIBLE
REGION
OPTIMUM
DECISION VARIABLE 1
FIGURE  IV-3. SATISFICING BEHAVIOR.
(i


IV_8
Day and Singh also introduce the concept of cautious optimizing.
H-ere- the farmer defines a region of safe-enough decisions, or what they
call a zone of flexible response (ZFR). Here rather than a target of
moving in the general direction of the northeast, the farmer wishes to
define a region in which serious mistakes might be avoided. In panel
(a) of Figure IV-4 some safe-enough solutions are also feasible. In
panel (b this is not the case; here the decision maker chooses that
alternative which is closest to the set of safe-enough solutions (point M).
Referring back to panel (a), once it is discovered that a safe-enough
decision is also in the feasible set, it is possible to make the-ultimate
decision on the basis of some other criterion--say profit maximization.
In this example the safe-enough decision might be to assure a certain
level of food consumption in the farm household. This constraint defines
a region as in panel (a). Then, within that set, the farmer may adjust
variables 1 and 2 so as to maximize profit; this is constrained optimiza-
tion where the constraint is some minimum diet for the family.   It is
also lexicographic decision making--once one dominant objective has been
attained (food sufficiency), another can become relevant (profit
maximi7ation).
To suwmarize so far, adaptive economic behavior is said to be
cHaracterized by feedback from learned behavior and experiences to assist
ci       in the selection of decision variables for the next planning horizon.   The
decision goals are said to be cautious optimizing,,or the selection of
safe-enough outcomes. This model of choice can be related to the preceding
chapter. To do this we must remember that feasible choice sets--or
opportunity sets--differ across farmers. Some will have a rather large
range of opportunities for decision making, while others will have a more


(a)                                                      (b)
ZONE OF FLEXIBLE                                          ZONE OF FLEXIBLE
RESPONSE                                                   RESPONSE
FEASIBLE               0CAUTIOUS SOLUTINS                 FEASIBLE        M-CAUTIOUS SOLUTION
REGION                                                    REGION
DECISION VARIABLE 1                                 -      DECISION VARIABLE 1
FIGURE  IV-4. CAUTIOUS OPTIMIZING.


IV-10
restricted set. An example can be found in the capital market. A
large farmer with secure land title has a greater choice of lending
agencies than does the poor tenant farmer; this is an example of differen-
tial opportunity sets as between the two. Likewise the large farmer may
have sufficient production that buyers will come to him/her and may actually
bid for the crop. In contrast the small farmer may be lucky to find one
buyer for the. small amount left overaftcr the family's subsistence needs
have been met. Finally, with respect to irrigation, the farmer at the
end of the water course will have fewer options than the farmer located
at the head of the system. Again the one at the end of the ditch has a
more restricted opportunity (choice) set than does the farmer at the head
of the canal. We can use the previous examples of feasible regions to
depict this phenomenon. But first it may help to graphically relate this
matter of differential vulnerability of farmers depending upon whether or
not they are in the commercial sector or in the subsistence sector. This
is shown in Figure IV-5. Here the arrows depict avenues of influence on
other agents in the economic system. This is a considerable simplification,
and yet the implication is important; the subsistence farmer (B) is seen
C           to be on the receiving end of influence by the farmer in the commercial
sector (A), by forces in the general economy, and by forces in the
agricultural economy. While the large farmer (A) is not immune from these
influences, it is also well known that such a farmer is not without some
ability to influence events in these other economic spheres. Moreover
if we introduce the interdependence which accompanies irrigated production--
and if we assume that 8 is at the end of a watercourse--then farmer A has
another avenue of influence not open to B.


FARM
AGRICULTURAL                       A                           NON-
SECTOR                                                  AGRICULTURAL
SECTOR
FARM    S.
FIGURE IV-5. LINES OF INFLUENCE OVER OPPORTUNITY SETS.


IV-12
The difference in opportunity sets of the commercial farmer and the
subsistei.ce farmer can be illustrated with Figure IV-6. Notice that there
is more ,of interest here than the mere difference in size of their
respective opportunity sets. Most importantly we see that farmer B has a
smaller zone of flexible response, as well as a feasible set which is
subject to change during the course of the crop season (depicted as two
sets). Farmer A can choose any combination within the feasible region,
or if wishing to be cautious can be confined to the set of cautious
solutions.
On the other hand B might start the season under the impression that
opportunity set I will exist throughout, only to find later on that now
opportunity set II exists. Instead of several cautious optimizing
solutions there is, in fact, only one.
Adaptive behavior then, can be combined with notions from Chapters
II and III about farmer interdependence and disadvantaged (powerless)
small farmers to illustrate the essence of subsistence farmers in the
developing countries. The learned response is one of cautious optimizing
in response to constantly changing opportunity sets. The range of choice
for the small farmer is extremely limited by the economic environment, as
well.as by-the influence exerted by more powerful neighbors. As seen
previously, not only are there important equity implications from this
situation, but serious productive inefficiencies are also present.
Economic development policy directed to the small irigator must address
both of these issues.


FARM A                                              FARM B
caJ
ca OPPORTUNITY SET I
OPPORTUNITY SET II
OPPORTUNITY
SET
DECISION VARIABLE 1                                       DECISION VARIABLE 1
FIGURE IV-6.  DIFFERE4TIAL OPPORTUNITY SETS.


IV-14
Irrigation development policy which recognizes institutional uncert-
ainty and cautious optimizing on the part of farmers would need to focus
on programs and projects specifically aimed at "tightening up" the economic
environment of the small farmer--whether that farmer is situated on a large
irrigation project or on a small one. More will be said on this matter in
subsequent chapters.


V-1
CHAPTER V. THE ESSENCE OF DEVELOPMENT
In,previous chapters we have talked of rural stagnation, and of the
adaptive behavior of farmers in the developing countries. This was given
specific content by talking about the problems of small-scale irrigators.
The solutions to irrigation problems in the developing countries cannot
be talked about in isolation from the normal concerns with economic
development. Hence, in this Chapter some attention will be paid to the
general concept olf economic development. There will be two main sections
in the chapter, the first will concern the question "what is development?"
while the latter will focus specific attention on the developmental
imperatives inherent in small scale irrigation.
A. WHAT IS DEVELOPMENT?
The concept of development is as varied as the number of authors who
have written on the subject. There is, additionally, some general confusion
in the.literature over the distinction between Rrowth and development.
It is not useful to become involved in a lengthy discussion of these
terms. Instead development will be defined, and will then..be related to
the foregoing discussion of subsistence agriculture and rural stagnation.
In tIe present context, economic development will be taken to mean
a succession of changes within the subsistence agricultural sector which
alters the.basic structural and technological aspects of economic life.
It must be clear that development is thus the result of some logically
prior changes--changes which permit structural and technological change.
This prior condition is the retention within the subsistence sector of an
C


V-2
economic surplus with which to finance the technological change. It
should also be clear that technological change fed by the existence of
an economic surplus cannot be expected to operate independently of the
economic infrastructure in the subsistence sector. What is this structure?
It is factor and product markets. It is, therefore, the means whereby
subsistence farmers have access to such inputs as credit, machinery, seeds,
fertilizer, extension advice, and markets for their products.
It should surprise few that much of our "development" assistance
has focused on these structural aspects; agricultural technology has also
received its share of attention. But usually the concern with structure
and technology has occured in the absence of a corresponding attention
devoted to the other essential ingredient--an economic surplus with which
to avail one's self of'the new technology, the factor opportunities,
and the more organized product markets.
Development must begin with a recognition of the hierarchical nature
of farmer decision making in the subsistence sector. Drawing upon the
previous chapter, we recall that there is a hierarchy of goals, here to
be referred to as:
1. assure survival---the subsistence goal
2. cautious optimizing---the safety goal
3. acquire cash for consumption and savings---the surplus goal
4. profit maximization---the speculative goal
These goals are considered lexicographic; the safety goal is not
considered until the subsistence goal has been attained; the surplus goal
C           is not considered until the safety goal is attained; and the speculative


V-3
goal is not considered until the surplus goal is attained. In view of
the problems in the subsistence sector, we will spend little time discussing
the speculative goal.
The nature of the lexicographic problem can be illustrated by
reference to a diagram from Day and Singh [1977]. In Figure V-1 we have
farmer i's working capital plotted against cash consumption. In stage I
of the Figure all of the working capital of the enterprise is required
merely to meet the subsistence needs of the farm family, plus the safety
goal. In stage II the farm has generated sufficient income in this period
to allow some consumption beyond subsistence and safety needs. In stage
III we finally reach a situation in which the farm family has sufficient
working capital to allocate some to non-current-consumption items (invest-
ments). It is this investable-surplus which was said to be missing in
instances of rural stagnation. Here, with an investable surplus a variety
of technological options can be pursued. The advent of these choices
will then begin to ripple through the rural economy altering its structural
characteristics.
Figure V-2 is also an adaptation from Day and Singh, and its purpose
C is to present another illustration of the lexicographic choice process of
the subsistence .farmer. In panel (a) we see the traditional crop plotted
along the horizontal axis, and the modern alternative along the vertical
axis. Implicit in the output of each is a "technological package" that
includes the nature and level of fertilizer, seed, pesticides, machinery;
clearly the implicit package for the traditional crop differs from the
implicit package for the modern crop. In this diagram we will consider


V-4
I            II             III           ;
SURPLUS FOR
INVESTMENT
WORKING CAPITAL
.FIGURE V-i. THE DEMANDS UPON WORKING CAPITAL.
(Adapted from Day and Singh, 1977).
C..


V-5
(a)(b)
LAND CONSTRAINT
C)                                                                   SUBSISTENCE
CAPITAL CONSTRAINT
TRADITIONAL CROP                              TRADITIONAL CROP
(c)                                             (d)
C'
0                                                              C
o 0
. :  SAFETY ZONE
TRADITIONAL CROP                                TRADITIONAL   CROP
(e)
N-     N
Lu
TRADITIONAL CROP
FIGURE V-2. LEXICOGRAPHIC DECISION MAKING.


V-6
only two constraints--land and capital; in the subsequent section we will
intrpduce irrigation water. The constraint lines delimit the zone of
feasible outputs of both in a variety of combinations.
Recall that the first-order goal of the farmer is subsistence, which
can be met through consumption or sale of either crop, though it is more
likely that the traditional crop will be consumed while the modern crop
would be sold on the cash market. Whichever option is chosen it is
possible to plot a "subsistence opportunity-income line" as in panel (b).
This line depicts the production of the two crops required in order to
cover subsistence needs (either through direct consumption in which case
there is an opportunity cost computed, or through sale in which case
income is generated which can then be allocated to the purchase of food
stuffs and necessary material needs).
Once subsistence is assured, safety becomes the relevant goal. We
can depict safety as we did in Figure IV-4 as a zone of flexible response
which can either intersect or be removed from the zone of feasible
combinations. In this example we will assume that there is correspondence
between the zone of feasible combinations and the zone of flexible
response. This is shown in panel (c) of Figure V-2.
With safety and subsistence adequately considered, it is now possible
ta turn attention to a surplus for cash consumption and savings. This
can be'depicted as a shift of the capital constraint to C' in panel (d).
Finally, in panel (e) it is possible to depict the profit-maximizing
choice where relative prices for the traditional and modern crop
become the dominant decision variable. Notice that until this time
relative prices were of no significance--
C


V-7
except implicitly in the calculation of the "subsistence-opportunity-
incoie line." In panel (e) only the relevant aspects are shown to high-
light the profit-maximizing combination of outputs, previously circumscribed
by the subsistence goal, the safety goal, and the surplus goal. Now, the
total revenue line reflects the relative prices of the traditional and the
modern crops,- and it is that parameter which will determine the ultimate
combination from the safety zone. Indeed, given the discrete nature of
the relevant zone of flexible response the final output combination would
be quite insensitive to several price ratios between the traditional
and the modern crops. This is perhaps more realistic than our classical
textbook case with all-around smoothness; the world is, after all, rather
lumpy, and discrete. Notice that it is only the safety goal which prevents
the farmer from specializing exclusively in the modern crop. With the
relative prices being what they are, a corner solution at M would be more
profitable than the indicated solution at N.
As. long as the subsistence farmer is constrained from the profit-
maximizing combination of outputs by a restrictive safety zone then
economic development programs must focus attention on the special problems
of the subsistence sector. Such efforts should concentrate on expanding
the safety zone., and will require at least two special aspects:
1. institutional arrangements to lower the private costs to the
farmer of a mistake;
and 2. institutional arrangements which are predictable and dependable
both within and between planning horizons.
f


V-8
What does each of these entail? With respect to the first, there
are several things which governments can do to reduce the costs of a
mistake., Partial contracting for output is one possibility. Programs
to assure availability of seeds, fertilizer, irrigation water and other
inputs will reduce the chance that the farmer will make a commitment to
modern crops only to find that necessary inputs are not available. The
guaranteed "loans" in U.S. agriculture are a possibility for reducing
price uncertainty. And there are other possibilities which will not be
pursued here.
But the major changes will come in the creation of institutional
arrangements which are dependable. By this one should not infer that
institutional arrangements must never change, nor does it mean that the
farmer should expect a perfectly predictable world. But it does mean
that institutional "tinkering" and continual "adjustments" must be
minimized. The subsistence farmer is constantly forced to second guess
the nature of the economic environment--and while all farmers must do
likewise, the limited safety zone of the subsistence farmer renders this
cautious behavior particularly stifling.
C.               In essence thm, economic development is the process of creating the
conditions.in which subsistence farmers can depend upon a small economic
sufrplus in excess of subsistence, safety, and consumption requirements.
C           This surplus--economic rents--can then be used to undertake technological
change at the farm level. Such modifications in traditional practices
then creates a demand for infrastructural changes in the wider rural
economy.


V-9
B. ECONOMIC DEVELOPMENT OF SMALL-SCALE IRRIGATION
.While the problems of the subsistence farmer have been talked about
previously, our special concern lies with small-scale irrigators. We are
reasonably safe in continuing to consider small-scale irrigators as part
of the subsistence sector, but one further refinement is called for.
That is, we must introduce the added uncertainty and constraint that is
present in irrigated agriculture. Consider Firgure V-3. Here panel (a)
is a replica of panel (d) from Figure V-2.
We are interested in the farmer located somewhere along a watercourse
other than at its head, and we assume that this farmer is susceptible to
the influence of at least one up-stream farmer. As outlined in Chapter II
this will show up in terms of reduced water receipts over that possible if
all irrigators received the amount implicit in project design for their
area to be irrigated. This can be depicted in panel (a) of Figure V-3
as a new constraint in output space--a constraint which takes precedence
over the previously binding land constraint. For the small-irrigator,
land is no longer the relevant constraint in crop choice and output.
'Instead, it is i  quantity and timeliness of water for the crops.
Because modern ciops are less forgiving of imprecise water application,
there is no need,for this constraint to be parallel to the now-irrelevant
lhnd constraint. Indeed there is good reason to assume that it departs
from the slope of the land constraint as shown in panel (a).
When the same safety zone is superimposed the full impact of uncertain
water receipts is highlighted. Now the small farmer is further away from


(a)                                a(b)
C
cr- o'
.K T
SUBSISTENCE                                                      M
TRADITIONAL CROP                                      TRADITIONAL CROP
FIGURE V-3. IRRIGATION AS THE BINDING CONSTRAINT.'


the capital constraint, indicating that capital is being under-utilized
even"on the very smallest farms. Additionally total revenue has fallen
by the magnitude (P M)(D)o as shown in panel (b).
The position of the water constraint warrants some further comment.
The traditional pattern in irrigation-project development is to strive
for a situation in which land and water availability are such that "normal"
crop practices can be followed on each farm located on the project.
This "lump-sum" allocation of water to a given area of land means
that neither should--in theory--be more constraining than the other.
When water receipts exceed that needed for the production of a crop then
land becomes the binding constraint of increased production. When water
is received in quantity less than that required then it--and not land--
becomes the binding constraint. This is the situation shown in Figure V-3.
In light of this, one possible policy objective might be to insure that
land and water are coincidental constraints on crop production.
C. SUMMARY
Economic development of small-scale irrigation will require the
explicit recognition of three conditions; programs conceived and implemented
in ignorance of these conditions are destined to failure. With an
abundant supply.of unsuccessful development programs world-wide, there
seems little reason to create more.
The first condition is the physical interdependence of farmers
along a watercourse. The second condition is institutional uncertainty.
The third condition is lexicographic decision making by small farmers,
with subsistence and safety goals dominating profit maximization.


V-12
Physical interdependence is an unavoidable fact in irrigated
agriculture. There is no way to provide water to more than one farmer--.
even with ground-water pumping--that does not result in this physical
linkage. Development programs for small-scale irrigation can only hope
to make the best of this situation.
Institutional uncertainty is a fact in the developing countries
because of the existence of the soft state. However, in contrast to the
physical laws which produce the interdependence among farmers on a water-
course, there is nothing immutable about the soft state and institutional
uncertainty; they exist because of an unwillingness to eliminate'them.
Lexicographic decision making exists because of the existence of
institutional uncertainty. While it is true that farmers everywhere are
cautious maximizers, the existence of the soft state compounds the usual
uncertainties of agriculture. This uncertainty, coupled with the marginal
existence of the majority of small farmers, makes subsistence and safety
dominant. When institutional uncertainty is reduced, subsistence farmers
will be able to become less safety conscious.


VI-1
CHAPTER VI. REGIONAL CASE STUDIES
The foregoing four chapters have been devoted to a discussion of
the theoretical issues in small-scale irrigation. In this Chapter we
will turn our attention to an examination of irrigated agriculture in
several regions. The intent is to lend realism to the rather abstract
discussions of the previous material. This treatment will also provide a
reference point for the following chapter which is concerned with program
possibilities for improving small-scale irrigation. Here we will discuss
irrigated agriculture in four disparate regions of the world: (1) Mexico;
(2) North Africa; (3) Pakistan; and (4) Asia. The literature to be
utilized in the discussion represents but a small fraction of the extant
writing, and also a small fraction of that reviewed in the preparation of
this report. However, it was selected for special notice because of the
issues it treats, and the problems it identifies.
A. AN EXAMPLE FROM MEXICO
In a carefully documented study entitled "Irrigation, Conflict, and
Politics: A Mexican Case" Eva and Robert Hunt present a relevant picture
of irrigation in a small Mexican town (San Juan in the state of Oaxaca)
[1978]. This is arid country in which irrigation is absolutely essential
for the production of corn, tomatoes, rice, mangoes, chicozapotes, and
sugar cane. Several major feeder canals serve the fields and orchards
which surround San Juan, with minor ditches spread over the outlying -
irrigated land of the municipio. In the town of San Juan there are two
town canals using water from the Chiquito River; one canal serves the


VI-2
house orchards and land around the town, the other serves the "Grasslands"
where most of the small landowners are located. In theory every resident
can have water from the town canallsi though only some avail themselves of
this particular source. Irrigation is under control of the town water
commission (La Junta de Aguas). This is a most unwanted position since
commissioners receive inordinate pressure for favors in water allocations.
When the poor are on the commission they suffer from the intreaties of
the powerful landlords to whom they are otherwise indebted. The rich do
not want to serve since it is not necessary in order to obtain their
water needs.
There are two "water masters" who make the everyday decisions, but
instead of one master per canal as intended, they work together to
discourage attack from irate i.rrigators. They are, however, rewarded for
the hazards of their job; their bribe income during the dry season is
sufficient to allow a relatively comfortable existence. Even after having
paid the bribe, an aspiring irrigator must stand on the ditch to ensure
that the bribed water masters indeed give him his water. Even though
evaporation losses are low during the night, few farmers choose to
irrigate then for this is to invite widespread water stealing. The
earlier discussions of the "soft state" refer to just ttis phenomenon.
Maintenance work is derived from the amount of water received by an
irrigator, however in practice only the lower classes perform any
maintenance. The wealthy hire laborers to perform their share of main-
tenance.


VI-3
A second irrigation network in 'the municipio is represented by two
privately owned canals which serve the land of the wealthy sugar farmers.
There is some selling of water from these two canals, though the bulk of
it is used by the landowners whom it is intended to serve.
The third system is the "ejido" canal which serves the land of the
poorest members of the community. In contrast to the other systems, the
officers in charge of water allocations, conflict resolution, and ditch
cleaning are not separated from the other roles central to ejido life.
The social organization of San Juan is as one would expect of a small
rural town. Approximately 10 percent of the 2,500 residents belong to the
elite class (la clase alta), another 10 percent belong to the "middle"
class (la clase media), while the remaining 80 percent are the poor
residual (los peones). The elite own the bulk of the land and the commer-
cial enterprises, as well as the best irrigated land. The "middle" class
derives income primarily from white collar service jobs and in the retail
establishments owned by the elite. This group owns a small amount of
irrigated land, but disdains agricultural pursuits. The peones own little
and derive their income from wage labor. There is a small amount of
irrigated land owned by this group, but it is largely insignificant.
Hences we have a situation which the Hunts maintain is rather typical
of Mexico in which land ownership--and hence water control--is in the
hands of approximately 10 percent of the local population. The elite own
all of the commercial establishments in town, which together with their
agricultural wealth probably accounts for 90-95 percent of all income


VI-4
earned in the local economy. To imagine that the small farmer has any
meaningful control over irrigation water in times of scarcity is difficult,
if not impossible.
B. IRRIGATION IN NORTH AFRICA
The discussion in Chapters III and IV about subsistence farmers may
have left an impression that innovation among this group of farmers is a
rare phenomenon. However it is necessary to point out that innovation is
not inconsistent with subsistence agriculture as long as that innovation
is perceived as improving the chances of the small farmer. A recent study
by Barnett [1979] in the Sudan concentrates on the innovative tendencies
of the subsistence farmers, and the unwillingness of bureaucrats to change.
There is an important lesson for irrigation in this case study as well.
The Gezira Scheme covers approximately 800,000 hectares between the
Blue and White Niles in the Sudan. This is an old system, with the
original irrigators coming from the ranks of the pastoralists. Given
their unfamiliarity with irrigated agriculture, there evolved a detailed
set of procedures, and a hierarchical administrative structure which was
also rather authoritarian. This control extended both to crop selection
and practices, and to water allocation.
Barnett points out that engineers--not agriculturalists--dominated
ea*rly'decision making and hence irrigation procedures tended to be more
responsive to physical aspects of the system than to agronomic requirements.
These precise engineering objectives gave way to bureaucratically inspired
and enforced "rules of thumb." One of the open conflicts revolves around
the bureaucratic interest in cotton production, and the farmer interest
in dura production for fodder.


VI-5
What has transpired in the Gezira Scheme is that the tenant farmers
have innovated significantly in irrigation matters, while the bureaucrats
have stuck to the traditional methods--and apparently refuse to recognize
that irrigation practices now differ markedly from what is articulated in
the "rule books." Specifically, night watering is now prevalent.
Additionally, the fil-ed rotation schedule has been altered to permit
watering when water requirements dictate rather than on a fixed rotation
which failed to recognize seasonal and age differences in plant water needs.
Finally, the farmers have developed an irrigation method which economizes
on the scarcest factor of production--labor; this too was in spite of
official rules to the contrary.
Barnett argues that these innovations have been a major factor in
contributing to the intensification and diversification of local irrigated
agriculture, even though the official irrigation bureaucracy refuses to
admit that farmers are not following the rules. The drive for innovation
came from the tenant farmers who are in a perpetual state of indebtedness
and cash-flow deficiencies. Moreover, the irrigation scheme has weakened
the extended family such that labor is the scarcest factor at several
critical stages of agricultural production. The farmers, in their efforts
to economize on.the most limiting factor, have innovated in several
imiportant respects. The government requirement that they grow cotton has
constrained them somewhat, but they then managed to innovate within the
confines of this constraint. It is the bureaucrats who lag behind.


VI-6
When water is scarce the farmers ignore the requirements to use it on
cotton and instead irrigate their own dura first. As a result the
prescribed 14-day rotation for cotton irrigation is sometimes extended to
20 days, and even to 25 days, Barnett indicates that, "Paradoxically they
seem to achieve better yield results with this practice.. .[1979, p, 66]."
The only remaining puzzle is how this flagrant abuse of official
irrigation practices and priorities can continue. Apparently, according
to Barnett, the news never reaches up to the top of.the very centralized
irrigation bureaucracy; the reports prepared at several levels below the
top fail to transmit this departure from authorized practices. In this
example, the usual bureaucratic "information loss" is turned upside down.
The standard situation is one in which irrigation management is seriously
deficient and hence system production is greatly inferior to what is
possible; but the top of the irrigation hierarchy somehow never seems to
find this out. On the contrary, Barnett tells of a system in which the
production and income of the system is probably quite good, and the top
of the bureaucracy is not told that it is because their guidelines have
been contravened.
C. IRRIGATION IN PAKISTAN
A team from Colorado State University has been studying irrigation
practices and problems in the Punjab and Sind zones of Pakistan. Data
collection took place in 1975-76 from a sample of 387 farmers in 16
villages on 40 watercourses. The sample was stratified along the water-
courses so that there are head, middle, and tail irrigators in the sample.
The findings of that extensive research effort will be summarized here
[Lowdermilk, et.al., 1978].


VI-7
The dominant aspect of Pakistan irrigation is revealed by reference
to major water losses in watercourses between the turnout from the main
canal (the mogha) and the entry point to the farmer's field (the nakka).
The research found losses ranging from 33 percent to 65 percent, with an
average loss rate of 47 percent. The average losses per 1,000 feet of
watercourse were 26 percent, or slightly over one-third cubic foot per
second. Water losses were found to be greatest in those watercourses with
the most water; where public tubewells augment watercourse supplies the
losses are greatest, while on those watercourses with private tubewells
the losses drop significantly.
The research reveals a high correlation between ample irrigation
water and overirrigation. Field application efficiency--an index of
the proportion of water entering the farmer's field which is stored in
the crop root zone--was highest for tail end farmers where water was less
available. Low field application efficiencies result in excessive tail
water which can contribute to both salinity and drainage problems. Field
application efficiencies were found to be correlated with farm size:
those farmers with over 10 hectares averaged 64 percent efficiency while
those with less than 10 hectares averaged 80 percent efficiency at the
field level. The larger farmers tend to have access to tubewell
supplies which contributes to overirrigation.
Yields of wheat, rice, and cotton were greater on those watercourses
with access to tubewells. It was also found that cropping intensities
were greater when tubewell water was available. Both of these differences
are attributed to the fact that tubewells permit greater water control--


VI-8
and hence reliability. The research reveals that timing of water receipts
is usually more important than the quanity of water received. Cropping
intensity was found to be a decreasing function of a farmer's distance
along the watercourse (from the head).
About 70 percent of the 387 sample farmers reported that they did not
receive word-from the irrigation bureaucracy in advance of closing canals
for maintenance. Agricultural extension workers were found--in this,
the world's largest contiguous irrigation system--to be untrained in water
management.
The soft state exists in Pakistan water management. The research
revealed that required watercourse improvements and procedures are ignored.
As with the Gezira Scheme, official sanctions against certain behavior--such
as water trading, water purchasing, modifications of moghas--are ignored
with impunity. Ironically, it is through such evasion that the system is
as productive as it is.
In a sample of 354 farmers the following question was asked: "What
do you presently perceive to be the major constraint to obtaining increased
per acre yields in your farm operation?". The responses to that question
are shown in Table VI-1. The importance of irrigation could not be more
vividly protrayed. Middle and tail-end farmers identified water problems
as serious about twice as often as did nead farmers.
The bulk of the farmers in the research sample receive no informa-
tion on plant water requirements, stages of plant growth, or the importance
of plant maturity for water requirements. Sixty one percent were found to
believe that wheat requires more water than cotton--or respond that they


VI-9
TABLE VI-1. ALLOCATION OF FARMER RESPONSE IN ANSWER TO QUESTION: "WHAT DO
YOU PRESENTLY PERCEIVE TO BE THE MAJOR CONSTRAINT TO OBTAINING
INCREASED PER ACRE YIELDS IN YOUR FARM OPERATION?", Sind and
Punjab regions of Pakistan (n=354)
Major  Constraint                                    Percent of Reports
Insufficient supplies of irrigation water                 73.0
Lack of fertilizer and improved seed                       9.3
Improved implements and farm machinery                     6.0
Lack of land                                               2.8
Lack of capital or credit                                  2.0
Lack of insecticides                                       1.7
Lack of extension services and improved,road               1.7
Seasonal labor shortages                                   0.9
No major constraint                                        2.6
100.0
Source: Lowdermilk, et. al. [1978].


VI-10
do not know. Of 378 farmers, 47 percent reported that they stopped
irrigating when water reached the far border, and 33 percent stopped
irrigating when all of the "high spots were covered." Farmers share a
common notion that crop roots penetrate only a few inches into the soil
and that 5 inches of water applied will infiltrate to a depth of only
12-24 inches.
Eighty-five percent of the farmers reported no contact with either
an agricultural assistant or a field assistant from the extension service
over the previous three months. It is not surprising, therefore, that
Lowdermilk, et.al. are at great pains to emphasize the need for improved
extension efforts in agricultural water management. The other obvious
deficiency is in farmer and bureaucrat discipline with respect to
compliance with established rules concerning water allocation, watercourse
maintenance, and dealing with instances of water stealing, bribery, and
ditch cutting.
In 'a second major study in the Pakistan Punjab Parker [1979] was con-
cerned with the institutional environment along a watercourse, as well as
farmer water receipts (quantity as well as timeliness) as a function of.
physical location along the watercourse. Parker utilized a three-tiered
model which depicted farmer water control as a function of laws and regula-
tions,.watercourse transport losses, location along the watercourse, the
farmer's ec6nomic status,"and several sociological factors. The second
tier considered the adoption of high-yielding varieties as a function of
water control (from tier one), input availability, knowledge of proper
techniques, and willingness to change. In the third tier crop yield was


VI-11
a function of technology used (from tier two), soil characteristics,
actual water received, and exogenous factors such as weather, disease, etc.
Parker selected two sections of the Puhjab province--Khanewal Tehsil*
in Multan District, and Lyallpur Tehsil in Lyallpur District. The primary
crops in Khanewal are wheat in the winter (rabi) and cotton in the summer
(kharif). The primary crops in Lyallpur are wheat in the winter and sugar4,
cane in the summer. Sample villages were chosen at random from the command
areas of single major canals within each of the two regions. Individual
fannerrespondents were chosen using a stratified random sample from the
head, middle, and tail sections of each watercourse associated with each
sample village. Ten respondents were chosen from each watercourse, allocated
to head, middle and tail in the same proportions as the total population of
irrigators on the watercourse.
Parker!s findings reaffirm the general picture concerning the important
variables in the ability of farmers to exercise reasonable control over
water receipts (farm size, farm location). Additionally, the results attach
special significance to variables concerning the total number of farmers
on a watercourse, and the number of farmers upstream from the respondent;
this is also modified by the number of farmers on the watercourse deemed.
to be "uncooperative." It was found that the number of uncooperative farmers
increased as the number of other farmers with whom the respondent must deal
over water receipts increased. The proportion of uncooperative farmers
had a negative effect upon-water receipts and satisfaction of watercourse
cleaning.


VI-12
In the second tier of the model Parker found that the adoption of
modern agricultural practices was greatly influenced by the quantity of
water received. However, the timing, reliability and security of water
receipts were instrumental in the choice of fertilizer use. In the crop
production aspect of the model the findings reaffirm the importance of
water control and of water-dependent technological variables. That is,
greater water control eihances the farmer's opportunity set.
Parker concludes by stating:
Farmers who are advantaged by size of land holding, by social
status, or by watercourse location (both in terms of physical
aspects of location as well as in relation to other farmers)
tend to have much better water control than do other farmers.
This superior water control, by boosting the yield levels of the
privileged large cultivators, tends to exacerbate farm income
differentials. Efforts to equalize water control between
farmers could be a promising method of improving income dis-
tribution in rural areas [Parker, 1979, p. 184].
D. IRRIGATION IN ASIA
When compared with the rest of the world, Asian rice irrigation is
usually considered to be the most efficient, the most rational, and the
least -ibject to the sorts of problems just described for Pakistan.
While it is true that there are outstanding irrigation projects in Asia,
it would be a fallacy of composition to generalize.
A-recent paper by Levine [1977] illustrates the differences which
exist in Asian irrigation. The basic model of Asian irrigation is
presented in Figure VI-1, with system water requirements plotted against
the degree of water control. Under climatic conditions present in most


V I-13'
2500
2000
 1500                             SYSTEM REQUIREMENT
1000            .
500                           EVAPOTRANSPIRATION
~ ~~     ~           ERCULATN~AfD ~SEPGE         -    -    -
0                           LAND  PREPARATION
INCREASING. WATER   CONTROL
FIGURE VI-1. SYSTEM WATER RE.QUIREMENTS AND WATER CONTROL.
Source: Levine [1977].


VI-14
of Asia, it is generally considered necessary to provide between 600 and
750 im of water per season of rice (between 95 and 110 days depending
upon the variety). With near "perfect" water control it is possible to
provide just this amount of water to the system. However, as water
control dim.hishes along the watercourse it becomes necessary to provide
increasing quantities to the system so that what remains at the farm
turnout is adequate to meet crop requirements, evaporation, percolation
and seepage, and other requirements. To highlight the differences among
Asian countries, Levine also includes a diagram upon which Figure VI-2 is
based. Here it can be seen that the Tou Liu project in Taiwan is the
epitome of good water control, but that the average for Taiwan is some-
what inferior. However, Taiwan is still superior to Malaysia, and especially
to the Philippines. That is, the efficiency of Philippine irrigation is
estimated to be on the order of 20-25 percent, for Malaysia it is estimated
at 40 percent, while in Taiwan the average is over 60 percent. Recall
that efficiency pertains to the percentage of water reaching the farmer's
field as a percentage of that turned into the system which is "alloted"
to that field.
The obvious question then becomes to determine why the Tou Liu
system is so efficient. Levine provides part of the answer in terms of
rotafional irrigation within 50 hectare units according to a strict plan.
Other aspects of the Tou Liu project include some of the laterals being
concrete lined, control gates and Parshall flumes at each 50 hectare
turnout, extensive networks of farm ditches, and 24-hour irrigation schedules.
In contrast, the Malaysian system is based upon continuous irrigation,


VI-15
2500 m
1400 mm
1000 mm
650 mm
E~-- 4
FIGURE VI-2. TYPICAL ASIAN IRRIGATION SYSTEM WATER REQUIREMENTS.
Source: Levine [19773.


VI-16
but control only within the primary and secondary canals. Beyond the
canals, water distribution is in the hands of the farmers, and few farm
ditches ,exist. Finally, the Philippines systems are also based upon
continuous irrigation, however there are few effective controls in the
channels and the turnouts, channels are not well maintained, their main-
tenance is deficient, there are few measuring devices, control over water
is only exercised five days per week (for eight hours each day), and
farmer cooperation is minimal at best.
These observations describe a situation, but they are mere symptoms
of something else--and Levine correctly identifies that other element as
the country's perception of the scarcity of water. There are two aspects
of scarcity which are relevant to water use in agriculture. The first is
the scarcity of water to the individual farmer and to the broader society,
while the second is the security of water receipts to those who "ordinarily"
use water. Each will be discussed in turn.
Dealing first with water scarcity, we begin by exploring the conditions
whereby water might be considered scarce by individual farmers; here it
is necessary to discuss the operation of the current irrigations systems
in the.various countries. Consider three types of farmers: (1) those
on an irrigation.system who regularly receive water; (2) those on a
system who only sometimes receive water; and (3) those near a system with
irrigable land who receive no water. Clearly for the latter two groups
water is indeed scarce. However, we need to distinguish between nominal
scarcity and real scarcity. And this is where we need to look at water
use by farmers in the first group. Agronomic research has determineed--


VI-17
for a variety of climatic conditions--water "requirements" for virtually
all -irrigated crops. While an economist might be interest in the optimal,
application of water vis-'a-vis other inputs, we must recognize that plant
stress gives some lower limit on water application. Once some reasonable
level of water has been provided to the irrigated crop, it is possible
to begin to explore the extent to which farmers apply more than this amount.
However, as part of that determination it is necessary to bear in
nir  .her aspects of the production process. For example, an effective
method of weed control on rice is flooding--here water is applied in
excess of direct plant requirements but it serves as a substitute for
other inputs such as herbicides or manual weeding. Or consider the issue
of leveling fields for rice; it is possible to imagine an array of
"minimum water applications" under various assumptions regarding field
leveling. Here, "extra" water is a substitute input for the time and
resources necessary to bring individual paddies to perfect (or near-
perfect) plane. The same applies to ditches that are ill designed, or
badly maintained. If water is cheap to the individual farmer then it
becomes a viable substitute for other inputs. This will be referred to
as nominally scarce water. That is, an irrigation system is operated
such that cheap,water to those fortunate enough to receive it regularly
replaces the use of other inputs. The price paid by other farmers is
one of no water for lands that are irrigable, or infrequent water for
those on land served by a distribution network. To the extent that water
is scarce to these latter two groups of farmers, they adopt crops and/or


VI-18
cultural practices where it is possible to survive without water; the
private cost of this situation is the foregone income to the individuals
from the lack of water.
But there is a social cost as well. Consider first the situation
where an irrigation system is designed and constructed to serve a certain
area and a specified number of irrigators. In any system design there is
a presumption of efficiency of water delivery over the system, and of
field leveling. If, in practice, these conditions differ from the design
assumptions, there will be less water for those at the end of the system;
this is compounded by a water allocation scheme which allows those near
the head of the system to assure themselves of their water "needs." The
upshot is an investment which has been undertaken on one set of production
possibilities, and realized on,an entirely different set. That this is
serious to the agricultural planning of a developing country should require
little elaboration here; to the extent that the investment was financed
by external debt it should also be obvious that serious foreigh exchange
implications attend such situations.
Another social cost worthy of discussion is the use of water to
replace labor in leveling and weeding operations; if labor is abundant-'
as is ofteri the case--this substitution of a socially scarce factor for
an offen abundant factor may be serious.
Yet another aspect of social cost is the aggregate production fore-
gone by those producers near an irrigation system who were not included
in the system because of an apparent lack of water to serve their needs.
This point relates to the set of design assumptions involved in determining


VI-19
the command area of the system. Or, it is possible that there is an
abundance of irrigable land near an existing system which is now vacant,
because it will not support agriculture in the absence of water. If
irrigated agriculture is a necessary ingredient to an agrarian .reform pro-
gram;,. then this apparent lack of water may stifle efforts on this front.
This aspect is even more serious if steep slope agriculture is practiced.
That is, if farmers are confined to steep slopes because of an apparent
lack of irrigable land upon which they might be settled, then they pay,
and the country pays twice. The farmer is impoverished because of a poor
resource base (and one which will--in all likelihood--get worse); the
country is absent the increased production which could result from
production under irrigated conditions, and the erosion and resource
depletion brought on by steep slope agriculture not only makes the nation's
land base poorer, but siltation may speed up the demise of existing
irrigation systems. This is a high price indeed for permitting a situation
of nominal scarcity to continue.
Water is only scarce in a real sense if irrigation systems are
designed and operated with a high degree of efficiency, and much attention
is devoted by those organizations responsible for irrigated agriculture
to assurean efficient allocation of water on lands with irrigation
potential.
Throughout the foregoing we have not mentioned the matter of compe-
tition for water between agriculture and energy production. With hydro-
electric production it is often possible that the use of water for one
acitvity precludes its use for another. If water is used in abundance on


VI-20
some farms that may also imply an additional social cost by precluding
its use for the generation of electricity. Similarly, if it is used for
the generation of electricity it may preclude its use in irrigation.
This issue confounds the determination of the real social cost of water
use.
Yet another aspect of the social cost of water use is that of
salinity and water logging of soils. For those farmers now receiving
water, the excess application has a cost both to them, and to others.
Water logging and salinity reduces their yields, but it also influences
the salinity of receiving waters. If others then utilize this saline
water their yields are reduced.
Thus, the issue of scarcity has both a private component and a social
component. Nominal scarcity refers to those situations in which water is
apparently scarce, yet that scarcity derives from the particular way in
which water is currently managed, or in which individual fields are leveled.
Real scarcity refers to situations where "good" water management and
irrigated agriculture are practiced and yet there exists irrigable land
which might be cultivated if more storage facilities were constructed.
Levine points out that as countries have begun to realize that good
arable land is not in infinite supply, that genetically superior rice
varieties are available, and that water supplies are becoming more difficult
to develop, they suddenly realize the scarcity of water--and the social
costs of its current use. Once that is recognized, then the effort to
begin improved water management suddenly appears in a more favorable light.
The difference between Taiwan and the other Asian countries studied


VI-21
(Malaysia and the Philippines) is that water has been recognized as a
soci-ally scarce input for approximately 50 years. Levine points out that
it took the impetus of the 1954-55 drought in Taiwan to drive this point
home definitively, but since that time there has been a relentless push
to increase the efficiency of water use on the island.
It would take us beyond our purpose here to study in great detail
the Taiwanese irrigation systems. However, there are some general principles
which merit brief discussion. A recent publication by Abel will be help-
ful in this regard [1977]. Abel identifies four essential factors which
contribute to the high degree of efficiency of Taiwanese irrigation.
First is the explicit recognition that water is a scarce factor of prod-
uction to be used as efficiently as possible. Second, the government of
Taiwan has evolved a system of centralized planning of irrigation invest-
ments, but decentralized management of the systems. Third, within the
irrigation assocations information systems have been developed which
permit the ready exchange of agronomic and engineering information between
farmers and the managers of the systems. Fourth, the irrigation associa-
. tions utilize systems of incentives for managers of the system as well
as for the users of the water.
These.four.conditions combine to create an irrigation system in
viiich'there is quick and accurate information available to all partici-
pants, a shared recognition of the importance of good water management,
and a recognition of the need for discipline, order, compliance, and
cooperation. Not all farmers receive all of the water they would like,
nor necessarily when they want to receive it. However, they know that


VI-22
others are treated similarly so there is less inducement to "break rank"
and vationalize it with reference to others getting more. Allocating
water among farmers is like dividing cake at a children's birthday party.
Chaos is apt to prevail in the absence of shared expectations of one's
share.
In Taiwan the systems are essentially owned and managed by the farmers,
and the water management personnel work directly for the farmers. The
cooperative water-user o'rganizations hire and fire managers of water based
on their performance. On the continuum of soft states there can be
little doubt that Taiwan is closer to the "hard" end. And it is-this
aspect which is necessary for a system to establish any expectation on the
part of the users that water will appear at a predetermined time, and in
some relation to the planned for quantity. Indeed, Abel points out that
farmers know how much water they will receive before the planting decision
is made. As indicated previously, there is an abundant literature on the
response of farmers to insecure water receipts. The major findings
indicate that water supply--which must include some notion of reliability
and security--is the most important factor in determining which crops to
grow and what areas will be planted. Moreover, water supply is often
the dominant variable in determining yield differences among crops. Also,
water supply is found to be a dominant factor in the adoption of new
inputs--including high-yielding varieties. In addition to the production
implications'there are seriods equity concerns in water security; the
least advantaged farmers are usually the ones to pay the highest price for
insecure water. With a loose organization controlling water allocation
C


VI-23
the poor and powerless are the least able to influence water distribu-
tion, and they are also the ones least able to mobilize an imaginative
response to the insecurity of water receipts.
As before, there is a private cost of an irrigation system which
cannot insure secure deliveries, and there is a social cost. One of
the private costs has already been discussed, that of the reduced produc-
tion from those farmers who do not adopt improved technologies. But the
aggregate cost of this failure to improve cultural practices is that total
production is less than it could be with improved management of the
systems. Given the risk aversity of poor farmers, the insecurity of water
deliveries is all the excuse necessary to induce them to continue their
traditional methods. If part of the agricultural strategy of a country
is to spread improved technology among a large number of farmers then
the degree of water security becomes an important limiting factor.
E. SUMMARY AND IMPLICATIONS
J                 It is impossible to provide a comprehensive overview of world irrigated
agriculture without the commitment of a great deal of time and space. The
examples offered here cannot hope to present more than general impressions
from a few scattered situations. Their inclusion is based on the fact'that
the common elements support the general theoretical issues developed in the
earlier chapters. In the Mexico setting we see that the wealthy and in-
fluential farmers do not need irrigation associations in order to receive
their water "needs." We see that when the poor farmers serve on the associa-
tion they are threatened by their powerful neighbors. We see :some violence,
bribery, .and a general lack of enforcement. We see a situation in which


VI-24
only the poor farmers spend time maintaining the irrigation system, with
the rich hiring others to do their work. While not necessarily bad in it-
self, too,much "absenteesim" can sometimes have a negative effect on the
viability of the water users' association. Finally, we see an extreme
maldistribution of income in the local economy, and this implies a mal-
distribution of power and control. The management of the joint input can-
not help but give rise to the sort of inefficiency discussed in Chapter II.
In the Sudan we see an irrigation bureaucracy and a set of rules con-
cerning water allocation which is out of touch with the needs and priorities
of the water users. Because of this we find widespread disregard for the
rules.
In Pakistan we find very low efficiency of water use, with extensive
over-irrigation in some fields, and insufficient water for others. We see
field efficiency -- the proportion of water reaching the field which is
stored in the root zone -- decrease with greater water availability and
with increased farm size. This measure of efficiency was Towest where
farmers had access to tubewells. The latter finding indicates that the
management of the system is grossly inadequate since the availability of
groundwater has not brought about a rational plan for 'conjunctive use of
both surface and ground water. Under ideal conditions plans for conjunctive
use would be developed so that each source is used to the greatest advantage.
The results from Pakistan also indicate that the timing of water receipts
is of much greater importance than is the quantity received. This point
relates to *the theoretical discussion in Chapter II. There, although the
diagrams are developed in terms of quantity,tthe underlying issue is one
of reliable expectations concerning quantity and timing of water receipts.
The Pakistan example also highlights the fact that cropping intensity
is a decreasing function of the distance from the head of the system. This,
C


VI-25
too, is a logical result from the uncertainty faced by those farmers near
the tail-end of the system. They soon come to expect the worst with respect
to water receipts and are unwilling to undertake the risks of a crop unless
water is reasonably secure.
The Pakistan example also calls attention to the conditions which are
present in the other settings. That is, there is little communication be-
tween the irrigation bureaucracy and the farmers. Moreover it was seen
that there has been no training in water management for the agricultural
extension workers. In light of this there can be little doubt that water
use is non-optimal with respect to-plant needs. Since this will usually
mean overwatering, we see again a possible reason why those near the head
of a system would be using more water than is necessary, and thereby de-
priving others less advantageously located of sufficient water.
Again focusing on the interdependence among system farmers, Parker's
research -- as well as that of the Colorado State team -- found that water
was a more crucial constraint for increased yields by those farmers at the
middle and tail sections of irrigat d systems.
Finally, in Asia, we find a very high degree of variability in rice
yields from country to country. The hypothesized causal factor here --
though we did not explore strict causality in this report -- is the efficiency
with,which water is managed within the country; the degree of intensifica-
tion .of irrigated agriculture. As in the previous examples, the inefficiency
of irrigation results from an institutional vacuum which means that the
joint input is not really managed but merely taken by those who have access
to it.
The uncertainty so prevalent in the irrigation systems of the developing
countries means that input usage among farmers is inefficient. When these


VI-26
inputs are extremely scarce both to the farmer as well as to the nation --
as is the case with fertilizer, pesticides, and modern seeds -- then the
country and the farmer pay a dear price for this inefficiency. This not
only means that individual yields are below what is possible, but aggregate
production is less than what it could be for each country. With the bulk
of the developing countries barely able to feed their expanding populations
this is a severe price to pay. This uncertainty shows up in farmer behavior
by causing extreme caution and a preoccupation with safety-first decisions.
The practical result of this is that experimentation -- the very essence
of a dynamic agricultural sector -- is stifled. When that dampening affects
the subsistence sector differentially -- as it does -- then the very sector
most in need of experimentation is held back.
In terms of the theoritical discussion of Chapter II, the agricultural
sector is interior to the production possibility frontier -- or, to say
the same thing, it is off of the conflict curve which is the locus of
efficient points. Distributional issues enter as well since it is the
small farmer -- or the one near the tail of a system -- who pays the greatest
price for the current mis-management of water. To the extent that small
farmers comprise an important target constituency then irrigation is an
important policy instrument.


CHAPTER VII: IMPLICATIONS FOR SMALL FARM DEVELOPMENT
A. INTRODUCTION
The small irrigation farmer is a cautious optimizer, which means that
safety is placed ahead of profit maximization. Given the fickle and un-
certain 'world in which the small farmer must operate, survival is enhanced
by such lexicographic decision making. When situated on an irrigation
canal with other farmers of equal socioeconomic status, the physical inter-
dependence introduces an added degree of uncertainty--and a requisite burden
of "transaction costs"--merely to obtain water in the needed quantities and
at the appropriate times. When the small farmer is situated on a canal
with farmers of higher socioeconomic status this troublesome situation is
exacerbated even further. Indeed, in many instances the powerless farmers
simply do not receive irrigation water.
This aspect of irrigated agriculture in the developing countries intro-
duces two costs to the economy. The first is an efficiency loss--aggregate
production is less than it could be if the irrigation systems were not
characterized by low-grade anarchy. The second is an equity cost in that
the avowed objective of many governments to look after the interests of the
small farmer is simply ignored. In a world in which small farmers already
pay higher per-unit costs for the same inputs than their lariger and wealthier
cohorts, this further burden from a technology that claims to promise scale-
neutral benefits is a cruel joke.
Survival in subsistence agriculture is the result of adaptive behavior.
It can take.several forms., but caution is one manifestation, as is distrust
of yet more government "assistance." Some of them can no longer afford:,to
be "helped" by government.


VII-2
The case studies illustrate--for a variety of situations-- the problems
faced by small irrigation farm2rs. The lesson is that there is either the
wrong type of control (as in Gezira) or no control (in the others). The
improvement of irrigated agriculture cannot proceed in this administrative
vacuum. The physical interdependence wrought by irrigation requires an
administrative system cognizant of this interdependence, and structured in
such a way that the interests of the small farmer are given protection.
I will refer to this general process as intensification.
The process of agricultural intensification has been detailed by
Boserup in her The Conditions of Agricultural Growth [1965], and .Wilkinson
elaborates on the theme in Poverty and Progress: An Ecological Perspective
on Economic Development [1973]. In a sense, these two books build on the
seminal work of Geertz in Agricultural Involution: 'The Processes of
Ecological Change in Indonesia [1963], although Geertz is cited in neither
study. Intensification is the increasing application of labor and capital
to a given land base in response to population growth and the need to
increase production of food. Boserup's thesis is that agricultural
development is caused by population trends rather than the other way
around, and that agricultural development is essentially the result of a
C


VII-3
process of intensification. We do not need to enter the dispute over the
direction of causality to report on some interesting correlations between
yields per hectare, agricultural land per capita, and the degree of
intensive agriculture (as reflected in the degree to which irrigation
water is carefully controlled).
The purpose in such a discussion will not be merely to advocate
better water control. The intent is to establish a relationship between
a productive agriculture--as renlected by yields per hectare--and the
application of management to the total agricultural enterprise (which
must include the management of water). For it is the proposition here
that +:,ose countries in which poor water management is occuring are
precisely those countries in which a minimal effort is expended toward
the intensification of agriculture. And it seems safe to conclude that
the only way these countries are to stay ahead of the growth in their
population will be to increasingly intensify their agriculture. This
does not rule out the bringing of new land into cultivation--though we
must hypothesize that the best land for agriculture is already so used.
But it does highlight the fact that these countries already have a
substantial area under cultivation, albeit with very low yields. Indeed,
we will see that the "land-abundant countries" (vis-a-vis population)
are currently experiencing rice yields per hectare of less than half
the yields in the "land-scarce countries."
First consider Figure VII-1. Here we see a plot of rice yields in
tons per hectare against the historic growth of rice yields in Japan; as
a norm few would doubt that Japan's agriculture is "productive," highly


VII-4
"developed," and that the degree of intensification--including water
management--is pronounced. It is also beyond dispute that Japan's
agriculture has not always been so intensive. One reads Figure VII-1
by noting that Laos is where Japan was in 900 A.D., the Philippines is
where Japan was in 1400 A.D., and Malaysia is where Japan was in 1900 A.D.
Now, using data on estimated agricultural land per capita, it is possible
to plot current rice yields in some of the countries in Figure VII-1
against this land availability. This is depicted in Figure VII-2. There
is little doubt that land scarcity is important in if one wants to form
an initial estimate of rice yields.' We are not, it should be clear,
talking of causality. We are talking about agricultural intensification
in countries where production at the extensive margin of agriculture is no
longer possible. And, we are talking about the yield potential that
intensification helps to realize.
While it would be possible to marshall a variety of statistics to
prove the degree of intensification in countries such as Japan, Taiwan,
South Korea, and China, our interest here is in water management. Recall
from Chapter VI that irrigation efficiency varied considerably as between
Taiwan, Malaysia, and the Philippines. We might consider the efficiency
of irrigation as an indication of the degree of intensification of a
coiuntey's agriculture, and we find that rice yields in these three countries
can be "explained" by irrigation efficiencies (Figure VII-3). As before
it should be emphasized that this discussion is not intended to establish
causality in the statistical sense. We are here talking about the merits
of devoting more time and financial resources to better water management.


VII-5
STRUCTURAL REFORM
PRIMITIVE            IRRIGATED        TECIINICAL
FARMING             CULTIVATION       INNOVATION
6                   2
JAf 'ANq
JUTH KO4A
NORTH KOREAz-
TAI 1,I
T AIWA N~
CNA
M ALAYSIA (
0                                        INDONESIA
PAK I,- AN,.
SRI LANKA r44
NEPAL       BANGLADESH
LAOS                       ILI  BIA
~KAPU/1T A LN D
K\MPUCHEA PHIL PPINES
600  700  800  900  1000 1100 1200 1300 1400 1500 1G00 1700 1800 1900 2030
FIGURE VII-1. HISTORIC RICE YIELDS IN JAPAN AND CURRENT YIELDS IN SELECTED COUNTRIES.
Source: Wortman and Cummings [1978].


VII-6
6          JAPAN
Lu              ,S. KOREA
,TAIWAN
w 4
#CHINA
44ALAYSIA
INDONESIA
*SRI LANKA                *PAKISTAN
2 * INDIA
BANGLADESH    NEPAL               -K*THPAILAND
PHILIPPINES    *APUCHEA               LAD)
0.10          0.20       0.30        0.40         0.50        0.60
S     .AGRICULTURAL LAND PER CAPITA (HECTARES)
FIGURE VII-2. RICE YIELDS PLOTTED AGAINST AGRICULTURAL LAND.


VII-7
5 .
TAIWAN
4.
PS
I-
* MALAYSIA
C3
2
* PHILIPPINES
* 1
25%             50%             75%           100%
IRRIGATION EFFICIENCY
FIGURE VII-3. RICE YIELDS AGAINST IRRIGATION EFFICIENCY.


VII-8
The point is that those countries in which such careful attention has been
paid to water management also seem to be those countries with extremely
high yields. To be sure, other things have gone along with better water
management; but this merely emphasizes the point made earlier that one
cannot consider water management in isolation from the total agricultural
enterprise.
The recent report of the Trilateral Commission Reducing Malnutrition
in Developing Countries: Increasing Rice Production in South and Southeast
Asia [Colombo, et. al', 1978] placed intensification at center stage in a
plan to reduce malnutrition in a part of the world containing the vast
majority of the world's poorest citizens, and where almost 3/4 of the world's
food grains are consumed. An earlier table also shows that the bulk of the
irrigated land is found in South and Southeast Asia. The cornerstone of
the Commission's plans to improve food production was better control over
irrigation water. Careful cost calculations were performed for several
irrigation improvement options. These were: (1) change uncultivated land
into adequately irrigated land; (2) change rainfed cultivated land to
adequately irrigated land; (3) change inadequately irrigated land to
adequately irrigated land; (4) change uncultivated land into inadequately
irrigated land; and (5) change rainfed cultivated land into inadequately
irrigated land.
The cost-effectiveness of the various plans was shown to vary consider-
ably, with options (1) and (4) being the least cost-effective. The lowest
capital costs for increasing paddy production by 1 ton per hectare per year
are through improving inadequately irrigated land to adequately irrigated
land (3), followed by improving rainfed cultivated land to adequately
irrigated land (2).


VII-9
The investments called for under the two favored plans were not large-
scale projects. Instead they consisted of digging out farm ditches and
keeping them well maintained, and good management of water within each
project (as well as among projects).
Donor agencies interested in assisting small farmers would be well ad-
vised to concentrate their efforts on intensifying irrigated agriculture.
This intensification must recognize the difference between irrigation systems
in which each farmer has an individual source of water--a pump from a river,
a pump from groundwater, or a diversion direct from a river--and those
systems where several farmers are linked to a common source. In either
system they will usually share a common drainage network. In the following
discussion this difference among systems must be kept in mind.
Another distinction which merits discussion is th'at between inundation
irrigation prevalent in lowland rice culture, and the intermittent irriga-
tion of wheat, cotton, and corn. While the specifics of any particular
project would need to-take these differences into account, the general dis-
cussion here will apply rather equally to both types of irrigation. More
detailed suggestions will require development on a case-by-case basis. We
now turn to a discussion of how small farmers might be helped.
B.  INTENSIFYING IRRIGATED AGRICULTURE
We will disaggregate irrigated agriculture into six components:
1. the farmer
2. the water transportation and control network
3. the agricultural.information system
4. the irrigation information system
5. the agricultural infrastructure
6. the irrigation infrastructure


VII-10
The Farmer
When undertaking projects or programs to improve irrigation the most
important element is the ultimate user of the water. In the construction
of new projects it is essential that all farmers agree--in general terms--
to a."constitution" wherein principles are adopted regarding water turns,
maintenance schedules and responsibilities, fee payments, and the like.
It is not essential that every detail of the water allocation system be
decided before the project becomes operable. But it is important that the
general issues be resolved prior to any farmer receiving water.
In the identification stage of the Project Cycle it will become
necessary to insist upon this sort of process as a normal part of the   !
general feasibility assessment. Also at this ttabe it will become necessary
to pay special attention to the differences among farmers on a proposed
system. Engineering studies have been optimistic in terms of the service
area for a project and this distorts the true aggregate production from
a system. Water deliveries are never quite what we hope they will be, and
the benefit-cost studies of proposed projects must be carried out more
critically than they have in the past. In this regard-, farm budget studies--which
form the foundation of a project's benefit-cost analysis--should be con-
ducted with an eye to a farmer's location within the system. It is unreason-
ab.le'to assume that tail-end farmers will have the same yields as those
near the head of a system. It would be preferable to design sample farm
budgets for three reaches of the watercourse: (1) head; (2) middle; (3) tail.
With different assumptions about water losses along the system, and about
the degree of water control, it will be possible to derive a more reasonable
set of expectations regarding project performance.


. VII-11
While the undertaking of entirely new projects is not expected to
comprise a very important part of future irrigation improvement efforts
for small farmers, the rehabilitation of existing projects will be signif-
icant. Consistent with the report of the Trilateral Commission, I consider
the enhancement of existing irrigation systems to be the most rational
policy for the next 5-10 years. This rehabilitation will have some con-
struction activity in terms of adding laterals and farm ditches. It will
also have some capital restoration components. Finally it will involve
careful work with groups of farmers and with the existing irrigation and
agriculture bureaucracies. More will be said on this below. But it should
be recognized that the same steps which are followed in the Project Cycle
are relevant to the rehabilitation of existing projects. We still must
pay attention to the problems of identifying the best candidates for
improvement; we still must offer special assistance in the preparation of
project plans. Countries will need special help in terms of identifying
objectives for'their irrigation program, and in assessing the role that
irrigation can play in the overall agricultural picture.
In the project appraisal stage the technical, institutional, economic
and financial aspects will require careful assessment, just as if a new
irrigation project were being undertaken. Part of this appraisal would
concern the likely viability of water user's organizations; another part
would concern the adequacy of the existing irrigation bureaucracy.


VII-12
. The economic appraisal will require careful attention to elisure that
monetary benefits received by small and low-income farmers are given
proper weights. If countries are serious about benefiting small farmers
then the,re is no reason why benefits received by rich farmers must be
weighted the same as benefits received by the small farmer. The benefit-
cost literature reflects this differential shadow pricing of benefits and
it might be employed in benefit-cost studies of projects where small
farmers are important target beneficiaries.
At the negotiation btage the Bank retains some ability to
influence the nature of proposed projects. The loan documents can reflect
the specifics of the project with respect to water management organiza-
tions in the government, the staffing of these organizations, the periodic
training to be received by its staff, general salary guidelines of various
1
levels within the organizations, the number and placement of water
measurement devices, the number and location of water control structures,
and rotation schedules. Also, maintenance and repair criteria can be
negotiated and become part of the legal documents of the project.
It may turn out that the host countries will not require much
persuasion to take steps which will enhance the operating efficiency of
the proposed projects. If this becomes a problem it can be minimized by
the Bank's willingness to cost share.
A frequent problem is that ditch tenders are so low paid that they are
easy targets for those wishing to influence water allocation by offering
side-payments.


VII-13
The Transportation and Control Network
It is the existence of this network which links farmers together
and thus introduces the physical interdependence outlined in Chapter II.
We may depict the basic options in water transportation networks by
making reference to Figure VII-4. The source can be either a diversion
from a river, a pump set which draws water from a river, or a pump which
draws under-ground water. The source can then go directly to either a
canal, a lateral, a ditch, or a farm. Water for irrigation can then move
either from farm to farm (as in some Asian rice culture) or travel along
a ditch to a number of farms and then become drain water.
What we have in the  Figure is the physical structure and its control
mechanisms, but not the way in which water is managed in that network.
It is strictly the engineering works of an irrigation project. This
would also include the engineering works at the source of the water, whether
a diversion from a river or a pump.
In the existing systems it may be possible to improve this network in
both an engineering sense, and in a socioeconomic sense. The former requires
little elaboration. As for the latter, we might be able to construct new
laterels in areas where water now moves from plot to plot, and we might
constrUct better (or more frequent) control structures. If these are
carefully placed so as to overcome current inequities in water allocation
then the interests of the small farmers will be enhanced. 2 Also if there
2 See Easter [1977].


由


VII-15
exists factionalism within a system the judicious placement of laterals,
ditches, and gates may be instrumental in ameliorating the problem.
In identifying sites for new irrigation projects special attention*
should be given to the compatibility of engineering "imperatives" with
socioeconomic considerations. While water will, indeed, not flow uphill,
there is no reason why canals, laterals, and ditches cannot be located
such that the currently poor farmers are not further disadvantaged once
the sy stem becomes operable. Sites in which the only feasible transpor-
tation network is one which places the already large and advantaged
farmers near the head of laterals and ditches should be avoided.
A second factor should be to match up--as closely as possible--
the basic physical subdivisions with existing sociological subdivisions.
To have a lateral serve two "rival" villages is asking for problems. If
it is not possible to engineer the system in such a way as to avoid
obvious conflicts then it will become increasingly important that insti-
tutional mechanisms be started at the same time that the initial surveying
is started. There should be meetings between farmers and the representa-
tives from the water management organization. There should be initial
efforts to establish farmer groups for managing the water once it arriv s;
it is essential that this institutional infrastructure exist before the
wate .flow; and 'behavioral patterns become established. That is, when
the system is new all farmers will have positive expectations about how
the system will function,.and some general notion about how they will


VII-16
personally benefit. If, during the first season, actuality departs
significantly from these expectations then trouble is virtually assured.
Current benefit-cost procedures are insufficient in both scope and
content to assess properly this mutual importance of engineering and
sociological considerations. What will appear as a "good" project in
our conventional view may indeed be a serious failure. This fact will
require an input on project identification efforts of anthropologists or
social psychologists with experience in the developing countries.
The Agricultural Information System
This includes the full array of information about agriculture which
is provided to the farmer, plus the information flow from the farmer back
through the system. This system includes the accumulated knowledge
about specific agricultural enterprises in a country, and the network
whereby that information is transmitted to the users. We generally refer
to parts of the system as "extension" but the notion employed here
comprehends more than the extension service in a country.
A program to enhance small farmer irrigation should pay special
attention to the informational needs of this group. Programs in which
water and crop practices are integrated would be a necessity.


VII- 17
The Irrigation Information System
-This represents a special class of information flows specifically
about water in agriculture. It is more narrow than the previous category,
and much more specific. Included here would be information about crop
response to alternative timing and quantities of water at certain stages
of plant growth, information concerning irrigation procedures to be used
in conjunction with fertilizers and herbicides, plus information about
good water management in general.
This information system--in conjunction with the agricultural infor-
mation system--comprises the totality of information exchange in'an
irrigated agriculture setting.
The Agricultural Infrastructure
This represents the full array of agricultural services beyond the
farm, but not including the formal information system. Here we have the
suppliers of agricultural inputs, the marketing channels, banks and
informal money lenders, and other commercial enterprises that are linked
to farmers. The work on farmer cooperatives, on agricultural credit, on
seed certification programs, and on improved farm-market roads would be
categorized here.
The .Irriga-tion -Infrastructure
The final element is the system of management and control over water
beyond the farm level; we might think of this component as the irrigation
"bureaucracy." In most countries this component represents an engineering
organization whose primary functions are planning, designing, constructing,
and maintaining canals and laterals; the dams are often constructed by an


VII-18
energy agency, with the irrigation bureaucracy getting some watdr for
irrigation. The irrigation infrastructure may also include a nominal
staff for the "management" of irrigation water, though their functions
only rarely exte!,d down to the level of a ditch in Fiqure VII-4.
Improving .the irrigation infrastructure can be discussed in both quantitative
and qualitative terms. The quantitative dimension pertains to the simple
lack of personnel in the developing countries to manage the shared input
water. In part this is the result of a frequent pattern of the failure
for any one agency to have an interest in--and control over--irrigation
water. Or, when that agency mandate does exist we find a lack of people
to carry out the required tasks.
A program in improved irrigation would, therefore, need to commence
by ascertaining whether the problem is one of a bureaucratic vacuum, or
a problem of insufficient personnel.
Bureaucratic Vacuum. The types of irrigation problems discussed in
Chapter VI for Mexico and the Philippines represent two aspects of a
bureaucratic vacuum. In Mexico we find a local irrigation system in which
ownership of canals/ditches is varied, while in the Philippines and in many
other places we find national jurisdiction over construction of canals and
laterals, but little else.
An assistan.ce model for situations such as we find in San Juan would
pFobabFly need to focus on the use of indigenous rural development/religious/
community organizations to provide some technical assistance to the
small number of local irrigators and their elected or appointed water
bureaucracy.


VII-19
For larger (national) systems the model would probably call for the
creation of a water management division which would contain sub-divisions
concerned with: (1) operation, which is essentially water control and
distribution; (2) maintenance, which is the routine work on control
structures, and ditch cleaning; and (3) repair, which is the more signifi-
cant overhaul of the physical facilities.
This division would need to be entirely separate from current
divisions which are occupied with water planning, construction, and the
like. There are even good arguments for why this type of a division
ought not to be located in the "construction" agency at all, but-instead
located in the ministry of agriculture. This makes some considerable
sense in certain countries, and this issue would need to be resolved on
a country basis. There are no-universal principles to guide us here.
This group of individuals works on the
transportation network--that is, manages it--to do one thing, and that is
to move water from the source to the farmers as efficiently, as predictably,
and as equitably as possible. It is quite possible to imagine some
bureaucratic performance criteria by which to evaluate the quality of the
job being done by such a division. The information system
is.then capable of operating as a link between farmers and
thbse 'Who manage water. Farmers indicate desires about water receipts,
the water managers indicate likely availability and timing; farmers
communicate conveyance system maintenance problems, the managers indicate
likely maintenance dates and work-detail needs and the farmers indicate


VII-20
availability for work parties, etc. The mere existence of the management
divition can thus be seen to create a communication system about irrigation.
In ,making the above suggestions for an irrigation management division
(either within-the existing irrigation agency or within the ministry of
agriculture) the temptation is almost irresistable to offer a flow-chart
of organizational possibilities with the conventional lines and boxes.
The temptation will be denied, but it will be helpful to consider some
general concepts.
It is essential that each segment of an irrigation system/project
on which are located 10-20 farmers have a water management person whose
primary (if not sole) responsibility during the irrigation season is
water control among those farmers. Depending upon the control structures,
the extent to which those 10-20 farmers are spread out, and the degree
of inherent cooperation among the farmers, it may be possible for this
water management individual to work two such subdivisions of a project
system.. Thus, an irrigation project containing 80-100 farmers ought to
have 3-4 such individuals depending upon local conditions. These
individuals should report to a "canal master" or some such.ana.ogue. If
one canal serves 3-4 "clusters" of farmers served by laterals then this
cana1 master would be in charge of perhaps 10-15 water managers. The
canal masters would report to supervisors in several possible patterns
depending upon the extensiveness and complexity of the project.
What must be recognized is the need for visible water managers
and canal masters on a daily basis during the irrigation season.


VII-21
Insufficient Personnel. The second quantitative aspect arises when
a water management agency exists but it is inadequately staffed. The pre-
vious discussion can be recalled for some indication in minimal staffing
needs. 'The usual situation may often find a weak canal master,: and one or
two water managers (or ditch tenders) for a very large number of farmers.
Turning to the matter of quality, the basic issue is that there is
a virtual dearth of qualified water managers in the developing countries.
There can be no greater priority than the development of a qualified
cadre of public servants to staff existing (or new) positions as described
above. This training would need to. be in agricultural areas, as well as
in engineering concepts pertaining to water movement and losses within a
system. It is difficult to define the ideal training program without
making specific reference to each country's (or region's) situation;
this would require detailed,diagnosi's of each project/system by a small
team of experts.
By way of.summary, the six components of the irrigation system can be
depicted as in Figure VII5. Nonirrigated agriculture is depicted as a
sphere involving the farmer, the agricultural infrastructure, and the agri-
cultural information system. This sphere is more developed in some
countries than in others, and the levels of intensification will vary as
well'.. The bulk of conventional agriculture assistance operates within--
or upon--this sphere.


VII- 22
INFRASTRUCTURE
TRANSPORTATION       --IRRIGATION INFORMATION
SYSTEM
AGRICULTURAL -          - ----  -FARMERS
INFRASTURCTURE
AGRICULTURAL
INFORMATION
FIGURE VII-5.
r*


VII-23
But irrigated agriculture interposes yet another sphere on the
farmer and that is one containing both the transportation network, and
the irrigation bureaucracy. As before, information about irrigation is'
important to this sphere. However, in this instance the link between a
water management organization and the farmer (via the transportation
network) is weak, if it exists at all. Having introduced this second
sphere, we can state that in most countries the intensification of agricul-
ture continues to operate within the first sphere, often ignoring the
irrigation sphere. Recall from Chapter VI the discussion concerning
Pakistani farmers and their knowledge about irrigation, and that 70 percent
of the farmers in a sample of 387 received no word in advance of canal
closings by the irrigation bureaucracy.
C. CONCLUSIONS
Irrigation is a technological innovation of unappreciated complexity.
The common inpression-seems to be that one only provides the transportation
systo.i and the.rest -- meaning watdr management -- will automatically
follow.' We know, of course, that this is not the case. When technology
precedes the institutional arrangements which define who controls the new
income streams made possible by that technology then those already in a.
position to enhance their economic and political advantage will move quickly
to d6.so. Such is the history of irrigated agriculture in the developing
countries.
In marry instances great wealth and considerable power are not necessary
preconditions for gaining at the expense of others -- all that is required
is that one be fortunate enough to have a farm at the head of an irrigation
system. With only this fortuitous accident the conditions are set for.a
significant income gain vis-a-vis the more distant irrigators. Of course,


VII-24
on some systems being at the head of the system is due to more than mere
luck; it is not unheard of for influential farmers to have some role in
the location of canals and laterals. But such overt influence is not
necessary.
Irrigation creates new income streams and the dominant fact of irriga-
tion in the developing countries is that those income streams accrue to
those fortunate enough to have some control over the application of water.
Programs to enhance the economic position of small irrigation farmers
can consist either of the construction of new irrigation projects to serve
small farmers, or the rehabilitation of existing projects. The hypothesis
here -- and one that is supported by the recent Trilateral Commission report --
is that the most cost-effective policy would focus on improving water alloca-
tion on existing irrigation projects. This emphasis is here referred to
as intensification. Donor agencies have a special role in this process.
Through the provision of both advice and loan funds there is a unique
opportunity to*encourage countries to devote more attention to intensifica-
tion. Financial incentives can be offered, but may prove to be unnecessary.
The obvious pay off should be sufficient to elicit the required cooperation
of recipient governments. But the demonstration of these pay offs will.
require careful attention. The network of international research centers
under-the Consultative Group provides one obvious mechanism. Many of these
have -programs concerned with improved cropping practices, of which irriga-
tion is an integral part.' There is now talk of the creation of a center
concerned exclusively with water management. It is not clear that the
institutional problems of water management lend themselves to the center
approach in the same way that genetics and cropping systems do. Yet the
center should be given serious consideration.


VII-25
And yet, the major impetus for improved water management is surely
to come from the donor agencies. Recipient governments have a long history
of cooperation with such agencies, and this should enhance the potential
for success.
The concepts developed hhere will hold for the vast majority of irriga-
tion systems in the developing countries. What will require tailoring is
the specific remedial action on a country-by-country -- if not project-by-
project -- basis. The components of irrigation systems presented here would
seem to offer a logical way in which to structure both diagnosis and treat-
ment. But it must be emphasized that project-specific and country-specific
programs must be developed. No sweeping generalizations will do when it
comes to rectifying years of mal-allocation of water.
If the diagnosis is undertaken with the conce.ptual model offered here,
then both economic efficiency and distributional justice can be the policy
objectives. One does not need to appeal to abstract notions of efficient
water use in an agronomic or an engineering sense. While these are import-
ant, policy makers are more inclined to listen to arguments which emphasize
the private and social costs of the current mis-management. Few of them
will want to improve the efficiency of water use unless we'convince them
that water is truly scarce. In the jargon of linear programming, they must
be convinced th'at an extra cubic-second of water has an extremely high
shadow price -- both privately and socially. When we can show them that
reallocating water is no.t necessarily a zero-sum game then that high shadow
price becomes even more irresistable.
This should be the primary mission of donor agencies. As that work
progresses, it will become time to devote attention to remedial programs.


VII-26
These efforts will require as much agency attention as now goes into the
planning and evaluation of new projects. But the payoff to the country and
to the agencies is almost certainly greater.
In conclusion, small farmer irrigation combines all of the elements
currently fashionable in agricultural development. It is concerned with
enhancing a country's ability to keep food production ahead of population
growth. It is concerned with creating a more dynamic and innovative sub-
sistence sector. It is concerned with distributional justice. And it is
concerned with spending donor and host-country resources in a manner that
is likely to yield high returns as.compared to other program options. The
era of massive capital infusion for dams, canals, and control gates has
yielded to an era of program performance and accountability. We are at
the threshold. The urban masses of the developing*countries -- not to
mention the rural landless -- demand ever increasing quantities of basic
foods7. With rice, wheat and corn comprising a good share of the irrigated
crops in the developing countries there is little doubt that programs to
enhance production will be popular with such governments. The subsistence
sector comprises an important opportunity to meet that nutritional imperative.
The intensification of agriculture at the small farm level through better
water management is not only a cost-effective policy, it is an ethically
compe1ling one as well.


LITERATURE CITED
Abel, Martin. "Irrigation Systems in Taiwan: Management of a-Decen-
tralized Public Enterprise," in: Water Resources Problems in
Developing Countries, Ed. by K. William Easter and Lee R. Martin,
Minneapolis: University of Minnesota, Economic Developmet Center,
Bulletin No. 3, July 1977, pp. 29-45.
Barnett, Tony. "Why are Bureaucrats Slow Adopters? The Case of Water
Management in the Gezira Scheme," Sociologia Ruralis 19:60-70,
No. 1, 1979.
Berry, R. Albert and William R. Cline. Agrarian Structure and Productivity
in Developing Countries, Baltimore: Johns Hopkins, 1979.
Boserup, Easter. The Conditions of Agricultural Growth, Chicago: Aldine
Publishing Co., 1965.
Bromley, Daniel W., Donald C. Taylor and Donald E. Parker. The Economics
of Water Reform: institutional Design for Improved Water Management
in the LDC's, Madison: University of Wisconsin, Center for Resource
Policy Studies, Working Paper No. 8, October 1977.
Bromley, Daniel W., Donald Taylor, and Donald E. Parker. Economic
Development and Cultural Change 28, January 1980.
Colombo, Umberto, D. Gale Johnson, and Toshio Shishido, Reducing Malnutrition
in Developing Countries: Increasing Rice Production in South and
Southeast Asia, New York: The Trilateral Commission, 1978.
Day, Richard H., and Inderjit Singh. Economic Development as an Adaptive
Process: The Green Revolution in the Punjab, Cambridge: Cambridge
University Press, 1977.
Easter, K. William, "Improving Village Irrigation Systems: An Example from
India," Land Economics 53:56-66, February 1977.
Geertz, Clifford. Agricultural Involution: The Processes of Ecological
Change in Indonesia, Berkeley: University of California Press, 1963.
Grabowski, Richard. "The Implications of an Induced-Innovation Model,"
Economic Development and Cultural Change 27:723-734, July 1979.
Hunt, Eva and Robert C. Hunt. "Irrigation Conflict, and Politics: A
Mexican Case," in Origins of the State, Ed. By Ronald Cohen and
Elman R. Service, Philadelphia: Institute for the Study of Human
Issues, 1978, pp. 69-123.
Levine, Gilbert. "Management Components in Irrigation System Design and
Operation," Ac!ricultural Administration, 4:37-48, 1977.
Lowdermilk, Max K., Alan C. Early, and David M. Freeman, Farm Irrigation
Constraints and Farmers' Responses: Comprehensive Field Survey in
Pakistan, Fort Collins, Colorado: Water Management Research Project,
Technical Report No. 48-A, Volume 1, September 1978.
Myrdal, Gugnar.. Asian Drama, New York: Vintage Books, 1972.
Newbery, D.M.G. "Tenurial Obstacles to Innovation," The Journal of
Development Studies, 11:263-77, July 1975.
Parker, Donald E., 1979. Irrigation Project Performance and Water Distri-
bution in the Less Developed Countries, Unpublished Ph.D. Dissertation,
Madison, Wisconsin, .University of Wisconsin.
Schultz, Theodore W. Transforming Traditional Agriculture, New Haven:
Yale University Press, 1964.
Wilkinson, Richard G. Poverty and Progress: An Ecological Perspective
on Economic Development, New York: Praeger, 1973.
Wortman, Sterling and Ralph Cummings, Jr. To Feed this World, Baltimore:
Johns Hopkins University Press, 1978.


ADDITIONAL REFERENCES
Benor, Daniel and James Q. Harrison, 1977. Agricultural Extension,
Washington, D.C.: World Bank, 1977.
Bennett, John W-., 1976. "Comment on Canal Irrigation and Local Social
Organization," Current Anthropology, 17:398-400, September.
Birkelbach, Aubrey W., Jr., 1973 "The Subak Association," Indonesia, 16:
153-69, October.
Bottrall, A.F., 1976. "Some Propositions About Irrigation Project and
System Management," paper presented to ODI "Workshop on Choices in
Irri,gation Management," University of Kent, Canterbury, U.K.
September 27 - October 1, 1976.
Chambers, Robert, 1974. "The Organization and Operation of Irrigation:
An Analysis of Evidence from South India and Sri Lanka," paper
prepared for a seminar on Agrarian Change in Rice Growing Areas of
Tamil Nadu and Sri Lanka, St. John's College, Cambridge, U.K.,
December 9-16, 1974.
Coward, E. Walter, Jr., 1977. "Irrigation Management Alternatives:
Themes from Indigenous Irrigation Systems," Agricultural Administra-
tion 4:223-37, 1977.
Cruz, Helen A., 1974. Conflict Among Farmers in an Irrigation System in
Iloilo Province, the Philippines, unpublished Ph.D. dissertation,
Ithaca: Cornell University.   .
Duewel, John, 1976. "A Structural Exchange Analysis of the Ulu-Ulu
System of Rural Irrigation in Java," (draft paper under consideration
for publishing by the Rural Development Committee), Cornell University,
Ithaca, New York.
Eyre, John D., 1955. "Water Controls in a Japanese Irrigation System,"
The Geographical Review, 45:197-216.
Freeman, David M. and Max K. Lowdermilk, 1976. "Community and Irrigation
in the Pakistan Punjab -- Physical and Sociological Dimensions of
the Water Management Problem," Paper presented at the Association for
Asian Studies Annual Meeting, Toronto, Canada, March 20., 1976.
Gillespie, Victor A., 1975. Farmer Irrigation Associations and Farmer
Cooperation. (Papers of the East-.est Food Institute No. 3),
Honolulu: East-West Center, June 1975.
Glick, Thomas F.*, 1976. "Comment on Canal Irrigation and Local Social
Organization," Current Anthropology, XVII (3):400-01, September.
Gustafson, W.E. and R.B. Reidinger, 1971. "Delivery of Canal Water in
North India and West Pakistan," Economic and Political Weekly
(Bombay), 6:A-157 to A-162.
Harriss, John C., 1974. "Problems of Water Management in Relation to
Social Organization in Hambantota District," paper preared for a
seminar on Agrarian Change in Rice Growing Areas of Tamil Nadu and
Sri Lanka, St. John's College, Cambridge, U.K., December 9-16, 1974.


Harriss, John C., 1976. "Problems in Water Management Policy and Objectives:
The Role of Local Institutions and Their Relationship with Bureaucracy,"
paper presented to ODI Workshop on Choices in Irrigation Management,
University of Kent, Canterbury, U.K., September 27 - October 1, 1976.
*        Hayami, Yujiro and Masao Kikuchi, 1978. "Investment Inducements to Public
Infrastructure: Irrigation in the Philippines." The Review of
Economics and Statistics 60:70-77, February 1978.
Hodgdon, Linwood L., 1974. "The Adoption of New Agricultural Inputs and
Practices by Indian Farmers," in Small Farm Agricultural Development
Problems, ed. by Huntly Biggs and R.L. Tinnermeier, Fort Collins,
Colorado State University, pp. 69-95.
Hunt, Robert C. and Eva Hunt, 1976. "Canal Irrigation and Local Social
Organization," Current Anthropology, 17:389-98, September.
Hunt, Robert *C. and Eva Hunt, 1976. "Canal Irrigation and Local Social
Organization: Reply,'` Current Anthropology, 17:405-11.
Hutapea, R. et.al., 1976. "The Organization of Farm Level Irrigation
Services in Indonesia," paper presented at ADC-SEARCA-IRRI Water
Management Seminar, Los Banos, Philippines, June 22-25, 1976.
Jay, Robert R., 1969. Javanese Villagers: Social Relations in Rural
Modjokuto, Cambridge, Massachusetts, the MIT Press.
Kumar, P., 1974. "An Economic Study of Water Management Program in
Sambalpur District (Orissa), Indian Journal of Agricultural Econo-
mics (Bombay), 29:43-51.
Lazaro, Rogelio, Donald C. Taylor, and Thomas H. Wi*ckham, 1977.
"Irrigation Systems in Southeast Asia: Policy and Management
Issues," Teaching and Research 'Forum, No. 6, The Agricultural
Development Council, Inc., Singapore University Press, May 1977.
Lees, Susan H., 1974.. "Hydraulic Development as a Process of Response,"
Human Ecology, 2:159-175.
Levine, Gil, L.T. Chin, and S.M. Miranda, 1976. "Requirements for the
SuccessfuT Introduction and Management of Rotational Irrigation,"
Agricultural Water Management 1:41-56, 1976.
Lewis, Henry T., 1971.  "Irrigation Societies: Zangjeras," in Ilocono
Rice Farmers: A Comparative Study of Two Philippine Barrios,
Honolulu, University of Hawaii Press (Chapter 12).
Lowdermilk, Max K., 1972. Diffusion of Dwarf Wheat Product-ion Technology
in Pakistan's Pubjab, unpublished Ph.D. dissertation, Cornell
University.
Lowdermilk, Max K., Wayne Clyma, and Alan C. Early, 1975. Physical and
Socio-Economic Dynamics of a Watercourse in Pakistan's Punjab:
System Constraints and Farmers' Responses (Water Management Technical
Report No. 42), Fort Collins, Colorado State University, December.
Miglani, S.S., et al., 1975. "Income Distribution in Relation to Farm
Size and Irrigation," Economic Affairs (Calcutta), 20:233-240.
Millon, Rene, Clara Hall, and May Diaz, 1962. "Conflict in the Modern
Teotihuacan Irrigation System," Comparative Studies of Society and
History, 9:494-521.


Mitchell, William P., 1973. "The Hydraulic Process: A Reappraisal,"
Current Anthropology, 14:532-34, December.
Mi,rz?, A.H., 1975. A Study of Village Organizational Factors Affecting
Water Management Decision-Making in Pakistan (Water Management
Technical Report No. 34), Fort Collins, Colorado State University,
March.
Moorti, T.V. and J.W. Mollor, 1973. "A Comparative Study of CostF and
Benefits 6f Irrigation from State and Private Tubewells in Utear
Pradesh," Indian Journal of Agricultural Economics, 28:181-189.
Orenstein, Henry, 19~65. "Notes on the Ecology of Irrigated Agriculture in
Contemporary Peasant Societies," American Anthropologist, 67:1529-32,
December.
Parks, Loren, 1976. Estimation of Water Production Functions and Farm
Demand for Irrigation Water, with Analysis of Alternatives for
Increasing the Economic Returns to Water on Chilean Farms, unpublished
Ph.D. dissertation, University of California at Davis.
Pasandaran, Effendi and Donald C. Taylor, 1976. "The Management of
Irrigation Systems in the Pekalen Sampean Irrigation Project, East
Java, Indonesia," (Research Note No: 01/76/RN), Jakarta: Agro-
Economic Survey, March 22, 1976.
Pasternak, Burton, 1968. "Social Consequences of Equalizing Irrigation
Access," Human Organization, 27:332-42, Winter 1968.
Pasternak, Burton, 1972. "The Sociology of Irrigation: Two Taiwanese
Villages," in Economic Organization in Chinese,Society. Edited by
W.E. Willmott, Stanford, Stanford University Press.
Radosevich, George E., 1975.' Water User*Organizations for Improving
Irrigated Agriculture: Application to Pakistan. (Water Management
Jechnical Report .No. 44), Fort Collins, Colorado State University,
December 1975.
Reidinger, Richard B., 1971. Canal Irrigation and Institutions in North
India: Microstudy and Evaluation, unpublished Ph.D. dissertation,
Chapel Hill, North Carolina, Duke University.
Reidinger, Richard B., 1974. "Institutional Rationing of Canal Water in
Northern India: Conflice Between Traditional Patterns and Modern
Needs," Economic Development and Cultural Change, 23:79-104, October.
RISS, 1975. "Research on Water Management at the Farm Level: An
Indonesian Case Study," paper presented by Research Institute of
S6cial Sciences, Satya Wacana University, Salatiga, Indonesia at an
IDRC Workshop on Technical and Social Progress in Rural Development,
Penang, Malhysia, December 16-19, 1975.
Small,'Leslie E,, 1973. Economic Evaluation of Water Control in the Greater
Chao Phya Project of Thailand (Cornell International Agricultural
Bulletin 27), Ithaca, New York, Cornell University, May 1973.
Steward, Daviel L., 1976.. Aspects of Chilean Water Law in Action: A
Case Study, unpublished Ph.D. dissertation, Madison, University of
Wisconsin.
Tantigate, Kanaengnid, 1976.  "The Conflict Between Rice anid Sugar Cane
Farmers Over Irrigation Water: A Case Study in Thailand," paper
presented at ADC-SEARCA-IRRI Water Management Seminar, Los Banos,
Philippines, June 22-25, 1976.


Taylor, Donald C., 1976. "Agricultural Development through Group Action
to Improve the Distribution of Water in Asian Gravity-Flow Irriga-
.tion Systems" (Teaching and Research Forum No. 1). The Agricultural
Development Council, Inc., Singapore: Singapore University Press,'
June 1976.
ThiesenHusen, William C., 1971. "Technological Change and Income
Distribution in Latin American" paper presented at 0AS Sixth Inter-
American Conference on Agriculture, Lina, Peru, May 27-June 2, 1971
(also LTC Paper No. 78).
Thorner, D., 1962. '"The Weak and the Strong on the Sarda Canal," in
Land and Labour in India, Bombay, Asia Publishing House.
Valera, A., D. Tabbla, R. Reyes, and R. Barker, 1975. "The Effect of
Improved Water Con:rol in Irrigation Canals on Crop Production."
(Paper No. 75-12), .os Ba 'os, Philippines: Agricultural Economics
Department, Interti'tional Rice Research Institute.
Vandermeer, Canute, 1968. "Changing Water Control in a Taiwanese Rice
Field Irrigation System," Annals of the Association of American
Geographers, 58:720-47.
Vandermeer, Canute, 1971. "Water Thievery in a Rice Irrigation System
in Taiwan," Annals of the Association of American Geographers,
-61:156-79, March.
VanderVelde, E.J., 1971. The Distribution of Irrigation Benefits: A
Study in Haryana, India, unpublished Ph.D. dissertation, Ann Arbor,
University of Michigan.
Wade, Robert, 1975a. "Administration and the Distribution of Irrigation
Benefits," Economic and Political Weekly (Bombay), 10:1743-47,
November 1.
Wade,.Robert, 1975b. -"Water to the Fields: Institutional Innovations in
India's Command Area Development Programme," South Asian Review,
8:301-21, .July/0ctober.
Wickham, Gekee Y., 1970. Sociological Aspects of Irrigation, unpublished
Master's dissertation, Los Banos, Philippines, College of Agriculture,
University of the Philippines, June 1970.
Wickham, G. and T. Wickham, 1974. "An Evaluation of Two Alternatives of
Water-Sharing Among Farmers," a compilation of unpublished papers,
Los Banos, Philippines: International Rice Research Institute,
August 1974.
Wickham, T. and A. Valera, 1976. "Practices and Accountability for Better
Water.Management," paper presented to ADC-SEARCA-IRRI Water Manage-
ment Seminar, Los Banos, Philippines, June 22-25, 1976.
Wittf6gel, Karl, 1957. Oriental Despotism: A Comparative Study of Total
Power, New Haven: Yale University Press.