EPTD DISCUSSION PAPER NO. 4 
 
 
 

 
Environment and Production Technology Division  

International Food Policy Research Institute 
1200 Seventeenth Street, N.W. 

Washington, D.C.  20036-3006 U.S.A. 
 
 
 
 September 1994 
 
 
 
 
 
 
 
 
 
 

EPTD Discussion Papers contain preliminary material and research results, and 
are circulated prior to a full peer review in order to stimulate discussion and critical 
comment.  It is expected that most Discussion Papers will eventually be published in 
some other form, and that their content may also be revised. 

  

 
WATER MARKETS IN PAKISTAN: PARTICIPATION AND 

PRODUCTIVITY 
 

Ruth Meinzen-Dick and Martha Sullins 



  
 
 

ACKNOWLEDGMENTS 
 

Research completed under USAID to Pakistan Grant Number 391-0492-G-00-
1791-00 for the Ministry of Food and Agriculture, Government of Pakistan. 
 

Special thanks are due to Naeem Sarwar and Khalid Saifullah for collecting 
additional data for this study, and to Manzoor Gill, Jafar Raza, Behjat Hojjati, and 
Sumiter Broca for assistance in analysis.  Sohail Malik and Zakir Rana provided valuable 
comments and insight, in addition to much assistance in arranging for the data collection.  
Pierre Strosser collaborated extensively in identifying the important issues to be 
addressed with regard to water markets in Pakistan, and shared detailed field data used 
for comparison in this paper.  Dr. Bashir Ahmed of Agricultural University, Faisalabad, 
also provided useful comments and access to National Input Output Survey data used for 
comparative purposes.  Steve Vosti, Peter Hazell, Mark Svendsen, and Mark Rosegrant 
commented on a draft of the paper.  Responsibility for any errors rests with the authors. 



  
 
 

 
 

ABSTRACT 
 
 

Water markets provide one of the most promising institutional mechanisms for 
increasing access to irrigation from groundwater, particularly for tenants and small 
farmers.  While water markets are found in all provinces of Pakistan, they are most 
prevalent in canal irrigated areas of Punjab and in NWFP.  This study reviews the 
emerging literature on water markets and uses farm-level survey data to examine the 
performance of groundwater markets, with particular emphasis on Faisalabad District in 
Punjab and Dir District in NWFP. 

Findings indicate that, while large landowners are more likely to own tubewells 
and pumps, smaller landowners and tenants are more likely to rely on purchases from 
other farmers' tubewells for access to groundwater.  The distance over which water can 
be transported provides a limitation to water market sales, but lined watercourses increase 
the distance over which tubewell water can be sold.  Contractual arrangements for water 
in the IFPRI study areas of Faisalabad and Dir districts include hourly charges, buyer 
providing the fuel plus a fee for wear and tear, and sharecropping for water.   
 
  While all types of irrigation--canal, purchased groundwater, and own tubewell 
water--are shown to increase yields of wheat, groundwater has a higher impact than canal 
water, and water from own tubewells, which provides farmers with the greatest degree of 
control, has a greater effect on yields than purchased groundwater.  Unreliability of 
access to purchased tubewell water was a problem for over half of water buyers in the 
study areas.  This analysis indicates that purchasers are more likely to have unreliable 
access to groundwater if they buy water from small-capacity, electric-powered tubewells, 
if they are young and own little or no land.   
 
 Policy measures to improve access to and reliability of groundwater through water 
markets include increasing the density of tubewells, especially by assisting small farmers 
to purchase tubewells; lining water delivery channels; and providing more reliable 
electrical power supply to rural areas.  Further research is needed on how water markets 
work in less favorable environments, such as those with salinity, waterlogging, or falling 
water tables, and to identify policy interventions that are appropriate under each set of 
circumstances. 



  
 
 

  
CONTENTS 

 
 
 
1.  Introduction.................................................................................................................... 1 
 
2.  Who Participates in Water Markets? ........................................................................... 11 
 
3.  Relationships Between Water Buyers and Sellers ....................................................... 27 
 
4.  Nature of Water Market Contracts............................................................................... 30 
 
5.  Agricultural Productivity Under Water Markets ......................................................... 35 
 
6.  Reliability of Irrigation Service ................................................................................... 45 
 
7.  Research and Policy Measures for Water Market Development ................................. 52 
 
References......................................................................................................................... 59 

 



 
 
 
 

WATER MARKETS IN PAKISTAN: PARICIPATION AND PRODUCTIVITY 
 

Ruth Meinzen-Dick1 and Martha Sullins***2 
 
 
 

 1.  INTRODUCTION 

Irrigation plays a key role in Pakistan's strategy for increasing agricultural 

productivity.  Surface irrigation has allowed the extension of cultivation into areas and 

seasons which lack sufficient rainfall for agriculture, and raised yields above what is 

possible under rainfed cultivation.  Groundwater irrigation is increasingly important in 

improving production, either alone or in conjunction with surface irrigation. 

Access to water in public irrigation systems (surface canals and public tubewells) 

is tied to ownership of land in the command area.  This land ownership entitles the farmer 

to a fixed turn of irrigation flow during a rotation cycle, to be used only on that land.  The 

rigidity of this system limits the productivity of surface irrigation and public tubewells, a 

limitation which is especially apparent in comparison with privately-managed 

groundwater irrigation, where farmers have more control of water timings (see Renfro 

and Sparling 1986). 

Access to privately-managed groundwater irrigation is dependent on investment 

in wells and pumping devices.  To the extent that large and wealthy farmers are most 

likely to own tubewells and small or poor farmers are unable to make the necessary 

                                                 
1 Research Fellow, Environment and Production Technology Division, International Food Policy Research 
Institute. 
 
2 Former Senior Research Assistant, Environment and Production Technology Division, International Food 
Policy Research Institute. 



  
 
 

2

investment, the latter may be excluded from the benefits of highly productive 

groundwater resources. On the other hand, widespread private ownership leads to over-

investment in wells and pumpsets, particularly where holdings are small or fragmented.  

Institutional arrangements are needed to spread access to groundwater to other farmers, 

and to increase agricultural productivity and equity of irrigation water resources. 

Water markets, in which farmers buy and sell irrigation water, provide one of the 

most promising institutional mechanisms for increasing access to irrigation with private 

groundwater, for providing vertical drainage, and for increasing the efficiency of water 

use in irrigation systems (see Rosegrant and Binswanger 1992).  While such markets are 

not formalized or officially recognized, the sale of water from private tubewells is a 

growing form of water allocation.  The sale and purchase of public canal water supplies, 

though legally prohibited under the Canal and Drainage Act, is another type of private 

water market transaction which takes place in Pakistan.  These are, however, much less 

common than tubewell water sales.1 

 This paper examines the nature and operation of groundwater markets in 

Pakistan.  It deals with the extent of water market development, who participates, the 

nature of transactions, the impact on productivity of irrigated agriculture, and the 

reliability of purchased irrigation services, with particular reference to Faisalabad District 

                                                 
     1  In the IFPRI sample, only 2 farmers reported purchasing canal water, compared to 74 purchasers and 
10 sellers of tubewell water.  A WAPDA (1990) study of water trading and sale practices in 100 water-
courses found 2-3 times as many farmers involved in sale of water from private tubewells as from public 
(canal and tubewell) sources.  The difference in amount of water sold was even more dramatic:   
 
Buyers          Sellers           
Number   Total hours Number  Total hours 
Public sources               30                161      21                      180 
Private tubewells          107  4,635   44  22,979 
 



  
 
 

3

in Punjab and Dir District in NWFP.  The final section suggests policy instruments for 

extending water markets and improving their performance. 

BACKGROUND 

Irrigation provides crucial water for agricultural production on over 80 percent of 

the gross cropped area in Pakistan.  Most of this irrigation comes through public canal 

systems, which, according to official reports, deliver surface water to approximately 70 

percent of the irrigated area.  However, throughout the 1970s and 1980s groundwater 

irrigation has been the most rapidly-growing source of irrigation: it now serves 

approximately 25 percent of the irrigated area, and provides over 36 percent of the 

irrigation water available at the farm gate (Pakistan: Ministry of Food, Agriculture and 

Cooperatives 1991; Rana and Shafiq-ur-Rehman n.d.).  Groundwater has become a 

crucial input, both as a sole source of irrigation and as a supplement to surface irrigation 

in canal irrigation commands. 

From the mid-1950s to 1980 public tubewells provided the primary  source of 

groundwater irrigation development in Pakistan, under the Salinity Control and 

Reclamation Program (SCARP).  According to the World Bank (1984:i): "The 

Government of Pakistan opted for public control of an extensive groundwater pumpage 

program based on the rationale that this arrangement would enable the Government to 

meet multiple groundwater objectives in an efficient and equitable manner."  These 

objectives included: 

1. providing vertical drainage to control waterlogging and salinity problems, 
especially in saline groundwater areas. 

2. increasing cropping intensities and agricultural production. 



  
 
 

4

3. capitalizing on the economies of scale in pumping technology.  By providing 
water to a larger area than that controlled by single farmers, public tubewells 
could use larger pumps which are, in theory at least, more efficient. 

4. reducing inequity in access to groundwater, by serving farmers with all sizes of 
holdings, regardless of financial resources for investment. 

 

In practice, institutional problems as well as technical difficulties resulted in 

disappointing performance of public tubewells.  Rising operation and maintenance 

expenses for public tubewells (which consumed 60 percent more than the entire national 

budget for canal operation and maintenance in 1983/84; see Aklilu and Hussain 1992:29), 

in conjunction with the poor performance of public tubewells in terms of timeliness and 

reliability of irrigation supplies, led the government to devolve responsibility for 

groundwater irrigation development from the public to the private sector.  The first two 

policies adopted in Pakistan's Revised Action Plan for Irrigated Agriculture (WAPDA 

1982: 16) were: 

Recommendation 1: Future Development of Usable Groundwater should be 

entrusted to Private Sector, . . . 

Recommendation 2: Present SCARP Tubewells in the Usable Groundwater Areas 

should be phased out and Replaced by Private Tubewells. 

 

Pakistan's experience suggests that private tubewell development can fulfill the 

objective of providing adequate vertical drainage, at least in areas of fresh groundwater 

(Chaudhry and Young 1990).  By 1982, approximately 80 percent of total groundwater 

was extracted through 186,000 private tubewells, compared to 20 percent provided by 



  
 
 

5

12,500 public tubewells (World Bank 1984).2  Increases in cropping intensities and 

agricultural productivity are greater under private tubewell ownership, primarily because 

private tubewells are more reliable (Johnson 1989:15; World Bank 1984:29).  The 

potential gains in technical efficiency derived from public tubewells using large-capacity 

pumps are offset, in practice, by frequent breakdowns and inadequate maintenance.  Most 

private tubewells provide greater water use efficiency by more closely matching water 

deliveries to crop needs, instead of adhering to the rigid schedules of public tubewell 

deliveries.  However, as the government closes public tubewells, the extent to which 

private groundwater development will benefit a large number of farmers and meet stated 

equity objectives remains an important question. 

There is increasing interest throughout much of South Asia in water markets as a 

means of increasing access to and use of groundwater for irrigation.3   Much of the 

literature on this subject is based on field studies conducted in India and Bangladesh.  

This paper reviews the issues identified in this literature, in order to set the stage for an 

empirical study of the performance of water markets in Pakistan. 

                                                 
     2 The large majority of tubewell development has taken place in Punjab.  In 1989/90, 88 percent of the 
private and 65 percent of the public tubewells of Pakistan were located in that province (Pakistan, Ministry 
of Food, Agriculture and Cooperatives 1991:194). 
     3 Water markets were a major topic in the World Bank Colloquium on How to Reach the Poor Through 
Groundwater, in Washington DC, April 12-14, 1989. 



  
 
 

6

Although water sales from private wells are a longstanding practice, these 

informal arrangements have only recently been recognized and empirically examined 

(Chambers, Saxena and Shah 1989:100-101).4  In most cases in South Asia, water sellers 

are farmers who sell surplus water after meeting the needs of their own fields.  In Gujarat 

State in India, where water markets are highly developed, individuals and even private 

water companies are investing in wells primarily to sell water to others.  Tubewell water 

sales have become a profitable enterprise for small farmers in Uttar Pradesh in India 

(Shankar 1992a) and even for the landless under the PROSHIKA program in Bangladesh 

(Wood and Palmer-Jones 1990). 

In Pakistan, water markets are reported in all provinces, but are most active in 

Punjab, where the greatest groundwater development has taken place.  By 1975, over 30 

percent of tubewell owners in Pakistan reported selling water, but the fraction of water 

sold was very small (World Bank 1984:35).  A study by Pakistan's Water and Power 

Development Authority in canal command areas of Punjab, Sindh, and NWFP found 

water sales in 43 of 100 watercourses (WAPDA 1990; see also Bajwa and Ahmad 1991).  

Sims' (1988) micro-level study in canal-irrigated villages in the Punjab found that 

approximately one-third of farmers purchased groundwater from neighbors, while 29 

percent owned tubewells. 

The potential advantages gained from groundwater market development lie in 

improving utilization of tubewell capacity, increasing access to irrigation water supplies 
                                                 
     4 There are numerous anecdotal reports and a growing number of studies of private water sales in Uttar 
Pradesh, Haryana, Punjab, Bihar, West Bengal, Orissa, and Andhra Pradesh in India (Shankar 1992a, b; 
Pant 1991a, b; Kolavalli and Atheeq 1990; Kolavalli, Naik and Kalro 1992; Kolavalli, Kalro and Asopa 
1989; Saleth 1991; Shah and Raju 1988).  Perhaps the most highly-developed private water markets in 
India are found in Gujarat state, where farmers invest in wells and underground networks of pipes for water 
deliveries (Shah 1985; Kolavalli and Chicoine 1989). 



  
 
 

7

(especially among farmers with small or fragmented holdings), and lowering water tables 

in areas of waterlogging (see Chambers, Saxena and Shah 1989; Chaudhry and Young 

1990).  By providing water to other farmers, tubewell owners can use a higher proportion 

of their well capacity than they would otherwise use on their own holdings.  The 

availability of hired tubewell services reduces the need for other farmers to install their 

own wells.  Because water markets increase the use of installed pumping capacity, they 

can improve the economic efficiency of private tubewell irrigation. 

Private well ownership tends to be concentrated among larger or wealthier 

farmers because of their ability to mobilize the necessary resources, including personal 

finances, credit, and government connections for electricity supplies (Chambers, Saxena 

and Shah 1989; Johnson 1989; Chaudhry 1990).  According to the World Bank 

(1984:35), 70 percent of tubewells are owned by farmers with over 12.5 acres, and half of 

all tubewells by farmers with over 25 acres, which seems "to point toward an adverse 

effect of private tubewells on income distribution within agriculture."  Water markets 

make it possible for those without wells to use groundwater for irrigation.5  The 

opportunity to sell groundwater can make it profitable for farmers to invest in wells even 

if their own holdings are too small to use the full pumping capacity (see Shankar 1992b).  

Dhawan concludes: 

the thrust of empirical research on groundwater markets, both in India and 
Bangladesh, has been to underscore the superiority of the institution of 
groundwater markets over the public tubewell system in catering to the 
irrigation needs of small and marginal farmers (1991:2). 

 

                                                 
     5 In examining inequality of irrigation distribution, Gill and Sampath (1992) note the effect of water 
trading and sales, especially in Rabi season, in promoting equity. 



  
 
 

8

Shah (1991) argues that the expansion of irrigation through water markets 

provides increases in cropping intensity and the demand for agricultural labor, which 

ultimately benefit the landless and those who rely on wage labor for household income.  

Increased employment opportunity is one of the biggest advantages for landless members 

of pump groups in Bangladesh (Wood and Palmer-Jones 1990). 

Other researchers on water markets voice concern regarding who appropriates the 

gains from irrigation (e.g. Pant 1991b).  Groundwater is an open access resource, but 

ownership of wells and pumps is required to extract the water.  The prospect of 

exploitative "water lords" has been raised, especially where control over water through 

well ownership reinforces inequality based on land and other assets (e.g. Barah 1992).  

Shah (1991) suggests that well owners extracting monopoly rents from the sale of water 

is most likely to be problematic where the markets are not competitive.  Since water 

transactions are restricted by topography and the distance between source and field, 

market competition is more difficult to achieve.  However, Shah (1991) suggests that the 

availability of groundwater resources and alternative irrigation supplies (especially canal 

water), a high density of wells, and the presence of lined conveyance structures can 

reduce the sellers' monopoly power and hence the price of water.6   

Empirical research and anecdotal evidence indicate a variety of contract forms 

and wide range of prices in water markets.  Buyers may be required to provide labor, 

fuel, or a share of the crop, though the tendency is to move toward a cash charge per hour 

of water supplied as water markets develop (Chaudhry 1990, Shah 1991).   

                                                 
     6  Pant (1991b: 277) also observed that the relative social and economic position of buyers and sellers 
affects water rates.  In Orissa State, India, small farmers selling water to large landowners charged less than 
nearby large farmers selling to small landowners. 



  
 
 

9

Research has shown that the energy pricing structure has a major influence on the 

price of water in private groundwater markets (Shah 1985, 1991).  In India, a flat monthly 

rate based on the horsepower of the engine is associated with a much lower water price 

than a unit charge based on electricity consumption.  With the flat rate power tariff, the 

marginal cost of pumping is dramatically reduced, and therefore well-owners maximize 

their profits by pumping and selling as much water as possible (within the limitations of 

groundwater and power availability).  Water from diesel wells is consistently sold at a 

higher rate than water from electric wells of similar capacity because of the higher 

energy, capital, and maintenance costs. 

Kolavalli (1989) notes that transactions are not impersonal, but are part of multi-

stranded linkages in which buyers may give preference to relatives or those with whom 

they have other relationships, either through lower water rates or priority for service.  

This may be a way of dealing with high transactions costs for water sellers, particularly 

where water is provided on credit (see also Bardhan 1984).  Detailed observations on the 

relationships between buyer and seller and their effects on price and reliability of 

irrigation services are difficult to collect, and therefore have not been carefully examined 

in previous studies. 

While several studies of water markets have dealt with the price of water (e.g. 

Kolavalli and Atheeq 1990; Shah 1989), there has been less analysis of the reliability of 

purchased private irrigation services and its impact on productivity.  Chambers, Saxena 

and Shah (1989) cite farmers' preference for purchased irrigation water above canal 

supplies as evidence of quality of service, but admit the difficulty of estimating adequacy, 

reliability, and other indicators.  Freeman, Lowdermilk and Early (1978) and Renfro 



  
 
 

10

(1982) studied water trading and sales in Pakistan as a means of increasing farmers' 

control over irrigation supplies.  The former used yields as a proxy indicator of quality of 

irrigation services; the latter used data on water and cash input use, cropping intensity, 

and gross income per unit area.  Both studies found that, while water purchases increased 

productivity over canal irrigation alone, they did not have as great an effect as tubewell 

ownership because tubewell ownership provided a higher degree of control. 

The following sections of this paper use empirical data to address both 

productivity and equity aspects of water markets in rural Pakistan.  Data on reliability of 

irrigation service and on agricultural production allow us to deal with the impact of water 

markets in greater detail than previous studies.  Household surveys on various aspects of 

agricultural production and rural poverty conducted by the International Food Policy 

Research Institute (IFPRI) in Faisalabad, Attock, Dir, and Badin Districts during 1990 to 

1992 provide the basis for much of the analysis.  While the latter three districts were 

selected to represent the poorest infrastructure development in Punjab, NWFP and Sindh 

provinces, Faisalabad was included to represent a leading agricultural district (see 

Alderman and Garcia, 1991).  Data on household assets and agricultural production are 

available from 1986 to 1992, but the last full survey round, covering the 1990-91 

agricultural year, provides the greatest detail on agricultural production and irrigation, 

and will therefore be used in this paper.  Plot-level data on soil characteristics from all 

sample farmers are also available from 1992.  A re-survey of all tubewell owners and 

households participating in water markets, conducted in 1992, provides detail on these 

transactions and on reliability of water markets.  While the study attempts to link findings 



  
 
 

11

from the study areas to patterns of water markets indicated in other, more nationally 

representative studies, the findings are largely based on Faisalabad and Dir Districts. 

 

 2.  WHO PARTICIPATES IN WATER MARKETS?   

Water markets are most prevalent in the Punjab and Northwest Frontier Provinces 

of Pakistan, where groundwater irrigation is most developed.  Punjab contains the highest 

number of well-owners who sell water, as well as the highest number of private wells.  

However, in NWFP a higher proportion of well-owners are involved in water markets: 

according to NESPAK (1991) data, 25 percent of sample well-owners reported selling 

water in NWFP, compared to 22 percent in Punjab, 3.5 percent in Sindh, and 2.5 percent 

in Baluchistan.7   

In the IFPRI study, water markets were found only in Faisalabad District of 

Punjab and Dir District of NWFP.  Attock District of Punjab is a barani (rainfed) area, 

and no Attock farmers in our sample owned or used tubewells.  The study villages in 

Badin District of Sindh are largely underlain by saline groundwater aquifers, which pose 

a serious constraint to groundwater irrigation and water market development.  Only one 

of the sample farmers from Badin owns a tubewell, but does not sell water from it; none 

reported buying groundwater.   

                                                 
     7 Provincial figures reported in the NESPAK (1991) final report are a simple average percentage of well 
owners selling water across all districts.  The percentages reported here are a weighted average of 
percentage selling, with the number of tubewells in each district as the weights.  



  
 
 

12

The study areas in Faisalabad District lie within the command area of the Rakh, 

Jhang, and Gugera Branches of the Lower Chenab Canal, with a relatively flat terrain.  

Canals provide the sole source of irrigation to 53 percent of the cultivated area in the 

Division (including Jhang and Toba Tek Singh Districts), with groundwater irrigating an 

additional 13 percent of the area, and conjunctive use of surface and groundwater on 31 

percent of the area (Punjab Bureau of Statistics 1988).  Groundwater use in the district is 

less than recharge, but there are some problems with groundwater salinity.  Annual 

rainfall is under 500 millimeters, so cultivation is heavily dependent on irrigation.  The 

major crops grown in Faisalabad are wheat, sugarcane, cotton, rice, and maize.  Dir 

district has relatively hilly terrain.  With higher annual rainfall (averaging 1364 

millimeters), the area is less dependent on irrigation.  The main sources of irrigation are 

small-scale surface systems, with some tubewells.  The major crops are wheat, tomato, 

onion, and maize. 

The pattern of water market participation found in the IFPRI study, with greatest 

activity in canal-irrigated areas of Punjab and in NWFP, is largely borne out by data from 

the National Input Output Survey of Major Crops, which interviewed 1700 farmers 

distributed across all agroecological zones in Pakistan,8 with the exception of 

Baluchistan.  Table 1 indicates that the highest proportion of farmers relying on 

purchased groundwater was found in the cotton/wheat zone of Punjab (Sahiwal, 

Bahawalnagar, Bahawalpur, R.Y. Khan, and Multan/Vehari districts), where over a third 

of all farmers buy tubewell water.  The mixed cropping zone of Punjab, which includes 

Faisalabad District, had the second highest proportion of tubewell water buyers, with 

                                                 
     8 For details on the classification of agroecological zones, see Pinkney (1989). 



  
 
 

13

21.8 percent.  Although 16 percent of farmers in the barani zone of Punjab, which 

includes Attock, purchase groundwater and 36 percent own tubewells, tubewell use is 

concentrated in Jhelum District, and is very low in Attock District itself (Punjab, Bureau 

of Statistics 1988:53).  



 
14

 
 T

ab
le

 1
--

T
yp

e 
of

 ir
ri

ga
tio

n 
an

d 
ac

ce
ss

 to
 tu

be
w

el
l w

at
er

 b
y 

ag
ro

ec
ol

og
ic

al
 z

on
e 

    
   

   
   

   
   

   
  P

un
ja

b 
Zo

ne
s  

   
   

   
   

   
   

  S
in

dh
 Z

on
es

   
N

W
FP

 Z
on

e 
 

R
ic

e/
 

 
C

ot
to

n/
 

Lo
w

 
 

C
ot

to
n/

 
R

ic
e/

 
Ex

ce
pt

 
 

W
he

at
 

M
ix

ed
 

W
he

at
 

In
te

ns
ity

 
B

ar
an

i 
W

he
at

 
O

th
er

 
D

I K
ha

n 
To

ta
l 

   U
ni

rr
ig

at
ed

 
3 

0 
4 

0 
24

 
0 

0 
71

 
10

2 
 

(0
.7

) 
(0

.0
) 

(1
.0

0)
 

(0
.0

) 
(4

8.
0)

 
(0

.0
) 

(0
.0

) 
(5

1.
4)

 
(6

.2
) 

 Pu
bl

ic
 ir

rig
at

io
n 

10
8 

27
 

10
6 

69
 

0 
11

9 
13

6 
59

 
62

4 
  s

ou
rc

es
 o

nl
y 

(2
6.

5)
 

(1
7.

3)
 

(2
7.

4)
 

(3
0.

4)
 

(0
.0

) 
(8

5.
0)

 
(9

7.
1)

 
(4

2.
8)

 
(3

7.
9)

 
 Pu

rc
ha

se
d 

tu
be

w
el

l 
39

 
1 

16
 

14
 

8 
3 

0 
5 

86
 

  w
at

er
 o

nl
y 

(9
.6

) 
(0

.6
) 

(4
.1

) 
(6

.2
) 

(1
6.

0)
 

(2
.1

) 
(0

.0
) 

(3
.6

) 
(5

.2
) 

 C
an

al
 a

nd
 p

ur
ch

as
ed

 
19

 
33

 
12

1 
54

 
0 

1 
1 

2 
23

1 
  t

ub
ew

el
l w

at
er

 
(4

.7
) 

(2
1.

2)
 

(3
1.

3)
 

(2
3.

8)
 

(0
.0

) 
(0

.7
) 

(0
.7

) 
(1

.4
) 

(1
4.

0)
 

 O
w

n 
tu

be
w

el
l w

at
er

 
12

8 
12

 
30

 
13

 
18

 
1 

0 
1 

20
3 

  o
nl

y 
(3

1.
4)

 
(7

.7
) 

(7
.8

) 
(5

.7
) 

(3
6.

0)
 

(0
.7

) 
(0

.0
) 

(0
.7

) 
(1

2.
3)

 
 C

an
al

 a
nd

 o
w

n 
11

0 
83

 
11

0 
77

 
0 

16
 

3 
0 

39
9 

  t
ub

ew
el

l w
at

er
 

(2
7.

0)
 

(5
3.

2)
 

(2
8.

4)
 

(3
3.

9)
 

(0
.0

) 
(1

1.
4)

 
(2

.1
) 

(0
.0

) 
(2

4.
3)

 
   To

ta
l s

am
pl

e 
si

ze
 

40
7 

15
6 

38
7 

22
7 

50
 

14
0 

14
0 

13
8 

16
45

 
 

(2
4.

7)
 

(9
.5

) 
(2

3.
5)

 
(1

3.
8)

 
(3

.0
) 

(8
.5

) 
(8

.5
) 

(8
.4

) 
(1

00
.0

) 
   So

ur
ce

: 
C

al
cu

la
te

d 
fr

om
 d

at
a 

in
 N

at
io

na
l I

np
ut

 O
ut

pu
t S

ur
ve

y 
of

 M
aj

or
 C

ro
ps

, c
ol

le
ct

ed
 b

y 
A

gr
ic

ul
tu

ra
l U

ni
ve

rs
ity

, F
ai

sa
la

ba
d.

  A
gr

oe
co

lo
gi

ca
l z

on
es

 
de

fin
ed

 a
s i

n 
Pi

nc
kn

ey
, T

. C
.  

Th
e 

de
m

an
d 

fo
r p

ub
lic

 st
or

ag
e 

of
 w

he
at

 in
 P

ak
is

ta
n.

  I
FP

R
I R

es
ea

rc
h 

R
ep

or
t 7

7.
  (

W
as

hi
ng

to
n,

 D
.C

.: 
 In

te
rn

at
io

na
l F

oo
d 

Po
lic

y 
R

es
ea

rc
h 

In
st

itu
te

, 1
98

9)
. 

N
ot

e:
  F

ig
ur

es
 in

 p
ar

en
th

es
es

 a
re

 p
er

ce
nt

ag
e 

of
 fa

rm
er

s i
n 

zo
ne

.  
a D

I K
ha

n 
D

is
tri

ct
 o

f N
W

FP
 is

 in
cl

ud
ed

 in
 P

un
ja

b 
lo

w
-in

te
ns

ity
 z

on
e.

 



  
 

15

Less than 3 percent of farmers use any groundwater in either of the zones in 

Sindh.  Of 138 farmers interviewed in the NWFP zone, only 5 percent reported pur-

chasing water from tubewells. 

The extent of water market participation among IFPRI sample farmers in 

Faisalabad and Dir districts is indicated in Table 2.  Nearly half of the sample farmers in 

Faisalabad District purchase tubewell water, more than twice the number who own tube-

wells.  Of the 22 sample farmers who own tubewells in Faisalabad, only 5 reported 

selling tubewell water.  In Dir, where groundwater irrigation is less prevalent, nine 

percent of all sample farmers purchase water, approximately the same proportion who 

own tubewells, but twice the number who sell tubewell water.  Water markets are most 

pervasive in Faisalabad, where groundwater purchasers or sellers were found in all six 

study villages (Table 3).  Within the villages, participation rates ranged from 0 to 16 

percent of farmers (0 to 100 percent of tubewell owners) selling water, and 0 to 81 

percent of farmers purchasing water.  Although all villages in Faisalabad fall within the 

command area of public canal irrigation systems, the watercourses in Jaranwala receive 

almost no surface water.  Thus many of the farmers in Jaranwala have invested in wells, 

often jointly with other farmers, giving it a significantly higher proportion of well owners 

(75 percent of sample farmers) than other villages.   

Only five of eleven study villages in Dir had any groundwater use among sample 

farmers, and groundwater markets were reported in three of those five 

villages (Table 4).  The water market is most active in  Katigram, where more 

than one-fourth of the farmers own tubewells, and nearly all well owners sell water.   



  
16

 T
ab

le
 2

--
Sa

m
pl

e 
fa

rm
er

 p
ar

tic
ip

at
io

n 
in

 tu
be

w
el

l w
at

er
 m

ar
ke

t i
n 

Fa
is

al
ab

ad
 a

nd
 D

ir
 D

is
tr

ic
ts

, 1
99

2 
  

   
  D

is
tri

ct
   

   
   

 
Fa

is
al

ab
ad

 
D

ir 
To

ta
l 

  Tu
be

w
el

l o
w

ne
r 

22
 

7 
29

 
 

(2
2.

2)
 

(7
.9

) 
(1

5.
4)

 
 Tu

be
w

el
l w

at
er

 
   

se
lle

r 
5 

4 
9 

 
(5

.1
) 

(4
.5

) 
(4

.8
) 

 Tu
be

w
el

l w
at

er
 

   
bu

ye
r 

49
 

8 
57

 
  

(4
9.

5)
 

(9
.0

) 
(3

0.
3)

 
 To

ta
l s

am
pl

e 
si

ze
 

99
 

89
 

18
8 

  So
ur

ce
:  

IF
PR

I s
ur

ve
y 

da
ta

, 1
99

2.
 

N
ot

e:
 

Fi
gu

re
s i

n 
pa

re
nt

he
se

s a
re

 p
er

ce
nt

ag
e 

of
 sa

m
pl

e 
fa

rm
er

s i
n 

th
e 

di
st

ric
t. 

 



  
17

T
ab

le
 3

--
W

at
er

 m
ar

ke
t p

ar
tic

ip
at

io
n 

am
on

g 
IF

PR
I s

am
pl

e 
fa

rm
er

s i
n 

Fa
is

al
ab

ad
 D

is
tr

ic
t, 

by
 v

ill
ag

e,
 1

99
2 

     
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

  V
ill

ag
e 

   
   

   
   

   
   

   
   

   
   

   
   

Sa
dd

oa
na

   
 S

in
gp

ur
a 

   
Ja

ra
nw

al
a 

   
 G

oj
ra

   
   

K
ha

lis
ab

ad
   

 S
um

un
dr

i  
   

To
ta

l 
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

  C
ha

k 
   

Su
ba

da
rw

al
a 

 
  Tu

be
w

el
l o

w
ne

r  
   

   
   

   
   

  0
   

   
   

   
   

 1
   

   
   

   
  1

5 
   

   
   

   
1 

   
   

   
   

  4
   

   
   

   
   

 1
   

   
   

   
 2

2 
   

   
   

   
   

   
   

   
   

   
   

   
 (0

.0
)  

   
   

   
(1

6.
7)

   
   

   
(7

5.
0)

   
   

   
(9

.1
)  

   
   

 (1
8.

2)
   

   
   

  (
5.

3)
   

   
   

(2
2.

2)
 

 Tu
be

w
el

l w
at

er
   

   
   

   
   

   
  0

   
   

   
   

   
 1

   
   

   
   

   
3 

   
   

   
   

 0
   

   
   

   
  0

   
   

   
   

   
 1

   
   

   
   

  5
 

   
se

lle
r  

   
   

   
   

   
   

   
   

 (0
.0

)  
   

   
   

(1
6.

7)
   

   
   

 (1
5.

0)
   

   
  (

0.
0)

   
   

   
 (0

.0
)  

   
   

   
 (5

.3
)  

   
   

 (5
.1

) 
 Tu

be
w

el
l w

at
er

   
   

   
   

   
   

15
   

   
   

   
   

 4
   

   
   

   
   

 5
   

   
   

   
 7

   
   

   
   

 1
8 

   
   

   
   

   
0 

   
   

   
   

45
 

   
bu

ye
r  

   
   

   
   

   
   

   
  (

71
.4

)  
   

   
   

(6
6.

7)
   

   
   

 (2
5.

0)
   

   
  (

63
.6

)  
   

   
(8

1.
8)

   
   

   
  (

0.
0)

   
   

   
(4

5.
5)

 
 To

ta
l s

am
pl

e 
si

ze
   

   
   

   
   

21
   

   
   

   
   

 6
   

   
   

   
   

20
   

   
   

  1
1 

   
   

   
   

22
   

   
   

   
   

19
   

   
   

   
 9

9 
  So

ur
ce

:  
IF

PR
I s

ur
ve

y 
da

ta
 1

99
2.

 
N

ot
e:

 
Fi

gu
re

s i
n 

pa
re

nt
he

se
s a

re
 p

er
ce

nt
ag

e 
of

 sa
m

pl
e 

fa
rm

er
s i

n 
th

e 
vi

lla
ge

. 



  18 
 
 

TUBEWELL OWNERS 

What are the characteristics of tubewell owners and water purchasers?  Table 5 

presents a logistic regression (logit) model used to examine tubewell ownership among 

sample farmers in Faisalabad and Dir districts.  The logit technique allows us to examine 

the effect of a number of variables on the underlying probability of a dichotomous 

dependent variable, such as the probability of owning a tubewell.  In this model, land 

ownership, age of head of household, whether a household has a member who has 

worked or is working abroad, and dummy variables for Jaranwala village and Dir district 

(areas with less access to canal irrigation) are hypothesized to influence the probability of 

owning a tubewell.  The dummy variables are proxy indicators for availability of 

alternative water supply.  Other indicators of water supply, such as position of farm 

relative to canal systems and discharge in the watercourse, might refine the model, but 

are difficult to quantify and measure. Household wealth or income are not included in the 

model because it is likely that tubewell ownership has contributed to wealth or income, 

rather than the reverse.  Similarly, factors such as cropping pattern, which influences 

demand for tubewell water, are not included because no indicator is available for farmers' 

desired cropping pattern, and availability of tubewell water has a stronger influence on 

cropping pattern than actual cropping pattern has on availability of tubewell water.  



 
19

 
  T

ab
le

 4
--

W
at

er
 m

ar
ke

t p
ar

tic
ip

at
io

n 
am

on
g 

IF
PR

I s
am

pl
e 

fa
rm

er
s i

n 
D

ir
 D

is
tr

ic
t, 

by
 v

ill
ag

e,
 1

99
2 

    
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

 V
ill

ag
e 

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
K

at
ig

ra
m

   
 B

at
an

   
 S

ha
h 

A
la

m
   

 K
ha

np
ur

   
  K

am
an

ga
ra

   
  6

 O
th

er
   

  T
ot

al
 

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
 B

ab
a 

   
   

   
   

   
   

   
   

   
   

   
   

   
   

  V
ill

ag
es

 
  Tu

be
w

el
l o

w
ne

r  
   

   
   

   
   

   
   

5 
   

   
   

   
0 

   
   

   
  1

   
   

   
   

   
  0

   
   

   
   

   
  1

   
   

   
   

   
0 

   
   

   
   

7 
   

   
   

   
   

   
   

   
   

   
   

   
   

  (
27

.8
)  

   
  (

0.
0)

   
   

 (1
4.

3)
   

   
   

   
(0

.0
)  

   
   

   
 (2

0.
0)

   
   

   
 (0

.0
)  

   
   

 (7
.9

) 
 Tu

be
w

el
l w

at
er

   
   

   
   

   
   

   
  4

   
   

   
   

 0
   

   
   

   
 0

   
   

   
   

   
  0

   
   

   
   

   
  0

   
   

   
   

   
0 

   
   

   
   

4 
   

 se
lle

r  
   

   
   

   
   

   
   

   
   

 (2
2.

2)
   

   
 (0

.0
)  

   
   

(0
.0

)  
   

   
   

  (
0.

0)
   

   
   

   
 (0

.0
)  

   
   

  (
0.

0)
   

   
   

(4
.5

) 
 Tu

be
w

el
l w

at
er

   
   

   
   

   
   

   
  1

   
   

   
   

 3
   

   
   

   
 0

   
   

   
   

   
  4

   
   

   
   

   
  0

   
   

   
   

   
0 

   
   

   
   

8 
   

 b
uy

er
   

   
   

   
   

   
   

   
   

   
(5

.6
)  

   
  (

25
.0

)  
   

   
(0

.0
)  

   
   

   
 (2

1.
1)

   
   

   
   

(0
.0

)  
   

   
  (

0.
0)

   
   

   
(9

.0
) 

 To
ta

l s
am

pl
e 

si
ze

   
   

   
   

   
   

 1
8 

   
   

   
 1

2 
   

   
   

   
7 

   
   

   
   

   
19

   
   

   
   

   
  5

   
   

   
   

 2
8 

   
   

   
   

89
 

  So
ur

ce
:  

IF
PR

I s
ur

ve
y 

da
ta

, 1
99

2.
 

N
ot

e:
  F

ig
ur

es
 in

 p
ar

en
th

es
es

 a
re

 p
er

ce
nt

ag
e 

of
 sa

m
pl

e 
fa

rm
er

s i
n 

th
e 

vi
lla

ge
. 



  
 
 

20

Table 5--Results of logistic regression model for tubewell ownership  
 
Independent    Wald 
Variables Units Coefficient T Ratio Statistic 
  
 
Size of land ownership acres .122 ** 4.03 16.24 
Age of head of household years .051 ** 2.34 5.48 
Relative abroad dummy 1.668 ** 2.15 4.63 
Jaranwala village dummy 4.458 ** 5.41 29.32 
Dir district dummy .036 .05 .00 
Constant                              -6.829 ** -4.35 18.90 
 
 
 Model Chi-Square =  66.9  **  with 5 degrees of freedom 
 Number of observations = 182.0 
 
 Classification Table for Tubewell Ownership 
 
                                                  Predicted              Percent  
                                 NOT OWNER       OWNER     Correct 
 Observed             
   NOT OWNER                  147                    7              95.4 
                     
   OWNER                           12                   16              57.1 
 
                                                                     Overall  89.6 
  
 
Source: IFPRI survey data, 1991/92. 
**  Significant at 5 percent probability level 
*   Significant at 10 percent probability level 



  
 
 

21

Results of this model indicate that land ownership has a strong positive effect on 

well ownership, implying that households owning more land are more likely to own 

wells.  The age of head of household has a significant positive effect on well ownership, 

perhaps because farmers invest in tubewells as the household becomes established.  

Households with a member currently working abroad or returned from abroad are also 

significantly more likely to own wells.  This seems to indicate that remittances are a 

source of financing for tubewell investment.9  The dummy variable for Jaranwala village 

has a large and significant coefficient.  As noted above, the lack of alternative canal 

irrigation supplies has pushed these farmers to purchase tubewells, and joint investment 

has enabled even small farmers in this village to own at least a partial share of a well.  

The pattern of well ownership in Dir district, however, is not significantly different from 

that of Faisalabad district.  This model correctly predicts the well ownership status of 90 

percent of all cases.   

                                                 
     9 The reverse direction of causality--that income derived from tubewell ownership has financed 
international migration--is unlikely.   Adams' (1992) study of migration among households in the IFPRI 
sample found that households with less than an acre of irrigated land were most likely to migrate. 



  
 
 

22

Although small holding size is not an insurmountable obstacle to well ownership 

(as demonstrated by a high proportion of joint well owners with small holdings in 

Jaranwala village), it is a constraint to widespread tubewell ownership.  Tenants, 

especially those with no land of their own, are at a disadvantage in tubewell ownership 

because they do not have secure enough rights to land on which to install a tubewell, and 

are likely to face credit constraints for tubewell investment (see Malik, Broca and Gill 

1992).  In the survey on water markets, over 60 percent of groundwater purchasers cited 

the expense of purchasing a tubewell as the reason why they did not have their own 

wells, but 25 percent cited a lack of land ownership or too small a holding size as the 

reason for not investing in a well (Table 6).  Groundwater quality problems also prevent 

farmers from installing their own wells.  Water markets meet a need for water among 

those who have too little land, cannot afford tubewells, or find the investment not 

worthwhile, and those who have problems with the groundwater quality on their own 

land. 

TUBEWELL WATER PURCHASERS 

A logistic regression model, similar to that for tubewell ownership, has been 

calculated to identify factors which predict who purchased tubewell water during rabi and 

kharif of 1991-92 (Table 7).10  Since remittances are less likely to have an impact on 

water purchases than on investment in tubewells, the variable for relatives abroad is 

omitted from this model, but a variable for season is added to see if there is a significant 

difference between rabi and kharif irrigation purchases.  As in the model for tubewell 

                                                 
     10 Because of the small number of tubewell owners in the sample, it is not possible to develop a model to 
predict which tubewell owners sell water. 



  
 
 

23

ownership, cropping patterns are not included  because they are determined largely by the 

availability of water.  The location of fields relative to the canal system and to tubewells 

whose owners are willing to sell, along with ground- water quality are likely to affect 

water purchases, but these factors are not included because the data are unavailable.11 

Table 6--Water buyers' reasons for not owning tubewells 
  
      District             

 Faisalabad Dir Total 

Too expensive 40 14 54 
 (54.1) (93.3) (60.7) 
Holding too small 11 0 11 
 (14.9) (0.0) (12.4) 
Not landowner 12 0 12 
 (16.2) (0.0) (13.5) 
Land not near canal 1 0 1 
 (1.4) (0.0) (1.1) 
Groundwater quality 6 1 7 
 (8.2) (6.6) (7.8) 
Only buy when own TW broken 3 0 3 
 (4.1) (0.0) (3.4) 
Purchase as supplement only 1 0 1 
 (1.4) (0.0) (1.1) 
Total number of water buyers 74 15 89 
  
Source:  IFPRI survey data, 1992. 
Note: Figures in parentheses are percentage of water buyers. 

                                                 
     11 Tubewell ownership was not included in the model because of the high multicollinearity between land 
ownership and tubewell ownership.  Similarly, operational holding size was not included because it is 
highly correlated with land ownership.  An alternative specification of the model, with operational holding 
instead of land ownership, did not show a significant effect of holding size on water purchases.   



  
 
 

24

Table 7--Results of logistic regression model for tubewell water purchase 
  
Independent    Wald 
Variables Units Coefficient T Ratio Statistic 
  
 
Size of land ownership acres - .039 ** -2.28 5.22 
Age of head of household years - .031 ** -3.30 10.87 
Season dummy - .355 -1.24 1.54 
Jaranwala village dummy -1.398 ** -3.38 11.46 
Dir district dummy -3.262 ** -8.15 66.50 
Constant  2.250 ** 3.98 15.82 
  
 
 Model Chi-Square =  118.4 **  with 5 degrees of freedom 
 Number of observations =  352.0 
 
 
 Classification Table for Tubewell Ownership 
 
                                            Predicted                     Percent 
 NOT BUYING BUYING Correct 
 
 Observed             
   NOT BUYING 221 31 87.7 
                     
   BUYING 36 64 64.0 
 
   Overall  81.0 
  
 
Source:  IFPRI survey data, 1990-1992. 
**  Significant at 5 percent probability level 
*   Significant at 10 percent probability level 



  
 
 

25

This model shows that, whereas the size of land ownership and the age of 

household head have a strong positive effect on tubewell ownership, these variables have 

a significant negative effect on water purchases.12  Thus younger households with less 

land are more likely to purchase groundwater than older households owning substantial 

amounts of land.  Tenants are likely to be dependent on other households' tubewells, as 

well as other households' land, for their cultivation.  Farmers in Dir are significantly less 

likely to purchase groundwater than those in Faisalabad, in part because of the lower 

availability of tubewells in Dir.13  Farmers in Jaranwala are also significantly less likely 

to purchase water.  In this village joint ownership, rather than water markets, provides 

small farmers with access to groundwater.  Because alternative sources of irrigation are 

not available, farmers in that village seek to assure themselves of access to groundwater 

by investing in wells rather than depending on groundwater purchases.  The season does 

not have a significant effect on water purchases: 30 percent of farmers reported 

purchasing irrigation in kharif, compared to 25 percent in rabi. 

Land ownership and age are indicators of overall status of farm households.  

Investment in tubewells are not made for prestige, but the status of households can enable 

them to mobilize the resources needed for tubewell investment, including financial 

resources and government assistance (such as electricity connections).  Therefore, it is 
                                                 
     12 In a logit model for groundwater purchasing, using cross-sectional survey data from 5 states in India, 
Saleth (1991) also found a significant negative effect of farm size on likelihood of purchasing in 3 of 5 
states.  The other two states are characterized by very small and fragmented holdings and higher average 
rainfall.  Thus the need for supplemental groundwater was less, and the larger farmers in those states were 
as likely to purchase groundwater as small farmers. 
     13 The effect of number and density of tubewells in a village was tested in an alternate specification of 
the model.  Tubewell density has a significant positive effect on the likelihood of water purchases, both in 
addition to the dummy variables for Dir and Jaranwala, and without the dummy variables included.  The 
variable for tubewell density is not included in the final model because there are multicollinearity problems 
when it is included along with the dummy variables for location, and tubewell density does not explain as 
much as the two dummy variables. 



  
 
 

26

not surprising that higher-status households are more likely to own wells, and lower-

status households are more likely to rely on tubewell water purchases.14  However, not 

only low-status households purchase water: 7 of 28 tubewell owners in the IFPRI sample 

also purchase water.  Water purchases may provide a backup when a farmer's own well is 

not functioning, or may be used to irrigate land that cannot be served by a farmer's well.  

In several cases, farmers preferred buying water to operating their own wells because 

purchased water from electric-powered wells was cheaper than using their own tractor-

powered tubewells. 

Results of these model indicate that, whereas private tubewells are likely to be 

owned by large farmers, water markets improve equity of groundwater use by making 

water available to small landowners or tenants and younger households--those farmers 

who are least likely to own tubewells.  In Faisalabad, where water markets are most 

active, they provide the only access to groundwater for approximately half of the farmers 

who own less than 12.5 acres. 

                                                 
     14 A comparable pattern of larger landowners owning wells and smaller landowners purchasing tubewell 
water, based on National Input Output Survey data from the Punjab mixed cropping zone, is shown in 
Appendix Table 1. 



  
 
 

27

3.  RELATIONSHIPS BETWEEN WATER BUYERS AND SELLERS  

PHYSICAL RELATIONSHIPS 

Water markets are not perfectly competitive markets in which buyers are free to 

choose among a number of sellers.  In many areas there are not a large number of water 

sellers, and, under most conditions prevailing in Pakistan, tubewell irrigation water is not 

a commodity which can be transported far from the source to the area of application.  

Conveyance losses between the tubewell and the field restrict purchasers to buying from 

tubewells located in close proximity of their fields.  The distance over which it is feasible 

and economically viable to transmit water depends on the soils, topography, and type of 

channel used to convey the water.  In the IFPRI study, the average distance between the 

tubewell and purchasers' fields was 600 meters in Faisalabad District, and 180 meters in 

Dir, which has more undulating topography.  Use of lined watercourses or field channels, 

which have lower transmission losses than unlined channels, increases the distance over 

which water can be transported.  Ten of the thirteen cases in which the distance between 

source and fields was over 1000 meters used lined watercourses (including one case in 

which water travelled down 3 kilometers of lined watercourse between the well and 

field).  

Lined canals and pipes ensure that water purchasers receive more of the water 

they pay for from the tubewell, and permit sales to a wider potential number of fields 

from each tubewell.15  Thus, they go hand in hand with the development of competitive 

water markets.  They allow purchasers to obtain water within a wider radius of their 

                                                 
     15 Underground pipes can even, to some extent, overcome topographic limitations to water sales, by 
enabling water to reach fields at a higher level than the 
tubewell. 



  
 
 

28

fields, thereby increasing the number of potential suppliers.  Shah and Raju (1988) report 

that as competitive water markets developed in Gujarat state, India, tubewell water sellers 

who wanted to maintain clients install lined conveyances to ensure that water reaches as 

many buyers as possible with low losses.   

Despite these advantages of water markets, there has been little private investment 

in lining or pipes in the IFPRI sample areas.  Four farmers reported using lined field 

channels to convey purchased tubewell water part of the way (though in two of these 

cases the lined field channels were only the first 20 meters from the tubewell).  Three 

sellers used underground pipes, which have the lowest conveyance losses.  Pukka (lined) 

watercourses are used more frequently than lined field channels or pipes because 

watercourse lining results from government and collective farmer investment, while 

lining field channels or installing pipes requires considerable private investment. 

Farmers use watercourses for tubewell water despite the Canal and Drainage Act 

provision that all who do so are assessed the water rate for canal irrigation use.  A 

NESPAK (1991) study suggests that this law restricts the sale of tubewell water, and 

recommends allowing farmers to use watercourses for tubewell water without charge to 

encourage water markets and the mixing of marginal-quality tubewell water with fresh 

canal water supplies.  The Canal and Drainage Act further restricts transporting tubewell 

water by prohibiting farmers from carrying water across public watercourses.  It may be 

worth lifting this restriction, provided the pipes or other structures used to carry the 

tubewell water do not block or erode the watercourse. 



  
 
 

29

Renfro's (1982) study notes an association between private tubewells and 

collective activity among farmers, particularly with regard to watercourse lining.16  

Renfro and Sparling (1986:206) suggest the reason for this is that "Farmers with 

cooperative neighbors are more likely to invest in tubewells, and the presence of private 

tubewells gives farmers new reasons to cooperate with each other."  However, this 

explanation omits the role of watercourse lining in reducing transmission losses of high-

value groundwater.  An alternative explanation is that, with tubewell use in general, and 

water market sales in particular, farmers recognize the value of canal lining and have 

greater incentive to reduce water losses.  With lined watercourses, tubewell water can be 

conveyed over a greater distance between the well and fields, thus expanding the 

potential market from tubewells along the watercourse.  Renfro and Sparling (1986:206) 

argue that watercourses with private tubewells are less in need of lining because they 

have additional sources of water.  However, watercourse rehabilitation not only 

conserves canal water, but by reducing transmission losses of tubewell water, lining can 

also encourage the development of competitive water markets.  Therefore, in areas where 

watercourse lining is to be done, it may be appropriate to give priority to lining 

watercourses with private tubewells as a means of fostering water markets.17 

SOCIAL RELATIONSHIPS 

Physical proximity is not the only relevant relationship which influences the 

development of competitive water markets.  Social relationships between buyer and seller 

                                                 
     16 Under the On-Farm Water Management projects implemented by the Government of Pakistan with 
USAID and World Bank assistance, farmers are required to organize into Water Users' Associations in 
order to receive government assistance for watercourse improvement (see Byrnes 1992).   
     17 Lining watercourses will also reduce groundwater recharge to some extent, and thus reduce 
availability of tubewell water.  However, there will continue to be recharge from irrigated fields. 



  
 
 

30

may also restrict the sale and purchase of groundwater if tubewell owners are only 

willing to sell to close relatives or those with whom they have other ties.  This does not 

appear to be a problem in the IFPRI study areas.  Only 22 percent of water market 

transactions in Faisalabad District were between close relatives.  Approximately one third 

(37 percent) of transactions were between members of the same biradari, which repre-

sents a broader social grouping.  Sale of tubewell water between kin is even more rare in 

Dir, where only 2 percent of transactions were between close relatives or biradari 

members.  There, water market transactions were more likely to follow patron-client ties, 

with 37 percent of sales reported between landlord and sharecropper (compared to less 

than 1 percent in Faisalabad District). 

As noted above, selling water to relatives is a means of controlling transactions 

costs and ensuring fee repayment.  Relatives may also have the closest land holdings, due 

to inheritance patterns.  However, transactions costs may be higher with relatives, either 

because of quarrels or difficulty in collecting payments.  Shah (1992) hypothesizes that as 

water markets develop, they become de-personalized.  This seems to have taken place in 

the study areas, because personal relationships neither restrict access to groundwater nor 

are shown to have a significant effect on the reliability of deliveries, as discussed below. 

 

 4.  NATURE OF WATER MARKET CONTRACTS 

A flat charge per hour of pumping is the most common form of water market 

contract in both Faisalabad and Dir districts (Table 8).  This type of arrangement occurs 

under diesel, tractor, and electric-powered pumpsets.  Water from diesel pumpsets in 

Jaranwala is commonly sold under an arrangement whereby the buyer supplies the diesel 



  
 
 

31

and motor oil for the pump, and pays an additional fee of Rs 4 to Rs 6 per hour to the 

well owner to cover the wear and tear on the engine.  Sharecropping contracts for water 

are used under both diesel and electric tubewells in Dir. 

 
Table 8--Water market contracts, by type of pump  
           Type of Pump            
 Diesel Tractor Electric Total  
Faisalabad District 
 

Flat charge per hour 9 19 44 72 
Buyer brings fuel 18 0 0 18 
Share of crop 0 0 0 0 
         
Total 27 19 44 90 

 
Dir District 
 

Flat charge per hour 13 0 0 13 
Buyer brings fuel 0 0 0 0 
Share of crop 2 0 6 8 
       
Total 15 0 6 21 

  
Source:  IFPRI survey data, 1992. 
Note: Table includes number of sellers' and buyers' responses about type of contract. 
 

 

Prices under the hourly charge system range from Rs 14 to Rs 80 per hour, 

depending on the pump type, capacity, and location (see Table 9).  The higher price of 

water from tractor tubewells reflects the higher cost of operating this type of pump.  The 

average price of water under the hourly charge system is approximately the same for 

diesel and electric tubewells, although the former are usually more expensive to operate.  

The average capacity of the electric tubewells is higher, though, so the cost per 

horsepower is lower for electric than diesel tubewells.  The  mean hourly cost of water to 



  
 
 

32

the purchasers from diesel tubewells is slightly higher under the buyer-brings-fuel system 

than under the flat hourly charge.  Water sellers with diesel pumps are apparently only 

recovering their own costs under either type of contract.18  The sellers' transactions costs 

in acquiring the fuel and operating or supervising the operations of the pump are 

presumably higher under the hourly charge contracts, but there may be an unwillingness 

to let some purchasers operate the pumps themselves under the buyer-brings-fuel system.  

Among IFPRI sample villages, the buyer-brings-fuel contract was only found in 

Jaranwala, where there is also a high incidence of joint ownership of wells.  However, 

this type of contract is also reported in other areas with conjunctive canal and tubewell 

irrigation, such as in southern Punjab. 

All water transactions under hourly charge contracts in Dir are found in Khanpur.  

The price is Rs 40 to 80 per hour, which is higher, on average, than in 

Faisalabad.  Several factors could account for the higher price in Khanpur: the 

well from which most farmers purchase is large-capacity, powered by a 113-horsepower 

truck engine, and therefore more expensive to operate.19  Irrigation water is also more 

scarce in Dir than in Faisalabad, where canal water is more readily available and 

groundwater tables are generally higher.  A final consideration may be that most water 

purchasers in Khanpur are buying water from their landlord, whereas that is only found in 

one case in Faisalabad.   

                                                 
     18 Unfortunately, much of the information on price of purchased tubewell water comes from water 
buyers, rather than from the sellers.  There are thus not enough data on tubewell operations costs and water 
delivery rates to determine the profit margin for water sales or the exact price per unit water pumped. 
     19 The price per horsepower is lower in Dir than Faisalabad, but this may be misleading because the 
relationship between horsepower and water delivery is not linear, particularly when comparing the pumpset 
powered by a truck engine with lower-horsepower engines designed for tubewells.   



  
 
 

33

Table 9--Average cost of tubewell water, by type of pump and contract  
             Type of Pump              
 Diesel Tractor Electric   
Faisalabad District 
 

Flat charge (Rs per hour) 29.44 43.95 27.82  
 (8.08) (7.56) (7.77) 

  
Buyer brings fuel (Rs per hour) 32.06  n.a. n.a. 
 (5.39) 

 
Dir District 
 

Flat charge (Rs per hour) 49.23 n.a. n.a. 
 (17.54) 

 
Share of crop (percent) 25.00 n.a. 23.15 
 (0.00)  (2.31) 

  
Source:  IFPRI survey data, 1992. 
Note: Costs computed from sellers' and buyers' responses. 
Figures in parentheses are standard deviations. 
n.a.    not applicable. 
 
 

A larger sample of water sellers and purchasers under different ecological and 

socioeconomic conditions would be necessary to estimate the effect of these factors on 

the cost of private tubewell water.  It does not appear, however, that the prices reported 

under the hourly rate or buyer-brings-fuel type of contracts represent a large profit to the 

tubewell owner.  Thus, concerns over water sellers appropriating the value of 

groundwater do not seem justified, particularly in the Faisalabad area. 

Sharecropping contracts for tubewell water are only applied to tomato, onion, and 

some maize cultivation in Dir.  The standard rate is 25 percent of the crop.  Three water 

sellers reported giving a different rate to their tenants: 20 percent as the share for the 

water (in addition to the share for the land), or 50 percent for the land and water 

combined.  The use of sharecropping contracts for provision of water in Dir may reflect 



  
 
 

34

the greater prevalence of sharecropping for land in that area than in Faisalabad.  

Sharecropping contracts for water are also reported for rice and berseem crops in other 

regions (Chaudhry 1990; Dr. Muhammad Jameel Khan, personal communication, March 

1992).  The extent of crop share contracts for water may be underestimated where the 

landlord supplies both land and tubewell water.  While tubewell irrigation may then be 

considered as one of the inputs which the landlord provides (such as fertilizer or plowing 

services), to do so masks the importance of timing and reliability of irrigation service in 

the overall production process. 

It is noteworthy that sharecropping for tubewell water is practiced under the 

cultivation of crops such as tomatoes, onions, or rice, which are sensitive to moisture 

stress at critical periods.  In sharecropping contracts, the water seller has a stake in the 

outcome of the crop, and a share in the risk if the water supply does not meet crop needs.  

Therefore, the seller has an incentive to supply tubewell water in an adequate and timely 

manner.  This is not as critical for crops less susceptible to the timing of irrigation, or 

where alternate sources of irrigation are readily available.  Chaudhry (1990) and Shah 

(1991) have suggested that, as water markets develop, there is a tendency to move toward 

cash contracts.  If, however, sharecropping for water offers a greater incentive for sellers 

to provide reliable irrigation service, this type of contract may remain for water-sensitive 

crops.  Further empirical research on the provision of tubewell water in different 

agroecological zones and for different crops is necessary to establish whether risk-sharing 

is a significant consideration in the choice of contract and timeliness of irrigation service. 



  
 
 

35

5.  AGRICULTURAL PRODUCTIVITY UNDER WATER MARKETS 

Previous studies have shown clear productivity gains to farmers purchasing 

groundwater over those using only public canal or public tubewell supplies, but the gains 

were much less than those obtained by tubewell owners.  The wheat and cotton yield 

increases of tubewell water purchasers (compared to those with canal water only) were 

half as great as the yield increases for tubewell owners in Freeman, Lowdermilk and 

Early's (1978) study.  For rice the gap was narrower: water purchasers obtained 78 

percent of the yield increases of tubewell owners (see Table 10).20 

A study of private tubewells by WAPDA (1980, cited in World Bank 1984) found 

that overall cropping intensity and the proportion of area under water-consumptive crops 

was higher for tubewell owners than for water purchasers.  There was also a yield gap 

between water purchasers and tubewell owners for sugarcane, rice, wheat, and vegetables 

(Table 11).  Part of the difference in yields may be due  to lower applications of irrigation 

water and complementary inputs such as fertilizer and insecticides by tubewell 

purchasers than by owners (even though tubewell purchasers used more inputs and had 

higher yields than non-users for almost all crops).  Renfro (1982) found that the cropping 

intensities, the proportion of area 

                                                 
     20 A WAPDA (1990) study also assesses the productivity impact of purchased water.  However, the 
yield differentials are based on farmers' assessments of what their yields would be with and without 
privately purchased water, and are thus not as reliable as comparisons of actual yields of farmers with and 
without purchased water. 



  
 
 

36

Table 10--Average yields of major crops, by water source in Freeman, Lowdermilk 
             and Early's (1978) study.  
 Canal Public Purchased Own 
Crop Only Tubewell Tubewell Tubewell 
  
 (kg/acre) 
 
Wheat 672 747 784 896 
Rice 522 709 784 859 
Cotton 261 299 373 485  
Source: D. M. Freeman, M. K. Lowdermilk, and A. C. Early.  "Farm irrigation constraints and farmers' 
responses:  Comprehensive field survey in Pakistan".  (Water Management Technical Report No. 48-E, 
Volume V, Colorado State University, Fort Collins, CO, 1978). 
 
Note:  All tubewell water is in addition to canal supplies. 
 
 
 



 
 

  

37

T
ab

le
 1

1-
-S

um
m

ar
y 

of
 in

pu
t u

sa
ge

 a
nd

 y
ie

ld
s f

or
 tu

be
w

el
l u

se
rs

 a
nd

 n
on

-u
se

rs
 in

 W
A

PD
A

 st
ud

y 
  Ite

m
 

U
ni

t 
Ty

pe
 

Su
ga

rc
an

e 
R

ic
e 

G
ar

de
ns

 
V

eg
et

ab
le

 
C

ot
to

n 
W

he
at

 
Pu

ls
es

 
O

ils
ee

ds
 O

th
er

s 
To

ta
l 

   1.
 C

ro
pp

in
g 

pa
tte

rn
 

%
 a

cr
es

 
O

*  
8 

21
 

4 
3 

8 
60

 
15

 
8 

18
 

15
7 

P*  
5 

16
 

2 
2 

8 
36

 
13

 
13

 
19

 
13

6 
N

u*  
3 

7 
1 

1 
7.

5 
50

 
16

 
11

 
15

 
11

3 
 2.

 P
er

 a
cr

e 
us

e 
of

: 
   

i. 
N

itr
og

en
 

50
 k

g 
O

 
1.

5 
1.

5 
0.

5 
0.

9 
0.

5 
1.

0 
- 

0.
2 

0.
5 

ba
g 

P 
1.

0 
1.

3 
0.

5 
0.

5 
0.

5 
0.

75
 

- 
0.

1 
0.

4 
N

u 
1.

0 
1.

0 
0.

4 
0.

5 
0.

2 
0.

5 
- 

- 
0.

2 
   i

i. 
Ph

os
ph

or
us

 
50

 k
g 

O
 

0.
25

 
0.

5 
0.

2 
1.

0 
0.

25
 

0.
75

 
0.

1 
- 

0.
2 

ba
g 

P 
0.

2 
0.

5 
0.

2 
0.

5 
0.

1 
0.

6 
- 

- 
- 

N
u 

0.
2 

0.
4 

0.
1 

- 
0.

1 
0.

2 
0.

2 
- 

0.
1 

  ii
i. 

Se
ed

 ra
te

 
M

au
nd

s  
  

O
 

67
 

0.
13

 
13

0 
30

0 
0.

15
 

0.
9 

0.
6 

0.
1 

50
 

or
 R

s 
P 

71
 

0.
12

 
15

0 
30

0 
0.

14
 

0.
8 

0.
6 

0.
1 

50
 

N
u 

53
 

0.
13

 
15

0 
30

0 
0.

11
 

0.
9 

0.
7 

0.
1 

50
 

  iv
. I

ns
ec

tic
id

e 
R

s  
O

 
14

 
11

.0
 

50
 

- 
25

 
1.

0 
- 

- 
- 

P 
18

 
10

.5
 

50
 

- 
13

 
- 

- 
- 

- 
N

u 
7 

11
.2

 
50

 
- 

17
 

- 
- 

- 
- 

   v
. C

an
al

 w
at

er
 

A
cr

e-
fe

et
 

O
 

1.
5 

1.
5 

2.
0 

1.
5 

0.
9 

0.
6 

0.
2 

0.
1 

0.
4 

P/
N

u 
1.

2 
1.

2 
1.

5 
1.

2 
0.

7 
0.

4 
0.

2 
0.

1 
0.

3 
  v

i. 
Tu

be
w

el
l d

el
ta

 
A

cr
e-

fe
et

 
O

 
2.

0 
2.

0 
1.

1 
0.

7 
0.

3 
0.

5 
- 

- 
0.

5 
P 

1.
0 

1.
5 

1.
0 

0.
6 

- 
0.

3 
- 

- 
0.

3 
 3.

 Y
ie

ld
 p

er
 a

cr
e 

M
au

nd
s  

  
O

 
59

5.
0 

32
.1

 
23

.6
 

24
50

.0
 

16
80

.0
 

9.
3 

26
.3

 
9.

3 
9.

1 
60

0.
0 

or
 R

s 
P 

48
5.

0 
29

.3
 

22
.9

 
25

73
.0

 
15

95
.0

 
8.

5 
21

.7
 

10
.4

 
8.

9 
60

0.
0 

N
u 

31
5.

0 
21

.4
 

18
.7

 
21

38
.0

 
10

30
.0

 
9.

2 
18

.5
 

10
.8

 
9.

7 
60

0.
0 

  So
ur

ce
:  

W
A

PD
A

, "
Pr

iv
at

e 
Tu

be
w

el
l a

nd
 F

ac
to

rs
 A

ff
ec

tin
g 

C
ur

re
nt

 R
at

e 
of

 In
ve

st
m

en
t,"

 A
nn

ex
ur

e 
IV

-3
.  

19
80

. 
* 

  O
 =

 T
ub

ew
el

l o
w

ne
r, 

P 
= 

Tu
be

w
el

l w
at

er
 p

ur
ch

as
er

 a
nd

 N
u 

= 
N

on
-u

se
r o

f t
ub

ew
el

l w
at

er
. 



  
 
 

38

under water-consumptive crops, and the gross income per acre achieved by tubewell 

water purchasers more closely approximated that of farmers who only received canal 

water than that of tubewell owners, even though their cash and labor inputs were virtually 

as high as those of tubewell owners (Table 12). 

In part, water buyers may have lower cropping intensities and yields than 

tubewell owners because buyers choose to use less water due to the cost of purchased 

tubewell water.  However, the price of water for tubewell purchasers in the IFPRI sample 

was not much greater than the cost to tubewell owners (except for owners of electric 

pumpsets with fixed electricity charges, who face a very low marginal cost and therefore 

have an incentive to pump as much water as can be used).  Whether tubewell water 

purchasers use less groundwater based on an input allocation decision or supply 

constraints is unclear.  As discussed below, there are numerous occasions on which 

tubewell water is not available to water buyers at any cost, despite their demand for it.  

The lower reliability of purchased tubewell water compared to owned tubewell water is 

likely to be a major contributor to any yield or income gap between tubewell owners and 

water purchasers.  Renfro (1982:83) concludes that, in comparison with water purchasers, 

"obviously actual sampled tubewell owners can exert more control over water supplies 

with favorable impacts on productivity." 

The differences in cropping pattern between tubewell owners, water purchasers, 

and non-users of tubewell water are not as clear in the IFPRI sample 

villages as in the WAPDA (1980) and Renfro (1982) studies.  Table 13 examines 

the cropping pattern in Faisalabad and Dir districts by source of irrigation.   

 



  
 
 

39

Table 12--Input use and agricultural productivity in Renfro's (1982) study  
 Canal Water Tubewell Tubewell 
 Only Buyers Owners Total  
Gross crop income 3018  3475  4659 3297 
(Rs/acre) (1081) * (1632) * (2029) (1453) 
 
Canal water use/ 26.3  26.2  25.2 26.0 
acre (acre minutes) (9.5)  (5.6)  (6.7) (9.2) 
 
Tubewell water use/ 0.0  14.2  31.4 7.9 
acre (acre minutes) (0.0) * (13.3) * (21.9) (14.9) 
 
Cash input expenditure 309  385  388 344 
(Rs/acre) (156)  (158)  (86) (198) 
 
Labor use 73.8  76.2  75.5 74.0 
(mandays/acre) (37.8)  (35.4)  (46.4) (37.3) 
 
Cropping intensity 160  168  184 164 
(percent) (25)  (28)  (23) (26) 
 
Percent high water 35  36  45 36 
using crops (17)  (22)  (20) (19) 
 
Sample size 69  50  10 129 
  
Source:  R. Z. H. Renfro.  "Economics of local control of irrigation water in 
            Pakistan:  A pilot study".  (Ph. D. diss., Colorado State University, Fort 
           Collins, CO, 1982). 
Note: Figures in parentheses are standard deviations. 
* Difference between categories significant at 0.05 probability level. 
 
 
In Faisalabad, the average total cropping intensity for tubewell owners, water purchasers 

and non-tubewell users is 176 percent.  In Jaranwala, where there is no alternative 

irrigation source, tubewell water purchasers have a slightly higher cropping intensity than 

owners (182 compared to 176 percent--though the differences are not significant).  In Dir, 

tubewell owners have the highest average cropping intensity (197 percent), followed by 

non-tubewell users (171 percent) and water purchasers (167 percent). 



  
 
 

40

The proportion of area under individual crops shows no clear pattern across the 

sample.  For example, tubewell owners in Faisalabad plant more fodder, wheat, and 

sugarcane than other crops, while wheat is the major crop for water purchasers, followed 

by fodder and sugarcane.  Non-tubewell users in Faisalabad plant the greatest proportion 

of their land in sugarcane, an extremely water-consumptive crop, followed by wheat.  In 

Jaranwala, both tubewell owners and water purchasers plant a greater amount of fodder 

than do farmers in other Faisalabad villages.  In Dir, tubewell owners plant 62 percent of 

their land in fodder, while water purchasers (who are largely tenants) plant only 5 percent 

and non-tubewell users plant 21 percent.  Dir farmers, particularly tubewell owners and 

water purchasers, also plant a large area under wheat and maize, as well as substantial 

area under tomato and onion.  Further analysis of the net returns to total farm operations 

would be necessary to understand the relationship between crop choice and type of 

irrigation. 



 
41

   T
ab

le
 1

3-
-C

ro
pp

in
g 

pa
tt

er
n 

by
 a

cc
es

s t
o 

tu
be

w
el

l w
at

er
, F

ai
sa

la
ba

d 
an

d 
D

ir
 D

is
tr

ic
ts

 
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

  P
er

ce
nt

 o
f T

ot
al

 A
cr

es
 U

nd
er

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

To
ta

l 
 

 
 

 
 

 
 

 
 

 
 

O
th

er
 

O
th

er
 

C
ro

pp
in

g 
 

W
he

at
 

R
ic

e 
Su

ga
rc

an
e 

Fo
dd

er
 

M
ai

ze
 

Pu
ls

es
 

C
ot

to
n 

O
ils

ee
ds

 
To

m
at

o 
O

ni
on

 
V

eg
et

ab
le

s 
C

ro
ps

 
In

te
ns

ity
 

  Fa
is

al
ab

ad
 - 

To
ta

l  
Tu

be
w

el
l o

w
ne

r 
59

 
2 

30
 

70
 

1 
0 

4 
0 

0 
0 

8 
0 

17
5 

   
(n

=2
2)

 
(2

4)
 

(6
) 

(3
1)

 
(4

4)
 

(4
) 

(0
) 

(1
1)

 
(0

) 
(0

) 
(0

) 
(2

1)
 

(0
) 

(2
2)

 
 Tu

be
w

el
l w

at
er

 b
uy

er
 

71
 

2 
25

 
49

 
8 

0 
15

 
0 

0 
0 

5 
0 

17
6 

   
(n

=4
6)

 
(1

9)
 

(7
) 

(2
4)

 
(3

1)
 

(1
8)

 
(0

) 
(2

0)
 

(1
) 

(0
) 

(0
) 

(1
7)

 
(1

) 
(2

7)
 

 N
on

-tu
be

w
el

l u
se

r 
48

 
1 

68
 

42
 

8 
1 

5 
1 

0 
0 

3 
0 

17
7 

   
(n

=2
4)

 
(3

0)
 

(4
) 

(4
7)

 
(4

7)
 

(1
1)

 
(5

) 
(1

0)
 

(4
) 

(0
) 

(0
) 

(1
4)

 
(0

) 
(3

7)
 

 Ja
ra

nw
al

a 
Tu

be
w

el
l o

w
ne

r 
57

 
4 

26
 

88
 

0 
0 

0 
0 

0 
0 

0 
0 

17
6 

   
(n

=1
5)

 
(2

8)
 

(7
) 

(3
6)

 
(4

0)
 

(0
) 

(0
) 

(0
) 

(0
) 

(0
) 

(0
) 

(0
) 

(0
) 

(2
1)

 
 Tu

be
w

el
l w

at
er

 b
uy

er
 

71
 

7 
26

 
76

 
0 

0 
0 

0 
0 

0 
0 

2 
18

2 
   

(n
=5

) 
(2

2)
 

(1
5)

 
(4

3)
 

(3
4)

 
(0

) 
(0

) 
(0

) 
(0

) 
(0

) 
(0

) 
(0

) 
(3

) 
(3

0)
 

 Fa
is

al
ab

ad
 o

th
er

 V
ill

ag
es

 
Tu

be
w

el
l o

w
ne

r 
61

 
0 

37
 

32
 

3 
0 

14
 

0 
0 

0 
26

 
0 

17
3 

   
(n

=7
) 

(1
3)

 
(0

) 
(2

0)
 

(1
8)

 
(7

) 
(1

) 
(1

5)
 

(0
) 

(0
) 

(0
) 

(3
3)

 
(0

) 
(2

5)
 

 Tu
be

w
el

l w
at

er
 b

uy
er

 
71

 
2 

24
 

46
 

9 
0 

17
 

0 
0 

0 
6 

0 
17

5 
   

(n
=4

1)
 

(1
9)

 
(5

) 
(2

1)
 

(3
0)

 
(1

9)
 

(0
) 

(2
1)

 
(1

) 
(0

) 
(0

) 
(1

8)
 

(0
) 

(2
7)

 
 N

on
-tu

be
w

el
l u

se
r 

48
 

1 
68

 
42

 
8 

1 
5 

1 
0 

0 
3 

0 
17

7 
   

(n
=2

4)
 

(3
0)

 
(4

) 
(4

7)
 

(4
7)

 
(1

1)
 

(5
) 

(1
0)

 
(4

) 
(0

) 
(0

) 
(1

4)
 

(0
) 

(3
7)

 
 D

ir 
- A

ll 
V

ill
ag

es
 

Tu
be

w
el

l o
w

ne
r 

50
 

0 
0 

62
 

49
 

1 
0 

0 
17

 
19

 
0 

0 
19

7 
   

(n
=4

) 
(4

1)
 

(0
) 

(0
) 

(9
3)

 
(5

6)
 

(2
) 

(0
) 

(0
) 

(3
3)

 
(3

2)
 

(0
) 

(0
) 

(7
) 

 Tu
be

w
el

l w
at

er
 b

uy
er

 
44

 
1 

0 
5 

40
 

5 
0 

4 
30

 
23

 
5 

11
 

16
7 

   
(n

=8
) 

(2
5)

 
(5

) 
(0

) 
(8

) 
(4

0)
 

(1
0)

 
(0

) 
(9

) 
(3

3)
 

(3
5)

 
(1

0)
 

(1
3)

 
(4

1)
 

 N
on

-tu
be

w
el

l u
se

r 
64

 
3 

0 
21

 
48

 
4 

0 
2 

16
 

8 
0 

6 
17

1 
   

(n
=7

3)
 

(3
5)

 
(1

4)
 

(0
) 

(3
8)

 
(3

8)
 

(1
0)

 
(0

) 
(7

) 
(2

6)
 

(1
8)

 
(1

) 
(1

8)
 

(4
2)

 
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

   
   

 
So

ur
ce

:  
IF

PR
I s

ur
ve

y 
da

ta
, 1

99
0-

19
92

. 
N

ot
e:

  F
ig

ur
es

 in
 p

ar
en

th
es

es
 a

re
 st

an
da

rd
 d

ev
ia

tio
ns

. 



   
 
 

42

In order to examine the impact of different sources of irrigation on yields, a 

production function was estimated using plot-level data for the 1990-91 agricultural year.  

Due to limited degrees of freedom for other crops in the sample, this equation was 

estimated for wheat only, a staple crop grown by nearly all farmers in both Faisalabad 

and Dir.  Table 14 shows the results of this estimation, using linear regression.21  Seeding 

rate, total fertilizer applications (defined as kilograms of elemental Nitrogen and 

Phosphorous per acre), total labor inputs, soil characteristics of pH, potassium, 

phosphorous, and salinity are included in the model along with the number of irrigation 

applications from own tubewells, purchased groundwater, and canal water.  Soil pH has 

been transformed to degree of alkalinity, a variable computed by subtracting 7 from to 

the original pH value.22   Salinity is represented by a dummy variable indicating if 

measured electrical conductivity levels are greater than or equal to 4 millimhos per 

centimeter, the threshold level for average crop tolerance of salinity. 

                                                 
     21 Alternative functional forms, such as Cobb-Douglas and log-linear, were tested, but did not fit the 
data as well as linear regression.  The large number of cases with values of 0 for one or more of the 
independent variables, notably the irrigation inputs, may account for the poor fit of the Cob-Douglas 
equation.  Quadratic and interaction terms were tested, but found not significant, and are therefore not 
included in the model. 
     22 Original pH on sample plots ranges from 7.0 (neutral) to 8.5 (alkaline). 



   
 
 

43

Table 14--Effect of irrigation applications on plot-level wheat yields in Faisalabad 
             and Dir Districts  
 
Independent  Standard T Variable 
Variables Coefficient Error Statistic Mean  
 
Seeding rate (kg/acre) 3.60 ** 1.82 1.98 40.93 
Fertilizer (N + P kg/acre) 4.67 ** .73 6.09 47.12 
Labor (person-days/acre) 1.89 * 1.12 1.68 27.86 
Degree of alkalinity (adjusted pH) 253.54 ** 69.20 3.66 0.62a 
Soil Potassium (ppm/acre) 1.25 ** .38 3.32 128.14 
Soil Phosphorous (ppm/acre) -4.50 4.82 -.93 10.07 
Soil salinity dummy -44.61 68.99 -.65 .13b 
Canal irrigations 31.14 ** 9.00 3.46 2.53 
Purchased tubewell irrigations 44.58 ** 16.66 2.68 .62 
Own tubewell irrigations 48.31 ** 18.45 2.62 .50 
Constant -63.16 112.20 -.56 
 
 
 Adjusted R Square            =   0.31  **   
 Number of observations      =  263.0                                     
Source: IFPRI survey data, 1990-1992. 
a Degree of alkalinity = soil pH - 7.0. 
b Variable equals 1 if electrical conductivity > 4 mmhos/cm, 0 otherwise. 
** Significant at 5 percent probability level 
* Significant at 10 percent probability level 
 
 
 

The seeding rate and amount of nitrogen and phosphorous fertilizer have a strong 

positive effect on wheat yields.23  Total labor inputs, including family, exchange, 

permanent hired, and casual hired labor are also associated with higher yields. The level 

of potassium in the soil and degree of soil alkalinity are positively associated with yields.  

Higher pH values influence yields because slightly alkaline soils (those with a pH above 

7.0) are characterized by greater nitrogen, phosphorous, and potassium availability.  The 

                                                 
     23 A single variable for the sum of nitrogen and phosphorous fertilizer is used because the levels of these 
two inputs are multicollinear.  If N and P are included as separate variables in the model, both have 
significant coefficients of approximately the same magnitude as the total fertilizer coefficient (4.4) in the 
final model. 



   
 
 

44

coefficients for other soil characteristics--phosphorous content and salinity are not 

significant.   

The lack of a significant effect of electrical conductivity on wheat yields is 

important because in areas of tubewell irrigation, secondary soil salinity induced by large 

amounts of groundwater use is a potential concern (Murray-Rust and Vander Velde 

1992).  Present levels of salinity, which average 0.8 millimhos per centimeter in Dir and 

3.2 millimhos per centimeter in the Faisalabad study area do not appear problematic, but 

higher levels may reduce productivity.   

After controlling for fertilizer input and soil fertility, all three types of irrigation 

inputs had a significant positive effect on wheat yields.  But the magnitude of the 

coefficients indicates that each irrigation application from own tubewells has the highest 

impact on yield, followed by purchased groundwater and canal applications.  

The number of applications is an imperfect indicator of irrigation, because it does 

not control for the volume of water used per application (though the area of crop irrigated 

is controlled for), nor for timing of applications.  The volume of water per application is 

usually lower for tubewell than for canal applications, and therefore would not explain 

the higher productivity of groundwater irrigation.  However, farmers have relatively little 

control over timing under warabandi rotations of canal systems.  Tubewell water can be 

adjusted to the crop needs and growing cycle, and therefore have a greater impact on 

production.   

The estimates of productivity of tubewell irrigation, particularly from own 

tubewells, are influenced by a number of tubewells in Jaranwala, a village within the 

formal command area of the canal system, but which actually receives little or no canal 



   
 
 

45

water (except through groundwater recharge).  While it may be argued that this is not 

representative, it is precisely in such areas that tubewell irrigation is likely to be most 

important in agricultural production. 

This production function, which focuses on seeds, fertilizer, labor, soil fertility, 

and irrigation, does not include all influences on productivity, such as weather, seed 

variety and quality, human and physical capital, and even water quality.  There are also 

selection processes underlying farmers' decisions to invest in tubewells and to purchase 

tubewell water, in addition to factors influencing whether farmers have access to canal 

water, which could be included in the model.  Furthermore, these relationships may differ 

under other agroecological conditions, such as areas with significant salinity problems.  

However, these results show that irrigation, and especially tubewell irrigation, has a 

strong impact on yields among sample farmers in Faisalabad and Dir Districts.  At the 

same time, they point to a productivity gap between the effect of own tubewell water, 

over which farmers have considerable control, and purchased tubewell water, over which 

farmers have less control.  The following section examines factors which affect the 

reliability of purchased tubewell water, in order to identify ways to improve the 

performance of water markets. 

 

6.  RELIABILITY OF IRRIGATION SERVICE  

The productivity of irrigated agriculture is not determined solely by the amount of 

irrigation water supplied.  The timeliness and reliability of water supplies is also critical.  

Timing waterings to meet crop evapotranspirative demand has a direct impact on yield, 

while the confidence farmers have in their water supply can affect their crop choice, level 



   
 
 

46

of fertilizer and labor use, and the application of other inputs.  Few studies of water 

markets have addressed the timeliness and reliability of purchased irrigation services, 

especially because they are difficult to quantify and measure.  However, availability of 

water throughout the season, especially at critical times, provides one indicator of 

irrigation service. 

Because tubewell water is not tied to a fixed warabandi rotation schedule, it is 

easier to match tubewell irrigations to crop needs, in order to provide more frequent 

irrigations during periods of peak demand, if necessary.  Tubewell water is also available 

throughout the year, except during periods of mechanical breakdown.  In the IFPRI 

sample, tubewell owners reported that pump or engine failures made groundwater 

unavailable for an average of 2 weeks per season in Faisalabad, and 1 week per season in 

Dir.  This compares favorably with the reported unavailability of canal water for an 

average of 4 weeks per season in Faisalabad and 5 weeks per season in Dir. 

Although the reliability of irrigation service under private tubewells is generally 

higher than under public sources such as canals and government tubewells, it is likely to 

be lower for water purchasers than for farmers with their own wells because tubewell 

owners sell surplus water after meeting the needs of their own crops.  Thus, the deficits 

created by shortages of groundwater or energy supplies are not shared equally between 

owner and purchaser, but rather reduce groundwater availability to purchasers first.  Such 

groundwater shortages compound groundwater unavailability due to mechanical failure 

of pumps, the latter affecting both owners and purchasers. 

Tubewell water sellers and purchasers in the IFPRI sample were asked whether 

water was always available when requested, as an indicator of reliability.  Farmers 



   
 
 

47

responded that water is most likely to be unavailable during periods of electricity 

shortage or load shedding, and during periods of peak water demand.  Not surprisingly, 

the water purchasers were more likely to identify problems with water availability than 

were the water sellers (Table 15).  Over a fourth of Faisalabad water buyers reported that 

they were unable to purchase water to meet crop needs during times of load shedding, 

although no sellers reported being unable to sell because of load shedding.  Times of peak 

demand are more problematic: nearly a fourth of all sellers and over half of all buyers 

reported that purchased tubewell water was not always available when needed during 

such periods. 



  
48

  T
ab

le
 1

5-
-R

ep
or

te
d 

av
ai

la
bi

lit
y 

of
 p

ur
ch

as
ed

 tu
be

w
el

l w
at

er
 

  Pu
rc

ha
se

d 
Tu

be
w

el
l 

   
   

 F
ai

sa
la

ba
d 

   
   

   
   

   
   

D
ir 

   
   

   
 

   
   

   
 T

ot
al

   
   

  
W

at
er

 U
na

va
ila

bl
e 

Se
lle

rs
 

B
uy

er
s 

Se
lle

rs
 

B
uy

er
s 

Se
lle

rs
 

B
uy

er
s 

D
ur

in
g:

 
R

ep
or

tin
g 

R
ep

or
tin

g 
R

ep
or

tin
g 

R
ep

or
tin

g 
R

ep
or

tin
g 

R
ep

or
tin

g 
   El

ec
tri

ci
ty

 lo
ad

 
0 

22
 

0 
0 

0 
22

 
   

sh
ed

di
ng

 
(0

.0
) 

(2
6.

5)
 

(0
.0

) 
(0

.0
) 

(0
.0

) 
(2

2.
4)

 
 Pe

ak
 d

em
an

d 
se

as
on

s 
1 

54
 

2 
1 

3 
55

 
 

(1
4.

3)
 

(6
5.

1)
 

(3
3.

3)
 

(6
.6

) 
(2

3.
1)

 
(5

6.
1)

 
 To

ta
l s

am
pl

e 
si

ze
 

7 
83

 
6 

15
 

13
 

98
 

  So
ur

ce
:  

IF
PR

I s
ur

ve
y 

da
ta

, 1
99

2.
 

N
ot

e:
  F

ig
ur

es
 in

 p
ar

en
th

es
es

 a
re

 p
er

ce
nt

ag
es

. 



  
 
 

49

What influences the reliability of purchased irrigation water?  Three factors can 

be hypothesized to have an effect: type of tubewell, characteristics of the buyer, and the 

relationship between seller and buyer.  Electric tubewells are more susceptible to power 

outages, and are therefore likely to be less reliable.  Larger-capacity tubewells and those 

which draw water from deeper levels are hypothesized to be more reliable.  Buyers with 

higher social status, indicated by land ownership and age, are also hypothesized to have 

more reliable access to purchased tubewell water.  If social ties influence reliability of 

water markets, farmers who buy water from close relatives or their landlords would be 

expected to receive more reliable irrigation service. 

A logistic regression model has been used to test these hypotheses, using buyers' 

reported availability of tubewell water whenever needed as an indicator of reliability.  

Dummy variables for Jaranwala village in Faisalabad and for Dir district are included to 

control for agroecological differences between these areas and the rest of the sample 

(most notably the differential availability of canal irrigation).    

Results of the logit model are given in Table 16.  As predicted, electric tubewells 

are significantly less reliable than those with diesel or tractor-powered lifts.24  Larger-

diameter tubewells were significantly more reliable than smaller ones, but deeper 

tubewells do not provide more reliable irrigation for purchasers--indeed, the coefficient is 

negative, though not significant.  Deeper tubewells may be located in groundwater-scarce 

areas, and therefore provide less reliable supplies to purchasers. 

                                                 
     24 In an alternate specification of the model, the difference was found between diesel and tractor 
tubewells was tested and found not significant. 



  
 
 

50

Table 16--Results of logistic regression model for reliability of purchased tubewell 
             water  
 
Independent    Wald 
Variables Units Coefficient T Ratio Statistic  
 
Electric tubewell dummy -2.189 ** -2.934 8.617 
Tubewell diameter inches 1.267 ** 2.368 5.587 
Tubewell depth feet -.026 -1.245 1.553 
Size of land ownership acres .072 ** 1.986 3.949 
Age of household head years .045 * 1.844 3.398 
Buy from close relative dummy .855 .866 .750 
Buy from landlord dummy 10.613 .435 .189 
Jaranwala village dummy .668 .726 .527 
Dir district dummy 4.209 ** 2.397 5.745 
Constant  -5.929 * -1.838 3.378 
 
 
 Model Chi-Square           =  58.2  **  with 9 degrees of freedom 
 Number of observations    =  96.0 
 
 
 Classification Table for Reliability of Purchased Tubewell Water 
 
  Predicted  Percent  
 NOT RELIABLERELIABLECorrect 
 
 Observed 
   NOT RELIABLE 49  6 89.1 
                     
   RELIABLE 9  32 78.5 
 
                                        Overall  84.4 
  
Source: IFPRI survey data, 1992. 
**  Significant at 5 percent probability level 
*   Significant at 10 percent probability level 

 

The amount of land owned has a significant and positive effect on reliability, 

suggesting that water sellers are less likely to deny requests for water from larger land-

owners than from small landowners or landless tenant cultivators.  This is due to the 

influence of the land owners, not the size of farm operated.  An alternate specification of 

the model with operational holding, rather than land ownership, showed no significant 



  
 
 

51

coefficient for size of holding.25  The age of purchasers' head of household, another 

indicator of status, has a positive effect on reliability.  The model does not show kinship 

or landlord-tenant relationships between water buyer and seller to have a significant 

effect on reliability of access to groundwater.  Jaranwala village does not differ 

significantly from other areas in reliability of purchased tubewell water, but buyers in Dir 

district reported significantly more reliable access to groundwater.  This is somewhat 

surprising because irrigation is less available in Dir than in the canal-irrigated areas of 

Faisalabad, and therefore demand for groundwater would be expected to be higher.  

However, rainfall is higher in Dir, and therefore groundwater may be more readily 

available to those farmers needing supplemental irrigation. 

While buyers' reported problems with unavailability of purchased groundwater is 

an imperfect indicator of reliability, this model points to important sources of problems in 

groundwater markets.  Purchasers are more likely to receive insufficient groundwater if 

they buy from small-capacity, electric-powered tubewells; if they are young and own 

little or no land; and if they live in Faisalabad District.  Improving the reliability of 

electric power or switching to diesel pumps are the most readily identifiable interventions 

to improve reliability of groundwater markets (as well as reliability of water for well 

owners).  Both of these options are, however, expensive.  Identifying the factors which 

lead to lower reliability in Faisalabad than in Dir requires further study.   

                                                 
     25 Both ownership and operational holding size could not be included because of multicollinearity 
problems. 



  
 
 

52

Although land ownership and age have strong influences on reliability, they do 

not appear to point to policy interventions that can improve reliability.  Improving the 

reliability of irrigation by increasing the land ownership and age of water-purchasing 

households is not feasible, or even desirable.  However, it may be possible to raise the 

status of purchasers relative to water sellers by encouraging smaller farmers to purchase 

tubewells.  This study does not have data on both the sellers' and buyers' characteristics 

for each relationship, but it is possible that farmers with less land will provide more 

reliable service, both because there is less of a status gap between them and the 

purchasers, and because tubewell owners with less land will not need as much water to 

meet irrigation needs on their own fields, and thus have surplus water available for sale. 

 

 7.  RESEARCH AND POLICY MEASURES FOR WATER MARKET 
DEVELOPMENT  

Water markets are largely autonomous, indigenous institutions which function--

and are likely to continue functioning--without a great deal of official intervention.  What 

type of attention, if any, should the government, researchers, and other agencies pay to 

water markets?  This final section highlights policy instruments and remaining research 

questions pertaining to water market development. 

First, understanding the role water markets play in mediating access to and 

control over groundwater resources can assist tubewell development programs in serving 

a larger number of farmers.  This study indicates that water markets improve the equity of 

groundwater use by increasing the access of small farmers, tenants, and younger 

households who are least likely to own tubewells.  Thus, neither public tubewells nor 



  
 
 

53

ownership of tubewells by all farmers are required to ensure widespread use of 

groundwater in areas where water markets operate.   

However, the study also demonstrates that water markets do not operate in all 

areas, raising the question of which factors influence the activity of water markets.  

Conditions that affect the amount of groundwater available for sale include the number of 

tubewells in an area, the proportion of tubewell owners willing to sell water, and the 

nature of the underlying groundwater resource.  Because, with current technology, water 

cannot be transported over large distances, wells must be available within close proximity 

to buyers' fields for groundwater markets to operate.  Chambers, Saxena and Shah (1989) 

suggest that a higher density of tubewells fosters the development of competitive water 

markets.  Thus, programs which assist private tubewell development, especially in areas 

of adequate, good-quality groundwater, are also likely to assist water market 

development. 

Making information about water markets more widely available both to 

individuals and groups of farmers may also stimulate further investment in tubewells.  

For example, knowledge of the potential for groundwater sales in a given area, different 

credit options for investment and contractual arrangements for water sales could be an 

important decision-making tool for potential investors.  This information could be 

gathered and disseminated by local groups, non-governmental organizations or projects 

such as the Private Tubewell Development Project. 

Due to the small sample of tubewell owners, this study was not able to determine 

factors affecting which tubewell owners will sell water.  Possible factors influencing the 

decision to sell include the size and location of tubewell owners' land holdings and their 



  
 
 

54

cropping patterns, as well as tubewell technology and energy costs.  Further research on 

this topic, as well as on the nature of the specific transactions occurring between buyers 

and sellers, would be useful in stimulating water markets. 

As noted above, large farmers are most likely to own tubewells, while small 

farmers are more likely to depend on water purchases, and consequently face less reliable 

access to groundwater.  Howev