. - ‘ .A: :‘ ', J-.v.“.. -
.IhLIL éul::‘ ! .: z : H l 5.
‘.. ,1

-.J

Y
I'I‘t‘r " u

  
   
  

V ‘ . . fig“.
“W ' Randy 1:
W33 1

 

' 1211.2». :13
' ‘ 3'r41':24"'3 - ' I
JE‘iW'm‘

'i-b, ‘
pmaf §

"1.” . .

n -:n

:3] “Nu .‘
' " L' “z".

V_F-

a.

v
-...
_,...¢..
l\‘-"V-‘ . -

w..—---

-—-~l-— 5-

' I
a.‘

“NW":

‘..
3.2};

u «r
>1»!an ‘

b
:31
-..-
"a."

V a ‘ ; - . _ .V
,“f - r: '1 . . r
.‘ . ....
. w .. _. .» .
' . ' w" ‘ _
.u. ; -
A m... - -

v _._.... L.’
W.-.”
. n... .‘
“5:1 “wily...“

‘ mm...

 

thm‘fm

5::

I.-v.-u "
“ 2 .
. a a. . '
- _; '~ ‘ ‘. . m.
I v' u
v D A ' '
. W, .L. .
. ’u‘n _
u . u
». m.

.m
I '

I ‘ '.
, i“_:’£ '. "
‘ . “j 9' ‘2'“; ‘v in: V1
‘ «1?§§i{§1f*' 1m:
,z 4* ‘

5:143‘

7 In. fl

 

. :44:-
F .
1.1-44— . . A .
1 ':
”mw
..
avg-:3.
. _
-‘ .. .rz.
3:3:— *

"u.
up“. .
v»

 

,.-u-m
‘0.—
.nth‘
”3..
v .

.£ . '1‘
"'3’1' W33,” '

J
I.’ . n ' WA
1' I“ l: H !
1222*:21W:
Lb” ‘ ‘

r

z}!
[’1
..

w-u

11‘
a
i- In “Em
‘ IN.'~Z'T.E‘.‘:;:'1:';W
it":

 

$3?“

-:

ZZZ-'3'

1&'"'."
1:25
wnvfitfi‘ A
3m:

 

mama».
1"“ "E"
-
‘ , n‘.‘ I: l ._
.. _ . 9:: -n‘fl‘-P’. -
"‘“m u“.
at: w
u.
I: ”a." a.
m -

.‘uw..., _
4‘ "
«a...
w
r--.»
.m-
“- ~
kg A .- V A w,
)fi . ‘
. I I.
A .

.
a...

-..L..“..m.~
“.9“?
“A . ”W

W... “1.. :-~ .

a.

'3
A .
ahlh

I y. .
.u‘
. 7'!"-

-.‘ 'qu F‘
v *3. 'L'
- 1. 1.:

,._.,_.-. . ,.. .4.

y-.‘.’ Q ..,I.- "L —

“-1....4 I ~mwup~
v‘ v.,-- t—- . .. .-

»

 

 

 

 

- .. ,_-‘p.--—r

 

 

H.
I:"¥”":
‘ l.

>_. .—--.r——-.

.~ .0...”- --

.9

.1
.

. puln". .
“-9— .

:-
".4

t .'

up

. -,
119'! .
‘J :M,

H};

-~
~ \
m.,.....

-._.--'r y».
"w...” 60."- a

_.‘ ..

_.,___ .

=25

up“

..

 

‘gzur.’

y.

. I}
; :1.‘
:

—...
..- . .
-3?» .

wag.
$3

. .—
A “w... Aw4.-

.‘X',

‘4 [if
I V

‘ pififlullfv
‘ | :

\
‘

4

.
Hr
0'

. < .14
o.‘~.-3. v—ow-é-wn

m

- ._
.. ..
.4,“

.... .4._‘

1g!

’.
u _'

l
&

, I v
. V
:g_ :3.
L -1331 —

 

1945315 , lilllllllllllllllllllllHlllllllllllllllllllllllllllllllllll

:2 302080 6091

 

 

 

 

2 V 0 ,
LIBRARY
Michigan State
University
This is to certify that the

dissertation entitled

CONTRACT THEORY AND ITS APPLICATION TO
CROWNDWATER MARKETS IN INDIA

presented by

Kei Kajisa

has been accepted towards fulfillment
of the requirements for

Ph. D. degree in Agricultural Economics

W

Major p'rofessor

 

[/0 / $3797

Date

 

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINE return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

m: i % 29°}
am 1 5 2003

 

 

 

 

Eflgtifl

 

 

 

 

 

 

 

 

 

 

 

11/00 c/CIRCIDateDuepss-m4

CONTRACT THEORY AND ITS APPLICATION TO GROWNDWATER MARKETS
IN INDIA

By

Kei Kajisa

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Agricultural Economics

1999

ABSTRACT

CONTRACT THEORY AND ITS APPLICATION TO GROWNDWATER MARKETS
IN INDIA

By

Kei Kajisa

This dissertation explores the mechanisms of contract choice and price determination
under bilateral transaction between an owner of irrigation systems and a non-owner at
groundwater markets. A sample of owners and non-owners in six villages in Madhya
Pradesh in India is used for empirical analysis. The analyses of contract choice show that
risk-sharing is the factor underlying the choice of sharecrop contract over either fixed-
rent or flat-charge contract. The analyses of price determination show that the unit water
price becomes higher under sharecrop contract since the water buyers pay a risk premium
to the sellers when risk is transferred from the buyers to the sellers by sharecrop contract.
The other notable result of price analyses is that in contrast to the conventional notion of
seller’s monopolistic pricing, the market structure rather works in favor of the buyers in
that they are charged no greater than the cost of irrigation. Thus, the water markets in the
study area are described such that while the water prices are not exploitative, in risky
environment sharecrop contract is chosen with a payment of a risk premium from the

buyers to their sellers.

C0pyright by
Kei Kajisa
1 999

To my parents

iv

ACKNOWLEDGMENTS

I would first and foremost like to express my gratitude and sincere appreciation to
my major professor Thomas Reardon for his unwavering support, guidance and
encouragement. I would also like to thank Dr. Takeshi Sakurai, who allowed me to join
his research project, for having the patience and generosity to guide me through entire
process of the research project beginning from designing the questionnaire, conducting a
field survey, to analyzing corrected data. I remain indebted the other members of my
committee, Professors John Strauss, Allan Schmid and Scott Swinton, given that I have
benefited enormously from their erudition. I also acknowledge funding this study from
the Global Study Program through the National Institute for Environment Studies,
Tsukuba, Japan and partially from the National Research Institute of Agricultural
Economics, Tokyo Japan. This paper would not have been possible without the
collaboration with faculty members and their associates of the Indian Institute of Forest
Management: P. Bhattacharya, Sangeeta Chourasia, Shailendra Das, Debashis Debnath,
Neeta Dubey, Hilaluddin, K. Krishna Kumar, Shailes Kumar, Ram Prasad, Kalpna
Rawat. I am gratefully acknowledge to my friends at Michigan State University,
especially Nazmul Chaudhury, Janet Owens, and Takashi Yamano who are the comrades
in a workshop which is informally organized by ourselves to discuss issues of
development economics. Lastly, I would like to especially thank to my parents and

dedicate this work to them.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ........................................................................................................... ix
INTRODUCTION ............................................................................................................... 1
CHAPTER 1
CHARACTERISTICS OF GROWNDWATER MARKETS IN SURVEY AREA ............ 3
CHAPTER 2
CONTRACT CHOICE AT GROWNDWATER MARKETS ............................................ 8
2.1 Introduction ............................................................................................................... 8
2.2 Methodological Issues ............................................................................................ 10
2.3 Variable Construction and The Hypotheses ........................................................... 11
Household Wealth ............................................................................................... 12
Management Ability ........................................................................................... 13
Monitoring Cost .................................................................................................. 15
Remarks on Flat Charge Contract ....................................................................... 18
2.4 Econometric Results ............................................................................................... 19
Comparison ofThree Types of Contract ....................................... 19
Comparison between Sharecrop and Non-sharecrop Contracts .......................... 20
2.5 Village-level Determinants ..................................................................................... 21
2.6 Concluding Remarks and Policy Implication ......................................................... 23
CHAPTER 3
PRICE DETERMINATION UNDER BILATERAL TRANSACTION ........................... 30
3.1 Introduction ............................................................................................................. 30
3.2 Structure of Water Markets in the Survey Area ...................................................... 32
Monopoly Power ................................................................................................. 32
Individual Price Variations ................................................................................. 33
3.3 The Model ............................................................................................................... 35
3.4 Variable Construction ............................................................................................. 40
3.5 Regression Results .................................................................................................. 44
Market Structure ................................................................................................. 44
Determinants of Unit Water Price ....................................................................... 46
3.6 Concluding Remarks and policy Implications .................................................................
SUMMARY AND CONCLUSION .................................................................................. 56
APPENDIX A
REVIEW OF LITERATURE ON CONTRACT CHOICEAT LAND RENTAL
MARKETS ........................................................................................................................ 59
Risk-sharing Model ....................................................................................................... 59

vi

Limited Liability Model ................................................................................................ 60

Screening Model ........................................................................................................... 61
Double-incentive Model ............................................................................................... 62
Monitoring Cost Model ................................................................................................. 64
APPENDIX B
SUPPLAMETNARY TABLE FOR CHAPTER 2 ............................................................ 68
APPENDIX C
CALCULATION OF THE WATER-CHARGE-COST RATIO AND THE RETURN TO
IRRIGATION INVESTMENT .......................................................................................... 69
APPENDIX D
THE QUESTIONNAIRE ................................................................................................... 71
BIBLIOGRAPHY .............................................................................................................. 92

vii

LIST OF TABLES

1-1 Characteristics of Irrigation in six Villages ................................................................... 7
1-2 Types of Water Contract across Villages ...................................................................... 7
2-1 Descriptive Statistics of Variables ............................................................................... 26
2-2 Multinomial-logit Regression Analysis of Determinants of Contract Choice ............ 27
2-3 Logit Regression Analysis of Determinants of Contract Choice ................................ 28
2-4 Types of Water Contract and Selected Characteristics Across Villages ..................... 29
3-1 Ratio of Water Charge to Irrigation Cost and return to Capital Investment in Different
Case Studies ....................................................................................................................... 52
3-2 Price Variations within and across Villages by Types of Input Goods ....................... 53
3-3 Descriptive Statistics of Variables ............................................................................... 54
3-4 OLS Regression Analysis of Determinants of Unit Water Price ................................. 55
B-l MultinomiaI-logit Regression Analysis of Determinants of contract Choice

(Constrained Models) ......................................................................................................... 68
C-1 The Water-charge-cost Ratio and the Rate of Return to Irrigation Investment .......... 69

viii

LIST OF FIGURES

A-l Transaction Cost and Optimal Contract Choice ......................................................... 67

ix

INTRODUCTION

Incompleteness and imperfection of markets are salient features in less developed
economies. Some markets such as credit and contingent markets are missing, and thus an
opportunity of transactions may be limited. Asymmetric information gives considerable
scope for all kinds of opportunistic behavior such as cheating, shirking, and moral hazard.
Consequently, transaction costs associated with the use of markets for exchange may
become very high and sometimes prohibitive.l In response to these circumstances,
exchanges may be limited to a face-to-face informal contract as opposed to the open
market transaction.

The most famous example in agrarian economies is the existence of various
arrangements of land-tenancy and labor-employment contracts (Hayami and Otsuka,
1993; Bardhan, 1989, 1984). Unfortunately, however, it is not well known how similar
informal contracts are observed for groundwater markets between owners of private
irrigation systems and non-owners. As well as land tenancy, the transactions of
groundwater markets carry interesting features which we cannot observe under perfectly
competitive markets (Saleth, 1998; Meinzen-Dick, 1996; Fujita and Hossain, 1995;
Kahnert and Levine, 1994; Janakarajan, 1993; Shah, 1993; Pant, 1991). First, different
types of contracts coexist. Second, variations in water price are very large not only
across villages but also across individual pairs of owners and non-owners. These two
characteristics of groundwater markets provide us with an opportunity to tackle the long-

standing research issues in the field of the economics of contracts: what are the economic

 

' Transaction costs typically involve the cost of information, search, negotiation, screening, monitoring,
coordination, and enforcement.

factors underling the choice of a particular type of contract; what are the determinants of
price under informal contracts. Yet, the groundwater markets have not systematically
been analyzed within the framework of contract theory.

The research objective of this paper is to explore theoretically and empirically the
mechanism of contract choice and price determination of groundwater markets. The
theoretical argument in this work is based on the economics of contracts. For empirical
study, I use our original data set from India. India provides an interesting case because
private irrigation systems have been rapidly proliferating since the cut back of public
investment in large scale canals in the late-19605, and consequently ground water
markets have been emerging in many villages (Rosegrant, 1997). I begin with an
analysis of contract choice in the following chapter; chapter 4 explores water price
determination under different types of informal contracts; chapter 5 synthesizes the

previous chapters' conclusions and policy implications.

Chapter 1

CHARACTERISTICS OF GROWMDWATER MARKETS IN SURVERY AREA

The six villages chosen for this study are located in two adjacent districts,
Hoshangabad and Narshimhapur, in Madhya Pradesh. In these villages, a research
project on an integrated analysis of natural resource management was conducted from
December 1997 to April 1988. The number of water buyers in each village is not so large
since most of the farmers own tubewells (Table 1-1). Among 569 households in the
surveyed villages, there are 41 water sellers and 79 water buyers (Table 1-1). The
number of sellers is smaller than the number of buyers since some sellers sell water to
multiple buyers. We also confrrmed who sells to whom in order that the samples became
matched data set.

The surveyed villages are homogeneous in terms of agronomic condition. The
climate is characterized as monsoon with deep-black soil.2 The farmers in the area
cultivate soybeans during rainy season (Kharif) without relying on irrigation, and they
cultivate mostly wheat and sometimes chickpeas with electric pumps and tubewell
irrigation systems during dry season (Rabi).3 The household samples I use in this study
are 26 sellers and 38 buyers who cultivate wheat.

All water sellers are self-cultivators using their own tubewells, and there is no
pure-seller who owns tubewells but do not farm by himself. Sellers are usually sell water

after meeting the needs of their own fields, and thus it is claimed that sellers are selling

 

2 There is no major difference in terms of annual rainfall, but it increases gradually as we move from west
to east.
3 Sugarcane is cultivated through a year but it is the minor activity in the study villages.

“surplus” water (Shah, 1993). All tubewells for irrigation are powered by electric pump
and there is no diesel pump in our survey field. By the government policy the electric
charge is free in village A, B and C. In village D, E, and F, if pump is smaller than 5
horsepower the well owners can use electricity for free, but if larger, the well owners
have to pay fixed amount of electricity charge by season.

All the buyers are also self cultivators and thus there is no specific case in the
sample wherein the landlords provide water to the tenants as a part of interlinked land
tenancy. In villages A, D, E and F, some farmers who do not own tubewells irrigate their
land by water from stream (Table 1-1). Among 38 water buyers, there is no buyers who
answered that they would irrigate their land by stream if current seller refused to sell
water, but 8 buyers answered that they would buy water from another tubewell owner,
and 30 buyers answered they have no water source other than the current seller. This
indicates that easiness of water access varies among buyers.

Although a wide variety of contracts exist in Indian water markets, the contracts
we observed in the study area are classified into three: sharecrop, fixed-rent, and flat-
charge contract.4 Under sharecrop contract, buyers and sellers share buyers' harvest for
water charge; under fixed-rent contract, water-buyers pay a fixed amount of cash for

certain acreage of irrigation for a season; under flat-charge contract, the buyers can

 

‘ Besides these three types of contracts, Shah (1993) observes several different types in Andra Pradesh in
India. First is labor contract under which the buyer provide labor and draft power to the seller in return for
water. Second is crop and input sharing contract under which the seller provides water, and shares buyer’s
input costs and the harvest (interlinked contract). Fujita and Hossain (1995) observed the contract called
“chaunia” in Bangladesh under which the well-owner rents in land during dry season from the landlord
who does not own irrigation systems. We observe no such contracts as those observed by Shah; there is
only one case in our data set which resembles a "chaum‘a" contract.

irrigate as much as they want at a given per-acre-water—price.5 Although each contract is
observed 17 %, 13 % and 8 % respectively over the study area, distribution of three types
of contracts in each village is skewed (Table 1-2). We observe all three types only in
village C and D. I talk about the methodological issue caused by this in next chapter.

In order to clarify the terms of contract, I have a few remarks on linkages of water
contract with other inputs. First, all water buyers cultivate their own land, and thus there
is no specific case in our sample wherein the landlords provide water to the tenants as a
part of interlinked land tenancy.

Second, there might exist linkages with other inputs such as animal, tractor or
labor. Water buyers might rent capital inputs (animal or tractor) from sellers, or buyers
might provide labor works to sellers as a part of payment. Unfortunately I do not have
this kind of information in our data set. In our sample, however, the farmers pay village-
homogeneous rental fee to capital inputs, and village-homogeneous daily-wages to hired
labors. Thus, even if buyers use seller's animal or provide labor works to sellers, there
seems not to be interlinked with water transaction. As for seed, fertilizer and pesticide,
all sample households purchase them at either public or private store, and thus there is no
possibility of interlinked contract through these inputs.

Third, there are several cases in our sample wherein sellers do not allow buyers to
operate irrigation systems (pump transportation, pump driving, and channel digging).
The salient feature of these cases is that all these sellers also do not handle their irrigation
systems by themselves but hire an irrigation manager who conducts all irrigation

activities. Since these hired managers are also in charge of buyers' irrigation operations,

 

5 A unit of measurement is the area irrigated. This is common in the area where farmers use electric
pumps, and electricity is provided for free or fixed amount payment per season, while hourly payment is

buyers do not operate irrigation systems by themselves. I discuss about this in next

chapter.

 

common for diesel pumps since diesel consumption is linked to hours of operation.

Table l-1: Characteristics of irrigation in six villages

 

A

<Total>

 

Number of ag household

Number of well-owner-
cum-non-sellers

Number of well-owner-
cum-water-sellers

Number of water buyers

Number of households
irrigated by stream

Irrigated agricultural
households (%)

70

40

71%

Village
C D
93 100
74 49
7 8
12 1 5
O 18

100% 90%

560
311
41
79
41

83%

 

Table 1-2: Types of Water Contract across Villages

 

A

Total

 

Number of Contract Observed in Village

(3) Sharecrop
(2) Fixed-rent

(3) F lat-charge

l

l

0

Village
C D
3 4
6 1
2 2

17
(45%)
13
(34%)
8
(21%)

 

Village total

11 7

38

 

Chapter 2

CONTRACT CHOICE AT GROWNDWAER MARKETS

2.1 Introduction

It is well known that there exist variety types of contracts for land tenancy all over
the world. A number of researchers have theoretically and empirically investigated the
relevant economic factors underlying the choice of a particular type of contract over
others.6 Unfortunately, however, the same amount of effort has not been devoted to the
analysis of the contractual choice for groundwater markets, despite the existence of
different forms of contracts. Moreover, few studies pay attention to the uniqueness of
groundwater market contracts

A unique phenomenon at groundwater markets is that we observe not only
sharecrop and fixed-rent contracts but also flat-charge contract under which water fee is
charged either hourly basis or acreage basis, and payment is done when water buyers use
irrigation systems (Saleth, 1998; Meinzen-Dick, 1996; Shah, 1993). Different from land,
un-bulky input like irrigation water can be sold according to volume. Still a certain
portion of sellers and buyers choose fixed-rent payment. There may exist economic

reasons underlying the choice of flat-charge contract over fixed-rent contract. On the

 

6 For review of theoretical achievements, see Hayami and Otsuka (1993) and Singh (1989). For
econometric empirical works on contract choice at land rental markets, see Tunali (1993), Allen and Lueck
(1992), Rozenzweig (1988), Datta et. al. (1986), Alston et. al. (1984). For descriptive empirical works, see
Sharma and Dreze (1996), Roumasset and Uy (1987), Robertson (1982), Bliss and Stern (1982), R30
(1971)

other hand, two contracts can be practically indifferent if rescheduling of fixed-rent
contract is allowed when unexpected events happen after they enter into the contract. In
this regard, the issues of contract choice at groundwater markets are not only an
identification of the determinants but also a clarification of the reasons of existence of
flat-charge contract. Yet these issues have not been addressed systematically considering
the past theoretical achievement of contract choice, nor adequately tested by data from
water markets.

The purpose of this paper is to explore the mechanisms of contract choice among
sharecrop, fixed-rent and flat-charge contracts at groundwater markets for a sample of
wheat-cultivating water sellers and buyers in six villages in Madhya Pradesh, India.
Recent theoretical treatments regard an incidence of a particular type of contract is a
response to risk, limited access to contingent market, input market imperfections,
asymmetries in information (Hayami and Otsuka, 1993; Singh, 1989). I examine whether
these factors significantly affect farmers' contractual choice among three options,
especially between fixed-rent and flat-charge contracts. The sample I use has a notable
advantage for this purpose; I can match water sellers and buyers, so that I can observe
both parties' characteristics. Thus, my econometric results do not suffer the omitted
variable problem which is due to using only one party's characteristics in an analysis of
two parties' choice problem.

Section 2.2 is devoted to the empirical strategy; Section 2.3 presents variable
construction and hypotheses; Section 2.4 and 2.5 present analytical results quantitatively
and qualitatively; Section 2.6 highlights the main conclusion and suggests relevant policy

implications.

2.2 Methodological Issues

An empirical strategy commonly used for multiple choices is either the ordered
logit or the multinomial logit model. According to the "agricultural ladder" argument,
sharecrop contract should be ranked lower than fixed-rent contract since sharecroppers
are regarded as more dependent on their renter in terms of agricultural management
(Hayami and Otsuka, 1993; Spillman, 1919). While this argument is widely accepted in
literature on land tenancy, we are not sure a priori where we can rank flat-charge contract
which is not observed in land tenancy. Thus, I use multinomial logit model for choice
among three types of contracts in this paper. I also perform likelihood ratio test in order
to check whether fixed-rent and flat-charge contracts are statistically indifferent, and then
if so, I conduct logit analysis for choices between sharecrop contract versus fixed-rent
and flat-charge contracts.

Another feature of the empirical strategy in this paper is that I include
characteristics of both sellers and buyers as explanatory variables. Since choice of
contract involves both parties, my econometric results suffer less from the bias caused by
problems associated with omitted variables than the empirical literature in the past that
uses either sellers' or buyers' characteristics alone.

The last is about treatment on village-fixed effects. Much of contractual
difference is undoubtedly due to complexity of historical, sociological, and political
factors peculiar to different places. These factors are difficult to be observed and thus
usually assmned to be captured by village dummies in cross-section analysis.

Unfortunately, however, as we mentioned in chapter 1, due to zero incidence of a

10

particular type of contract in several villages in our data set, I have to give up using
village dummies in my econometric analysis. The omission of village dummies would
result in potential bias in estimates, even if the six villages are located in agriculturally
homogeneous area. For compensation of this shortcoming, I also provide descriptive
analysis in section 2.5, where I investigate the relationship between dominant contract

forms in the villages and village level characteristics.

2.3 Variable Construction and The Hypotheses

In this section I list covariates I use in this paper and hypothesize them based on
the models presented in the literature on contract choice (for the detail of each model, see
Appendix A). However, flat-charge contract is not discussed there simply because land
tenancy contract is the literature's main interest and there is no flat-charge in land tenancy
contracts. Hence, I first review how tenancy literature discusses the choice between
sharecrop and fixed-rent, and then make remarks on peculiar character of flat-charge
contract at the end of this section. I divide covariats into variables that attempt to
measure household wealth, management ability, alternative opportunity, and factor
related with moral hazard. The descriptive statistics of covariates are presented in Table

2-1.

11

Household Wealth

Wealth plays an important roll in two fundamental models of contract choice: ,
risk-sharing and limited liability models. The former recognize contract as a risk-sharing
device in environments characterized by missing contingent markets and asymmetries in
information (Stiglitz, 1974). Sharecrop contract has benefit of transferring risk from
buyers to sellers, while it could cause the Marshallian inefficiency problem. On the other
hand, fixed-rent contract let buyers shoulder the entire risk without Marshallian
inefficiency problem. Given this tradeoff, benefit from sharecrop contract would become
bigger as buyers get more sensitive to the risk and sellers get more willing to bear some
of it. Assuming risk-averse is decreasing in wealth, the larger the buyers' wealth, the
more likely is the buyers to shoulder the risk and thus the more likely fixed-rent contract
is to be chosen, whereas the converse holds for seller's wealth.

Limited liability explains the relationship between wealth and contract choice in a
different way (Sengputa, 1997; Basu, 1992; Shetty, 1988). Due to non-existence of
contingent markets in an agrarian economy, liability of buyers' payment is limited by
their initial wealth. Sellers will know that the default on fixed-rent may occur when the
harvest fails and buyers are poor, while the possibility of complete default is lessened by
sharing the harvest. Hence, for buyers with large wealth, fixed-rent contract would be
more likely to be chosen.7 Meanwhile, sellers' wealth does not matter in this argmnent
since sellers are receivers of the payment.

The wealth-variables I use in my empirical analysis are buyers' and sellers' last
year's animal holdings (the number of oxen plus cows) and total land holdings (acre).

These variables may not be truly treated as exogenous because farmers adjust these

12

variables in their long term decision process. However, they can be still treated as
“quasi”—exogenous for cross-section estimation process. Land holdings are seldom
adjusted. The last year’s ntunber is given to the current year activity, although the

number of current year's animal is a part of the entire production decision.

Management Ability

I use management ability in my regression for three reasons. First, the risk-
sharing argument supports the inclusion of this variable. Management ability is
considered to be closely related with vulnerability toward risk in production. The agent
can make the risky environment less risky if he has the management ability. Thus, the
higher buyers' (sellers') ability is, the less likely buyers (sellers) hesitate to shoulder the
risk by fixed-rent contract (by sharecrop contract), resulting in getting positive (negative)
sign on buyers' (sellers') ability variables for the choice of fixed-rent over sharecrop
contract.

Second, importance of managerial inputs in contract choice is also pointed out by
Eswarn and Kotwal (1985). It has been noted that resource owners often provide
managerial inputs to sharecroppers, while fixed-rent farmers are more independent in
terms of their farm management. If water buyers have little experience in irrigation
agriculture, they would expect suggestion from water sellers who have longer experience
of irrigation agriculture and know better about their irrigation systems. On the other
hand, if buyers are already experienced, they would rather choose fixed-rent contract to
manage their cultivation by themselves. Thus, the higher the buyers' ability, the smaller

the benefit from sharecrop contract, while the larger the sellers' ability, the larger the

 

7 This hypothesis holds without assuming risk-averse agents (see review in Appendix 2 for the detail).

13

benefit of receiving their managerial input through sharecrop contract. The other way
round, I predict positive (negative) sign on buyers' (sellers') ability variables for the
choice of fixed-rent over sharecrop contract.

Third, the model called screening model claims that sellers can design a menu of
contracts such that high ability buyers will prefer fixed-rent contract and low ability
buyers will prefer sharecrop contract (N ewbery and Stiglitz, 1979; Hallagan, 1978).
Sellers have incentive to do this since they cannot directly observe buyers’ true ability
and since buyers might mimic themselves as low ability in order to be charged low
prices. Thus, buyers' management ability would be positively correlated with the
incidence of fixed-rent contract, while sellers' management ability does not matter in this
argument. The model is, however, criticized that the ability is generally revealed sooner
or later in small rural community (Singh, 1989; Eswarn and Kotwal, 1985). Thus, the
sign on buyers' ability variable may not be explained by this logic. There is also
difficulty in empirical procedure since true ability is very difficult to measure correctly.
One way to circumvent this problem is to find a good proxy, but if good proxy exists,
sellers screen buyers by that but not by a menu of contracts, (so that sellers do not have to
propose the menu of different types of contracts). Nevertheless, the sign hypothesized by
this model does not conflict with those derived from previous two explanations, and thus
my prediction does not change.

The management ability variables I use are buyers' and sellers' schooling years
and age of household head, which are assumed to measure managerial knowledge and

experience. Arguably, these variables are not current year's choice variable.

14

Monitoring Cost

Monitoring cost is inevitable when there is asymmetry of information between
sellers and buyers and thus there is a possibility of cheating in production activities
(Allan and Lueck, 1995, 1992; Roumasset and Uy, 1987; Datta et. al., 1986; Alston et.
al., 1984). For example, labor-effort must be monitored if there is an incentive to Shirk;
usage of irrigation systems must be monitored if there is an incentive to overuse them;
sharing the harvest also must be monitored if there is a possibility of underreporting. A
number of researchers claim that monitoring of these activities is costly and that a
particular type of contract is chosen to minimize the sum of these monitoring costs.

Under sharecrop contract, if monitoring of buyers' labor-effort is very difficult,
buyers put less labor-effort than the optimal since they receive a portion of their marginal
products (the Marshallian inefficiency). On the other hand, since the entire harvest is
obtained by buyers, fixed-rent contract entails no cheating in labor-effort. Hence, given
other things being equal, the more costly it is to monitor labor-effort, the more likely a
fixed-rent contract is offered. Moreover, the more labor-intensive crop cultivation is, the
larger the potential losses from cheating, and thus the more likely fixed-rent contract is to
be chosen.

Sharecrop contract, however, has advantage in reducing the possibility of buyer’s
overuse of irrigation systems since increment of outputs achieved by overuse must be
shared with sellers. Meanwhile, under fixed-rent contract buyers may overuse sellers’
irrigation systems since the entire harvest goes to buyers. Thus, the more difficult to

monitor irrigation overuse, the less likely fixed-rent contract is to be chosen. Sellers

15

would be concerned about buyers' overuse if their irrigation systems are vulnerable and
precious. Thus, the larger the potential damages on irrigation systems, the less likely
fixed-rent contract is to be chosen.8

The variables related with monitoring in our data set are the total amount of
investment to irrigation systems, whether seller and buyer belong to the same caste
group, whether a buyer has alternative water seller, and whether buyers operate irrigation
systems by themselves.9 Regarding the amount of investment, the more expensive the
systems, the larger the potential damages by buyers’ overuse, resulting in low possibility
of fixed-rent contract.

The dummy for caste matching has multiple effects. If both parties belong to the
same caste group, it might be easier to build trust between them. Buyers’ overuse of
sellers’ irrigation systems is less likely to occur, resulting in high possibility of fixed rent
contract. Meanwhile, cheating in labor-effort may be also less likely to occur within the
same group, and thus fixed-rent is less likely to be chosen. Hence, the sign on this
variable is an empirical question.

The dummy for existence of alternative water seller may or may not have effect
on buyers’ cheating activities. The buyer who has alternative seller may have larger
incentive to do cheating activities since he can go to another seller in case of rejection

from current seller afier a disclosure of his cheating activities. If so, this dummy, which

is related both with labor-effort shirking and irrigation overuse, has multiple effects as

 

8 Monitoring the harvest-sharing arouses cost, too. Buyers know exact amount and quality of their harvest
since they harvest by themselves, but sellers have to guess it. If the cost of physical measurement and
division of harvested crop is very high, buyers would have incentive to underreport their harvest to the
sellers. Thus, the more costly the harvest sharing, the more likely fixed-rent contract is to be chosen. Since
I use the samples of wheat cultivator, difficulty of measurement does not vary across samples. Thus, we do
not discuss the monitoring cost aroused by harvest-sharing.

16

caste matching dummy does. In a small village, however, loosing reputation would
reduce the opportunity of transactions with other villagers. The buyer may not do
cheating activities regardless of the existence of alternative seller. Thus, the empirical
question is whether this dummy becomes significant or not, and if significant, how the
existence of alternative buyer effect incentive of cheating and then contract choice.

I also allow for dummy whether buyers’ irrigation operation is conducted by the
irrigation manager who is hired by sellers. If so, sellers do not have to worry about
buyers' overuse of their irrigation systems even under fixed-rent contract. Hence, this
dummy may have positive sign for the choice of fixed-rent over sharecrop contract.
However, as I mentioned in chapter 1, managers operate buyers’ irrigation only when
sellers hire managers for sellers' own irrigation activities. Moreover, there is neither the
case such that sellers themselves operate inigation systems for buyers nor the case such
that sellers hire someone just for buyers irrigation operation, which would have been
observed if sellers just wanted to reduce the possibility of irrigation overuse by buyers.
Hence, the variation of this dummy may not due to the reason claimed by monitoring cost
model but probably due to management inertia, resulting probably in undeterministic sign
on this dummy.

Before concluding this subsection, it is better to mention about potentiality of the
endogenous variable problem of the dummy for hired irrigation manager. Since this is a
part of production decision, estimates would be biased. Given limited information in our
data set, however, I can neither adequately address the nature of potential bias in the

parameter estimates due to this problem, nor can I adequately instrument these

 

9 Total investment to irrigation systems consists of the cost of well digging, electric pump, and sprinkler
systems.

17

endogenous variables. Alternatively, I run regressions without these dummies to see

robustness of the results.

Remarks on F lat-charge Contract

There are differences and similarities between fixed-rent and flat-charge contract.
The most remarkable difference is the nature of risk each party faces. Sellers receive
certain amount of payment from buyers for sure under fixed-rent contract, while under
flat-charge contract sellers may not be sure since buyers can stop buying water from him
in the middle of the season if they get enough amount of irrigation water by rain. Thus,
sellers also face the risk, but the nature of that risk is different from the one transferred
from buyers to sellers under sharecrop contract which depends on buyers production
outcomes but not on buyers decisions about buying or not buying. Therefore, flat-charge
contract may have different coefficient on the household wealth and management ability
which are assumed to be related with the attitude toward risk. Besides, seller will not
worry about buyers' default (limited liability) under flat-charge contract since payment is
done when buyers use irrigation systems. This may also change the coefficient on
buyers' wealth.

The structure of monitoring cost does not vary between fixed-rent and flat-charge
contract. Under either contract, buyers have incentive to overuse irrigation system unless
detected, while there is no possibility of cheating of labor-effort. Thus, coefficient on the
variables related with monitoring cost may not be different.

Furthermore, there seems to be no practical differences between two contracts. In

land tenancy contract, it is observed that landlords reschedule fixed-rent contract when

18

ll

unexpected events happen (Ishikawa, 1975). If the amount of fixed-rent can be
rescheduled according to the actual frequency of water application, fixed-rent contract
becomes akin to flat-charge contract. Sellers and buyers would not strictly distinguish
two contracts. Therefore I have no clear-cut hypothesis as to flat-charge contract, but
have empirical question of whether fixed-rent and flat-charge are practically different,
and if so how wealth and ability variables have different effect on the choice of fixed-rent

over flat-charge contract.

2.4 Econometric Results

Comparison of Three Types of Contract

The results of multinomial logit analysis on three types of contracts are given in
Table 2-2. The comparison group is sharecrop contract. Model 1 is without dummy for
hired manager and Model 2 is with it. I perform the likelihood ratio tests to examine
whether all coefficients are equal across fixed-rent and flat-charge contracts (x2 statistics
and associated p-values are in the table). The null hypotheses is that all coefficients
across fixed-rent and flat-charge are same but different from sharecrop contract. The test
results indicate that I cannot reject the null hypotheses at any acceptable significant level
(p-values are 0.43 and 0.49).lo Given these test results, I should not call the difference
between fixed-rent and flat-charge significant. As I talked in the previous section, sellers

and buyers in our study area do not practically distinguish the two contracts. They

19

probably answered the closest option between fixed-rent and flat-charge, while the actual
payment schedule may be more flexible depending on contingent events. Meanwhile, the
farmers still distinguish sharecrop from non-sharecrop contracts. I discuss this issue in

next subsection.

Comparison between sharecrop and non-sharecrop contract

I combine samples of fixed-rent and flat-charge contract into one group. The
results of logit regression analysis on choice of fixed-rent and flat-charge contracts over
sharecrop contract are given in Table 2-3. Since the qualitative results are not altered
dramatically between Models 1 and 2, I will confirm my discussion to the model with the
dummy for hired manager.

The variables significant at 10% level are seller’s animal holdings and schooling
years which have negative effect on the choice of fixed-rent and flat-charge contracts,
consistent with my conjecture from risk-sharing model and division of managerial input.
Allen and Lueck (1992, 1995) claim that empirical evidences do not support risk-sharing
model. However, their argument is based on the data from the United States of America
and mainly from non-agricultural sectors where contingent markets are relatively better
organized and people are more tolerant to risk than any agrarian economies in less
developed countries. My results tell us that risk-sharing can still be a motivation of
choosing a particular type of contract in agrarian economy in developing countries. On

the other hand, the limited liability model is not supported by our data since the effect

 

’0 First, I run multinomial logit regressions with the constraint such that null hypothesis is true (see
Appendix B for the results). Then, likelihood ratio statistics are calculated from unconstrained and
constrained results.

20

from buyer 's wealth is statistically very weak. This is probably because the possibility of
complete default at water markets can be circumvented by flat-charge contract.

The factors affecting monitoring cost do not have very strong statistical
relationship. Thus, cheating activities in terms of labor effort and irrigation overuse are
not serious problems in a small community where people know each other and reputation
plays important role in transaction. The statistical relationship of manager dummy is also
very weak, implying that the reason of managers' operation for buyers is just due to

management inertia.

2.5 Village-level Determinants

I will explore qualitatively the village-level determinants underlying contractual
choice. The village-level effects would be clearly recognized by focusing on two
extremes, namely the villages B and E where only fixed-rent-and-flat-charge contracts or
sharecrop contract exists in large (Table 2-4). The village characteristics I use are
weather shock, credit access, remoteness, and health condition presented from rows 4 to
10 in Table 2-4.

The weather shock is measured by the number of drought in the last ten years,
district-level means and standard deviations of rainfall from 1901 to 1950. The
remarkable feature is that weather shock is harsher for the village B than for the village
E. The village B had experienced droughts, having lower mean and higher standard

deviation of rainfall (rows 4, 5, and 6). However, the village B seems to be reckless of

21

this climatic risk as nobody chooses sharecrop contract. One potential reason is that all
sample farmers are irrigated and thus are not vulnerable to weather shock. Another
potential reason is the existence of well-firnctioning credit and asset markets for
consumption smoothing. In drought years, the suffered farmers in the village B either
borrowed money or sold their asset to compensate their losses (not shown in Table 2-4).
As for credit markets, they claimed that 90 percent of villagers have access to formal
credit in the village B (row 7 in Table 2-4). On the other hand, only 47 percent of the
villagers have access to formal credit in the village E, implying that means of relief are
very limited.

In general, the accessibility to the means of relief is also measured by the
remoteness of the village. The remoteness, which is measured by distance from the block
headquarter (row 8), distance from the nearest agricultural market (row 9), and
accessibility to vehicles (row 10), tells us that the village E is not only located the farthest
from the block headquarter but also seasonally limited to the access to vehicles. While
the village B is not the least remote village, the stretch to the city is not so long since it
still has access to vehicles all seasons.

In sum, while the village B faces risks in agricultural production due to unstable
weather condition, their annual income would be rather stable since they have access to
several means of relief such as credit and asset markets. One the other hand, the village
E faces limited access to the means of income smoothing. Therefore, it is natural to
conjecture that the village-level factors associated with risks have significant effects on
contractual choice. Thus, omission of village dummies in regression analysis would

cause the omitted variable problem; all coefficients might be biased. Nevertheless, it is

22

consistent with village-level explanation in this section that the risk covariates in
regression analysis (i.e. relative animal holdings and schooling years) have strong
statistical relationship with contractual choice. Getting individual and village level
analysis together, what I can still insist is that risk-sharing motivation is significantly

affecting contractual choice.

2.6 Concluding Remarks and Policy Implications

In this paper, I explore the factors underlying the contractual choice at the
groundwater markets for a sample of wheat-cultivating water sellers and buyers in the six
villages in Madhya Pradesh, India. In the study area we observe three types of contracts:
sharecrop, fixed-rent, and flat-charge contracts. A notable feature of water markets is the
existence of flat-charge contract which is not used for land tenancy. However, a
flexibility of informal contractual agreement in agrarian economy makes us wonder
whether the villagers practically distinguish flat-charge and fixed-rent contracts. The
results from multinomial logit regression analysis show that there is no statistical
difference in the choice between flat-charge and fixed-rent contracts. This is probably
because it is allowed to reschedule fixed-rent payment according to the actual frequency
of irrigation. Unfortunately, our data set does not have the information on contractual
rearrangement; further detailed investigation of this issue would be required. Meanwhile,
the results’obviously show that the farmers distinguish sharecrop contract and non-

sharecrop contracts.

23

The logit regression analyses are conducted to quantify the determinants
underlying the choice between these two options. Given the availability of variables, the
explanations I employ are risk-sharing, limited liability, managerial input sharing,
screening, and monitoring cost models. Putting individual-level econometric analysis
and village-level qualitative analysis together, what I can safely state is that risk-sharing
is among the motives of choosing sharecrop contract in the study area. While Allen and
Lueck (1992, 1995) rejected risk-sharing motivation by using the data from United States
of America, it is still an important determinant of contract choice in developing countries
where contingent markets are not adequately developed. The econometric analysis shows
that sharecrop is more likely to be chosen when sellers own animals (i.e. being less risk-
averse), and when sellers are educated (i.e. having risk management ability). The village-
level analysis shows that sharecrop contract is skewed in the village that faces limited
access to means of income smoothing.

Given imperfection of contingent markets and asymmetries in inforrnaion,
sharecrop contract is a "locally optimal" outcome. However, since there may exist the
Marshallian inefficiency problem under sharecrop contract, the "local optimum" does not
necessarily correspond with the " global optimum" which could have been achieved under
perfect markets and no asymmetries in information. The empirical literature on land
tenancy shows the existence of inefficiency under sharecrop contract in India (Shaban,
1987; Bell, 1977). If there exists the Marshallian inefficiency also at groundwater
markets under sharecrop contract, policy makers had better to consider how to reduce the
production risk in villages and also to increase access to means of risk hedge. Further

detailed investigation of production environment and detection of inefficiency would help

24

guide policy makers in designing appropriate strategies to obtain "global optimum" at

groundwater markets in Indian villages.

25

Table 2-1: Descriptive Statistics of Variables

 

 

Variable Mean Standard Deviation
Household Wealth
Buyer's last year's animal holdings 2.26 1.81
Seller's last year's animal holdings 3.13 1.66
Buyer's land holdings (acres) 5 .07 4.79
Seller’s land holdings (acres) 9.63 9.95
Management Ability
Buyer’s schooling years 5.58 4.99
Seller's schooling years 4.32 4.29
Buyer's age 44.42 11.01
Seller's age 45.97 15.04
Factor related with Moral Hazard
Total amount of investment (000Rs.) 6.08 3.87
Caste matching dummy 0.61 0.50
Existence of alternative seller (dummy) 0.24 0.43
Hired manager dummy 0.11 0.31

 

26

 

Table 2-2: Multinomial-logit Regression Analysis of Determinants of Contract Choice
(The comparison group is sharecrop contract)

 

With Hired manager dummy Without Hired manager

 

 

 

 

dummy
Fixed-rent Flat-charge Fixed-rent F lat-charge_
Household Wealth
Buyer's animals 0.112 -0.090 0.066 -0.155
(0.34) (0.19) (0.19) (0.32)
Seller's animals -1.113 -1.627 -l.214 -1.658
(1.37) (1.77) (1.38) (1.74)
Buyer’s land holdings 0.096 0.025 0.097 0.015
(0.60) (0.1 I) (0.58) (0.06)
Seller's land holdings 0.152 0.208 0.137 0.190
(1.17) (1.31) (1.02) (1.20)
Management Ability
Buyer's schooling 0.032 -0.088 0.038 —0.107
(0.24) (0.50) (0.28) (0.58)
Seller's schooling -0.307 -0.214 -0.365 -0.190
(1.56) (0.78) (1.58) (0.60)
Buyer's age -0.075 -0.075 -0.055 -0.069
(1.22) (0.86) (0.86) (0.70)
Seller's age -0.001 0.065 -0.001 0.072
(0.01) (0.83) (0.01) (0.87)
Factor related with Moral Hazard
Total investment -0.1 14 -0.469 -0.085 -0.449
(0.47) (1.51) (0.34) (1.40)
Cast matching dummy 2.088 5.360 1.598 5.010
. (0.87) (1.94) ‘ (0.65) (1.82)
Alternative seller dummy 2.603 4.371 2.601 4.171
(1.02) (1.51) (1.03) (1.47)
Hired manager dummy . . 1.872 0.748
(0.96) (0.27)
Constant 4.217 1.901 3 .854 1.892
(1.01) (0.35) (0.93) (0.33)
Likelihood ratio test
:3 statistics 11.12 11.49
(l-Io: coefficients are same) [0.43] [0.49]
Observations 38 38
x2 statistics 30.78 31.81
[0.10] [0.13]
Pesudo R2 0.38 0.39
flLikelihood -24.69 -24.17

 

The comparison group is sharecrop contract

Numbers in parentheses are absolute value of z-statistics. Numbers in brackets are p-

values.

"‘ significant at 10% level; ** significant at 5% level, *** significant at 1% level

27

Table 2-3: Logit Regression Analysis of Determinants of Contract Choice
(1 if non-sharecrop (i.e. fixed-rent or flat-charge) is chosen)

 

Without hired manager

With hired manager dummy

 

 

 

dummy
Household Wealth
Buyer's animals 0.078 0.047
(0.25) (0.15)
Seller's animals -1.304 -1.299
(1.73)‘ (l .69)*
Buyer's land holdings 0.070 0.077
(0.47) (0.50)
Seller's land holdings 0.180 0.159
(1.42) (1.26)
Management Ability
Buyer's schooling -0.003 —0.001
(0.03) (0.01)
Seller's schooling -0.297 -0.330
(1.65)‘ (1.68)"'
Buyer's age -0.076 -0.057
(1.31) (0.93)
Seller's age 0.018 0.016
(0.33) (0.29)
Factor A fleeting Monitoring Cost
Total investment -0.244 -0.204
(1.08) (0.90)
Caste matching dummy 3.315 2.823
(1.44) (1.23)
Alternative seller dummy 3.443 3.461
(1.37) (1.44)
Hired manager dummy . 1.444
(0.83)
Constant 4.405 3.808
(1.1 1) (0.96)
Observations 38 38
X’statistics 19.66 20.41
[0.050] [0.060]
Pesudo R2 0.38 0.39
Log Likelihood -16.30 -15.92

 

Numbers in parentheses are absolute value of z—statistics. Numbers in brackets are p-
values.
"' significant at 10% level; ** significant at 5% level, *** significant at 1% level

28

Table 24: Types of Water Contract and Selected Characteristics Across Villages

 

 

 

Village
A B C D E F Total
Number of Contract Observed in Village
(3) Sharecrop l 0 3 4 9 0 17
(45%)
(2) Fixed-rent l 3 6 1 0 2 13
(34%)
(3) Flat-charge 0 4 2 2 0 0 8
(21%)
Weather Shock
(4) Number of drought in the l 3 l 0 0 0 -
last ten years
(5) Mean of rainfall (mm/year) ' 1295 1295 1295 1361 1361 1361 -
(6) Standard Deviation of 297 297 297 217 217 217 -
rainfall (mm/year) ‘
Access to Credit
(7) Percentage of villagers who 100 90 100 95 47 25 -
can access to formal credit b
Remoteness
(8) Distance from block 30 25 4 13 50 17 -
headquarter (km.)
(9) Distance from the nearest 30 30 4 12 13 17 -
agricultural market (km.)
(10) Accessibility to Vehicles 8" a" all a" seasonal All -
$085011 season $685011 season . $685011

 

Notes
a District data

b Formal credit institutes includes commercial bank, cooperative bank, governmental

bank.

29

Chapter 3

PRICE DETERMINATION UNDER BILATERAL TRANSACTION

3.1 Introduction

In South Asian countries including India, private lift irrigation systems, i.e. wells,
pumps and water conveyance networks, have been rapidly proliferating since the cutback
of public investment of large scale canals in the late-19605 (Rosegrant, 1997).
Consequently, groundwater markets are emerging in villages, and we observe informal
but repeated water transactions between LIS owners and non-owners. In India, the area
irrigated through water markets is estimated to be reaching nearly 50 percent of gross
irrigated area with LIS, making significant contribution to increasing agricultural
production (Shah, 1993).

There has been concern, however, about water sellers' exploitative pricing
behavior toward water buyers, which stems from the conventional wisdom of treating
water sellers as "water-lords" (Bagchi, 1995; Campbell, 1995; Kahnert and Levine, 1994;
Pant, 1992; Dhawan, 1988). In response to this concern, a number of researchers have
investigated the relationship between sellers and buyers at groundwater markets, and they
have also addressed the factors underlying high water prices. Most of the case studies on
these issues has been in terms of village-level comparisons. A consensus that has
emerged from these case studies suggests that the water sellers' pricing behavior is not

necessarily monopolistic, but rather depends on village-level factors such as depth of

30

water table, rainfall pattern, energy source, and density of wells (Shah and Ballabh, 1997;
Fujita and Hossain, 1995; Shah, 1993). At the same time, noticing price variations within
villages, several articles have emphasized the effect of individual characteristics of sellers
and buyers on price, albeit the statistical evidences is limited (Saleth, 1998; Meinzen-
Dick, 1996; Janakarajan, 1993; Pant, 1991). In fact, as I talk later, in small agrarian
communities where both sellers and buyers know each other and usually have some
forms of personal relations, they often complete exchange through a face-to-face
transaction as opposed to an open-market transaction. Furthermore, even within the same
village, there exist different forms of contract for water transaction purpose. Thus, a
village level analysis may be inappropriate for an analysis of groundwater markets, but
rather, we should also focus on individual-specific differences. Groundwater markets
have not been analyzed either theoretically or empirically at the individual level in a
systematic way within a bilateral transaction framework that takes into account of the
differences in contracts.

The objective of this paper is to explore the seller-buyer relationship and the
determinants of water price in groundwater markets. Theoretically, I use Nash’s (1950)
two-person bargaining model, and specify a reduced form groundwater price function.
The bargaining model is chosen since it encompasses a commonly used model -- the
principal agent model -- as its special case. I also compare pricing mechanism under
different types of contracts. Empirically, I conduct regression analysis on our original
data set from Madhya Pradesh in India. A notable advantage of this data set is that it

allows us to match water sellers and buyers, so that I can involve both parties' individual

31

characteristics in the analysis, and with this new perspective, I can examine the claim that
sellers are not necessarily exploiters.

Section 3.2 provides a description of the groundwater markets in the study area
with a comparison with past empirical studies; Section 3.3 lays the formal model of
bilateral transaction; Section 3.4 and 3.5 present variable construction and econometric
results; Section 3.6 highlights the main conclusions of this study and suggests relevant

policy implications.

3.2 Structure of Water Market in the Study Area

Monopoly Power

One of the simple but prevalent methods of analyzing a market structure is to
estimate the price-cost ratio. Researchers use the ratio to judge whether markets are
monopolistic or competitive. Table 3-1 shows the summary of the past case studies and
estimation results from our study area. Shah and Ballabh (1997) and Shah (1993)
conclude that the water markets in areas with high ratio are not competitive but
monopolistic. On the other hand, even though Fujita and Hossain (1995) obtain high
value of the ratio (2.59 or 2.00), they regard the "rent" as return to seller’s capital
investment in irrigation systems, given that the rate of return to capital (69 %) is close to
the interest rate on the informal financial market (38 — 61 %). They conclude that if we

take into consideration the risk of investment in irrigation systems, the water price is

32

economically reasonable, and “the groundwater market cannot be characterized as
monopolistic, but rather it is competitive and efficient.”(pp. 456)

Following the method used by Fujita and Hossain (1995), I calculate the weighted
average of water-charge-cost ratio and the rate of return of irrigation investment in our
study field (see chapter 2 for general characteristic of the water markets in our study
field).ll While the ratios in our study area are relatively high (3.43 and 2.15), the
difference between the rate of return to irrigation investment (29 %) and the interest rate
of the credit program that well-owners have used for irrigation installation (12 — 15 %) is
not as large as the difference in Fujita and Hossain (1995); the difference is not large
enough to warrant concerns of monopolistic behavior. ’2 Thus, similar to F uj ita and
Hossain’s implication, the rate of return in the study area on average does not radically
exceed the interest rate at the formal credit market which the buyers would have used if
they borrowed money to have their own irrigation systems. Therefore, we may have

reason to doubt monopolistic water market in our study area.

Individual Price Variations
The absence of monopoly, however, does not necessarily imply existence of a

competitive market which the past literature implies. In fact, water prices vary across

 

” First, I calculate the ratio of per-acre water charge to per—acre operational cost for all individual pairs of
sellers and buyers, and then take average weighted on the buyer' s area of irrigation. Operational cost
compose of such item as the cost of electricity, the cost of repairs and necessary parts, and wage payments
to labors. Wages are paid for such works as excavating channels, operating pumps, and supervising buyer's
pump-use. A shadow wage (i.e. village wage) is used for cost calculation when sellers do these works by
themselves. The ratio of per-acre water charge to the sum of operational and depreciation costs is
calculated in the same way. The depreciation cost of irrigation investment is estimated by the constant
amount method assuming a life of 20 years of the system. The rate of retina of irrigation investment is per-
acre profit over per-acre profit, where profit is per-acre water charge minus per-acre operational cost. The
weighted average of each pair's rate of return is the number reported in the text. See Appendix C for detail.

33

pairs of sellers and buyers depending on specific characteristics and relationships
between sellers and buyers, implying existence of bilateral transaction instead of single
market in each village. Table 3-2 shows simple regression results of individual-level
input prices on village dummies. If a market is competitive, we should not see large price
variations at least within a village. As for seed, urea, super phosphate, and DAP,
significant variations in prices are explained by the village dummies (R-squared = 0.41
for seed, 0.68 for urea, 0.70 for super phosphate, 0.73 for DAP), implying that these input
prices do not significantly differ within villages. On the other hand, the regression of
water price shows that there remain huge part of unexplained variations even afier
removing the village-level variation (R-squared = 0.20). One potential reason is
existence of only one price study area with large variation due perhaps to measurement
errors. It also seems natural to conjecture that each water transaction carry a different
water price which could be explained not by village-level factors as the past literature
suggests, but by individual-specific characteristics of sellers and buyers.

In sum, the seller may be neither monopolist nor a price-taker at a competitive
market. Given this fact, now the question arises: what is the sellers' position in the water
markets and whether individual characteristics of the seller and the buyer affect the water
price, and if yes how do they determine the price. These issues need to be investigated
empirically, but prior to the econometric analysis a formal model is presented to underpin

the empirical work.

 

’2 Interest rates at informal credit market vary from 36 % to 60 % per year in our study area, which is
usually used for the relief from short term liquidity constraint such as wedding and purchase of modem

34

3.3 The Model

I assume an economy consists of two individuals: a water seller (who owns
irrigation systems and provides water for production) and a water buyer (who has no
irrigation systems but provides other inputs such as land and labor). There is uncertainty

in production; the state of nature is represented by the random variable, 0, which is

|_.r.r al.].

treated as a multiplicative factor distributed with E9 = 1 and finite variance. The
payment from the buyer to the seller is assumed to be expressed by the following linear

function:

 

a0f (a, x; h,) + wa

where a is a parameter representing the output share to the seller; f(.) is the buyer’s
production function which depends on the amount of irrigation water (a) and the vector
of inputs (x) given the buyer’s production technology and agrarian characteristics (hb),
and is assumed to be a concave and linearly-homogeneous with the usual properties of
positive but decreasing marginal product; w is unit water charge in case of fixed-rent or
flat charge contract.13

For simplicity the seller is assumed to bear the entire cost of irrigation while the
buyer takes care of other inputs. Hence, the seller’s and the buyer’s profit from water

transactions are respectively defined as:

 

agricultural inputs instead of long term investment.

' Under fixed-rent contract, w and a are set by the seller, and only w is set by the seller under flat-charge
contract. There is no difference between these contracts when the contractual parameters are determined
cooperatively between the seller and the buyer.

35

71's 5 a9f(a. x; hi) + W - 0(a; h.)

m, E (1 —a)6f(a, x; h) — wa — pxx
where c() is the irrigation cost function which depends on the amount of water (a) given
irrigation technology and seller’s agrarian characteristics (hs); px is the vector of input
prices; the output price is normalized to one. Following previous work on agrarian
contracts, I further assume a risk-neutral water seller and a risk-averse water buyer
(Stiglitz, 1974; Bell, 1989)14 Thus, the seller’s utility is measured by expected profit
(E 713) and the buyer’s utility is expressed as Eu( 7a,) where u ’>0 and u ’ ’<0.

Both parties do not enter into the contract unless it provides them with utility at
least equal to their reservation utility. The reservation utility is formally defined as the
utility that a party would get, in case of disagreement, from the best alternative usage of
his resource which is currently in use. Thus, the seller’s reservation utility may be a
function of the amount of water currently used by his buyer (a) and also on productivity
of the alternative way of water usage which is determined by the capacity of his irrigation
facility and ability of himself (hs). Likewise, the buyer’s reservation utility may be a
function of the amount of resources he takes care of (x) and his ability to utilize them

(hb). Due to the different attitudes toward risk, each reservation utility is written as

7r..(a,h.)

17(7—{b (x: hb))
where if, (.) is increasing with respect to a and h, for the seller, and with respect to x and

h), for the buyer.

 

'4 The data of our study field shows that the sellers are wealthier than the buyers in terms of lands, animals,
and tractors, and this holds generally in Indian water markets (Saleth, 1998). Given the widely accepted
idea of decreasing risk aversion in wealth, this assumption would not be unacceptable.

36

The principal-agent solution has been the most common framework for examining
bilateral transactions (Salanie, 1997; Hayami and Otuska, 1993). It assumes that the
principal (either the seller or the buyer) has ultimate power of bargaining, and the agent is
not free to propose another contract. However, there may be the case that both parties
have some power of bargaining and then the contract is formed through bilateral
negotiation. We should not exclude this kind of market structure a priori when we look
at newly emerging markets. Otherwise, an econometric specification would miss
necessary explanatory variables and then have model specification bias in the results.
Thus, we should start from a more general model.

The solution concept to be employed in this paper is Nash’s (1950) solution to
two-person bargaining problem.15 This solution is more general than the principal agent
approach in that it covers not only the principal-agent solution as its polar case but also
the case that both parties have some power of bargaining. Now define

F E {E 71.01.w.a,x;h.hs)-7r.(a,h.)}5 x

{E u(n1(a,w,a,x;h. hb))— 17(7r1,(x;h1,))lHS
which is simply the product of the two parties’ gains from transaction, relative to their
reservation utilities with weight, 6, where 0 S 6 S 1. Putting weight is the natural

generalization of Nash’s program when the parties have different bargaining powers

 

’5 Bell and Zusman (1976) is the first paper which introduces the bargaining solution to tenancy problems.
Bell (1989) shows more generalized model by adding fixed-rent contract option into the previous paper in
which only sharecrop and fixed-wage are contractual options. Frisvold and Caswell (1995) apply this
solution concept to water markets.

37

(Dasgupta, 1993). Assuming cooperative bargaining, Nash’s solution is the contract that

yield the greatest value of 1" subject to both parties’ incentive constraints.l6

max r = {E rho-mm}; u(.>—z7(.)}“"

a ,w,a,x

st. Err..(-) 2 7?..(-) , E u(.) 2 i7(-)
Note that when 6 = 1, the seller is assumed to possesses ultimate bargaining power and
the bargaining problem becomes equivalent to the principal-agent problem in that the
seller, as a principal, maximizes his profit such that the buyer, as an agent, receives at
least as much as his reservation profit. The converse holds when 6 = 0.

The unit water price is defined as

p. E a9f(-) , w

a
Assuming relative strength of bargaining power, 6, is exogenously determined by both
parties technologies and agrarian characteristics, namely h,, and hb, then the equilibrium-

unit-water-price becomes the function of all exogenous variables, i.e.

p..* = p..*(p.;h.,hs) (1)

This function implies that the unit-water price may vary across the pairs of the sellers and
the buyers depending on their technologies and agraian characteristics. I estimate a linear
approximation of this function in the empirical section.

Another distinctive implication from this model is that the unit-water price would
become higher under sharecrop contract. Under sharecrop contract (0 < 01 < 1, w = 0),

equation (1) is more specifically written as

 

'6 The cooperative solution implies that there is no shirking in input provision. This may not be
unacceptable assumption in the small agrarian community where people easily observe each other’s activity

38

_1_
a(.)

 

pI.(_.,..,.,=[c(.)+2r.(.)+ 5 1 {Eu(.)-r(.)}] (2)

1—6 Eu'(.)9

where the functions on the right hand side are all at equilibrium points. This equation
tells us that the seller receives the irrigation cost and his reservation profit so that he
enters the contract, plus the extra profit that is transferred from the buyer depending on
the relative strength of the seller’s bargaining power, (6/1-6). Under fixed-rent or flat

charge contract, it becomes '7

. 5 1 _
pw(ftx/flat) = [C() + is () + SEW—(5 {Eu() - “()I]

A

3
a(.) ( )

Since Eu ’0 < Eu’, the third term becomes larger under sharecrop contract.18 Intuitively
the term l/Eu( )0 (or 1/Eu(.)) converts buyer’s utility to the seller’s utility (i.e. monetary
term). '9 The inequality, Eu ’0 < Eu’, means that when the seller also shoulders the
production risk (sharecrop contract), one unit of buyer’s utility is more appreciated in
terms of money. In other words, if buyer can get the same amount of utility he is willing
to pay more when the risk is shared. Thus, the difference of the third term between
equations (2) and (3) may be considered as a risk premium payment from the buyer to the
seller in return for the partial transfer of the production risk through sharecrop contract.
Consequently, the unit water price under sharecrop contract becomes higher than under
fixed-rent or flat-charge contract, given other things being equal. This implication is

consistent with the principal-agent solution in the past literature (Hayami and Otsuka,

 

(Bardhan, 1984, chapter 7). Empirical evidence is presented by Otsuka et al (1986).

’7 Under the assumption of cooperative bargaining, amount of water also becomes choice variable. Thus,
there is no difference in solutions between fixed-rent and flat-charge. Actually, this is consistent with my
empirical analysis which shows that there is no statistical distinction whether choosing fixed-rent or flat-
charge contract (see chapter 3).

39

1993). A comparative statics with respect to p,, h,, and h), is not informative because
changes of these parameters shifts both the seller’s supply schedule and the buyer’s
demand schedule as well as their bargaining power. In the empirical analysis I allow the

data to inform us on the signs of the effects of the different parameters.

3.4 Variable Construction

The dependent variable I use is unit-water-price, which is calculated such that the
ex-post total payment for water is divided by area irrigated and by the number of water
applications, and then divided by wheat price to convert to wheat term.20 Researchers on
water markets are concerned about qualitative difference in irrigation water. The sellers
usually irrigate their own plots first and then allow the buyers to use their irrigation
systems, and thus the buyers’ irrigation may be physically constrained in terms of timing
and amount. As I explain later, I control for this as much as I can in my econometric
analysis. Besides, I do not find notable difference in water quality in our samples,
although the available information is limited.21

Turning now to the explanatory variables, for one of the seller’s characteristics,

h,, I use the total investment in irrigation systems which consists of the investment of

 

’8 Eu'G =Eu'E9+cov(u’, 9). Since u" < 0, cov(u’, 0) < 0. Since EG =1, Eu'G =Eu'+cov(u', 0) <Eu'.

'9 Note that the numerator is the seller’s marginal utility which is equal to one.

2° The farmers in our study area use water rates on an acreage basis which is common in the area irrigated
by electric pump with free or fixed power cost. Hourly bases are prevalent measurement units in areas
where people use diesel pump or electric pump with pro rata power cost (Saleth, 1998; Shah, 1993).

2' I suppose that the constraint is reflected by the number of times of water application by the buyers. The
data, however, indicates no big difference in frequency of water application among buyers. The buyers
cultivating wheat apply water 4.4 times on average (a = 1.3; min = 2; max = 6). Chick pea cultivators’
average is 2.1 (o = 0.9; min = 1; max = 3).

40

well digging, electric pump, and sprinkler systems. This variable attempts to measure the
capacity of irrigation systems such as horsepower, length of pipelines, and availability of
sprinkler systems. The capacity would determine not only the cost structure of irrigation

systems to provide one unit of water, c(.), but also the seller's reservation utility 7?, (.) ,

because it also constrains the number of potential buyers that the seller can supply.
Another variable I use for h, is the total area irrigated by others. One well
irrigates multiple plots. One of them is obviously the buyer’s plot. The others are the
plot irrigated by the seller himself and sometimes by other buyers. For example, let's
assume that a seller has one well, irrigates 5 acre of his own plot and sells water to two
buyers; the first buyer irri gates 3acres and the second buyer irri gates 2 acres. When we
look at water price charged to the first buyer, the area irrigated by others means the sum
of his seller's and the second buyer's irrigation area (i.e. 5+2=7). Thus, this variable
measures the demand for water in the absence of the buyer we are looking at. The higher

this demand, the higher the seller’s reservation utility, 7?, (.) and thus the higher the water

price. At the same time, this variable would also effect the quality of irrigation water.
The larger the area irrigated by others, the more constrained is the buyer to get sufficient
amount of water at correct timing, resulting in low quality of water and a cheaper price.
The net effect would be empirically determined.

Unfortunately, however, this variable may not be exogenous because the seller
himself is among the decision makers. If so, the bias in this variable would bias all the
other coefficients. Given limited information in this data set, I can neither adequately
address the nature of potential bias in the parameter estimates due to problems associated

with endogenous/omitted variables, nor can I adequately instrument this endogenous

41

variable. Alternatively, I run regression without (Model 1) and with (Model 2) this
variable to see the robustness of the results.

The other variables which attempt to measure h, are the seller’s agrarian
characteristics such as last year’s animal holdings (oxen and cows), land holdings (acre),
schooling years of household head, and age of household head. As I mentioned, all
sellers in our field use their irrigation systems also for their own plots, so that if a current
buyer exits from the transaction (i.e. disagreement), at least seller himself would become
less constrained in terms of his own irrigation activity. The amount of profit from
utilization of that additional water (and hence his reservation utility) depends on the
seller’s agrarian characteristics mentioned above. These variables may not be truly
treated as exogenous because the farmers adjust these variables in their long term
decision process. However, they can be still treated as “quasi”-fixed for cross-section
estimation process. Arguably, schooling, and age are not current year’s choice variables.
Land holdings are seldom adjusted. Since the farmers sell their animals for input
purchase occasionally, the number of current year’s animals is endogenous. But the last
year’s number is given to the current year activity.

The explanatory variables on the buyer’s side are their agrarian characteristics
which consist of the same variables as the seller’s agrarian characteristics. It seems to be
straightforward that these variables specify the buyer’s production function, f(.). Since

buyer's reservation utility, 17,, (.) , depends either on the performance of his agricultural

activity without irrigation water or on the profit from non-farm activities, these variables
would also affect water prices through changes in reservation utility. I also introduce

dummy variables if the buyer has alternative water sellers or not. Since all buyers

42

currently choose to irrigate their plots, in case of rejection by the current seller, the
buyers who have alternative water sellers could have higher profit thus higher reservation
utility than the buyers who have only one seller.22 The existence of alternative water
sellers available to a particular buyer is exogenously determined.

Besides the variables suggested by the model, I use several covariates as
controlling variables. First, kinship is among the most important socioeconomic factors
affecting an outcome of bilateral transaction. Transactions among closer relationships are
usually claimed to reduce transaction costs by ensuring a payment, although there is
always the converse story of costly transactions with relatives due to difficulty of
collecting a payment (Meinzen-Dick, 1996). Unfortunately, our data set does not contain
data on kinship relationship. Alternatively, what I introduce is a dummy variable for
whether the seller and the buyer belong to the same caste group. Although this broad-
meaning proxy may have complicated effects, I still think it plays an vital role in the
Indian context.

Second, I control for village fixed effect. Since significant variations in input
prices, p,, are also captured by the village dummies (Table 3-2), I do not include other
input prices in my regression analysis, while they appear on the theoretical model.

Finally, I allow for a dummy variable to capture whether the payment schedule
chosen by pairs of sellers and buyers is sharecrop or not. The payment schedule,
however, may be chosen simultaneously with other endogenous variables. There are no
extra exogenous variables which determine the payment schedule but do not appear on

water charge function (i.e. under-identified) since sellers and buyers regard the payment

 

22 Buying water from an alternative seller is a contingent event so that his existence does not change
current production technology f(. ).

43

schedule and water charge as a set of contract. If this is the case, we have no way but
estimating the reduced form water charge function, and thus we can not know how
sharecropping is related with water charge. On the other hand, we may claim that they
make their decision stepwise: they first choose the payment schedule and then talk about
how much the water charge is. The payment schedule may be persistent when it is
decided in the past. If this is the case, the system becomes recursive one, and then I can
use the sharecrop dummy as predetermined exogenous variable in our water charge
function. Nevertheless, we can never deny the simultaneity of their decision process, so
that I run regressions with this dummy for a comparative purpose (Model 3 without the
area irrigated by others, Model 4 with it). The descriptive statistics of the variables I use

are summarized in Table 3-3.

3.5 Regression Results

Market Structure

The results from OLS regressions and joint hypothesis tests are given in Table 3-
4. A robust result across the models is that the variations in unit water price are
statistically explained by the seller’s characteristics, h_,, but not by the buyer’s
characteristics, hb. We can see this by checking significance levels of each variable.
Joint hypothesis tests are also performed to see whether variables classified under h, (h)

are jointly zero or not (see the bottom of the table). The test results indicates that the

44

seller’s variables as a whole have quite significant effect on the unit water price, while
the buyer’s variables do not.

This result implies two potential market structures. The first but less plausible
one is the case that all buyers have the same demand schedule for irrigation water so that
the price is determined solely by the shifi of the seller’s supply schedule. As we see there
is considerable heterogeneity among the buyers (Table 3-3), thus this type of market
structure would not be plausible.

The other potential structure, which is also claimed by the existing case studies, is
that the sellers sell "surplus" water after meeting the needs of their own fields, and thus
do not ask more than the cost of irrigation (Saleth, 1998; Meinzen-Dick, 1996; Shah,
1993). In this case, regardless of the buyers' demand schedule, only the sellers'
characteristics which are related with irrigation cost becomes significant.

It may sound too good for buyers that no matter how strong the sellers bargaining
power they do not charge more than the cost just because water is "surplus." However, in
closed society like agrarian villages "maximizing profit from water sales could lose
goodwill and cost the seller more in the long run (Meinzen-Dick, 1996, p. 39)." Saleth
(1998) also claims that "the sellers may not exercise their power because of social
constraints (p. 198)." Thus, sellers would set price so as not to enjoy too much excess
profit. This is consistent with my result in section 3.2; on average the sellers do not enjoy
large amount of extra profit from water sales. Furthermore, the sellers' bargaining power
may not strong enough to enjoy excess profit from water sales. As I mentioned,
irrigation facilities are not scarce in the villages. Besides, informal talks with the

respondents during the survey revealed that the villagers felt that aquifer levels were

45

sufficient even during dry season. Taking into account of these stylized facts at
groundwater markets, there appears to be no empirical justification of characterizing
seller's water pricing behavior as exploitative, i.e. "water-lord." Rather, price is set
according to the cost structure of the irrigation systems which is determined by sellers

characteristics.

Determinants of Unit Water Price

We now turn to interpretation of determinants of water price. It may be better to
first note that the magnitude of the coefficients do not vary across models when they are
significant under all models. The results discussed below are robust across the models.

The results show that variations in unit water price are explained not only by
village-level factors (see joint F test for village dummies) but also by individual
characteristics. Comparing with the regression result in section 3.2 (OLS regression of
water price on village dummies), the adjusted R-squared and the F statistics of the
models in Table 3-4 are improved. One possibility of the low explanatory power of the
regression of water price in section 3.2 would be due to measurement errors. However,
the results in Table 34 do not support this conjecture, but do support the idea that price
variations are explained by individual-specific characteristics as well as village factors.

Looking at each explanatory variable, the amount of total investment in irrigation
systems has positive and significant effect on unit water price, which may be explained
by two reasons. First, this variable represents the capacity of the inigation system. The
capacity relates with the quality of irrigation water since water is conveyed without loss if

pipelines are available, and water is applied timely if horsepower is strong; the value of

46

water becomes higher in this case. Second, the absolute amount of seller’s loan
repayment in each year is high when the amount of investment is large, and thus the
seller would ask high price to make his buyer partially shoulder the repayment.

The area irrigated by others have very weak statistical relationship with water
price, implying that the sellers’ reservation profit does not change even when the demand
for water by others is high. This supports the fact that sellers sell “surplus” water. If the
water sold is “surplus,” reservation profit from that water is always nil. Thus, the
variable related with the amount of reservation profit has no effects on water price.

Among seller's agrarian characteristics, animal holdings and schooling years are
statistically significant. The positive sign on schooling years is consistent with my
conjecture claiming that increased schooling increases the seller's reservation profit due
to his high ability to utilize additional amount of irrigation water. However, again, the
sellers seem to sell "surplus" water, so that the returns from additional amount of water is
zero regardless of their ability. Thus, the reasoning based on the seller's reservation
utility is not relevant for schooling years. Since the seller's schooling years may measure
other factors, they probably become significant for different reasons. One possibility is a
quality-related reason; the irrigation systems owned by the sellers with high management
skills may have better maintenance, and thus the quality of water provided by them may
be reasonably good to ask higher price. The other possibility is that the sellers might sell
irrigation water with their water management skills, so that the sellers with high ability
can provide better irrigation service and can ask higher price.

The animal holdings are negatively correlated with unit water price. The

reasoning based on reservation utility argument is not convincing under the market

47

structure in our study field, rather a specific feature of the rural credit/asset markets
explains the situation well. In rural India, people regard animals not only as inputs for
agricultural production but also as asset, and they often sell animals when they need
money (Rosenzweig and Wolpin, 1993; Walker and Ryan 1990). Thus, the sellers who
hold enough animals could completely or partially self-fmance their investment in
irrigation systems. Small or no burden of loan repayment would let the sellers ask low
water price to the buyers, resulting in negative sign on animal holdings. Then, why the
coefficient on sellers' land holdings, another important form of asset, is not significant?
Note that the role of land and animals are different. Land is not sold as ofien as animals,
but it plays an important role as collateral for a loan especially at informal credit markets.
However, the sellers who need credit for investment in irrigation systems use formal
programs (such as inigation development program) which charge relatively
homogeneous interest rates, instead of informal credit. Thus, at water markets land is not
related with the amount of loan repayment and thus with unit water price.

Belonging to the same caste group has no statistical relationship with unit water
price. The descriptive statistics in Table 3-3 tells us that 39 % of transactions are
conducted between different caste groups. However, the percentage sharply declines to
13 % if we look only at transactions between very "distant" caste groups (ex. between
Brahmin and scheduled caste/tribe). Thus, water transactions may be restricted among
individuals belonging to the same or closely related caste groups. We have to wait for
further empirical investigation to answer the questions of whether caste group actually
restricts water transaction, and if so, how is water price determined when caste groups are

"distant."

48

Lastly, it is worth noting that sharecrop dummy is positively related to unit water
price in the Models 3 and 4, albeit I am ignoring the potential endogenous problem. As
shown in the theoretical section, this may be taken to imply the existence of a risk
premium payment from buyers to sellers. Shah’s (1993) observation in India, however,
suggests another potential interpretation. He claims that, as inigation systems have been
disseminated, there is a tendency of disappearance of sharecrop contract. If so, farmers
in the less irrigated area (therefore, high water price area) might selectively choose
sharecrop contract, giving multiple meanings to this dummy variable: not only a risk
premium payment but also water scarcity. However, since water scarcity is already
controlled by area irrigated by others and also by village dummies, we can not deny the

existence of a risk premium payment in our study area.

3.6 Concluding Remarks and Policy Implications

Based on the two-person bargaining model, this paper analyzes the structure of
the groundwater markets and price determination for a sample of water sellers and buyers
in highly irrigated villages in Madhya Pradesh, India. I do not find strong evidence that
supports seller’s exploitation of buyer. The market structure in the study area rather
seems to work in favor of the buyers in that water charge is based on the cost of irrigation
systems. This is partly because social constraints do not allow the sellers to enjoy
unacceptable amount of excess profit. The result may also stem from over investment in

private lift-irrigation systems. In other words, the number of irrigation systems in the

49

villages may be more than the village-optimal (so that sellers have no bargaining power
to enjoy excess profit). All farmers do not necessarily have to have their own irrigation
systems as long as there exist adequate water transactions. It is worth reevaluating the
current irrigation development schemes which could be inducing farmers' over-
investment in private lifi-irrigation systems. The supportive schemes may be better
restricted to the farmers who have difficulty in access to irrigation water due to either
geographical or social reasons.

My analysis also clarifies whether and how individual-specific characteristics
affect unit water price, albeit, with potential bias in the parameter estimates due to
problems associated with endogenous variables. The results show significant effects of
individual characteristics on water pricing. The seller '3 characteristics -- amount of total
investment in irrigation systems, animal holdings, and schooling years -- are important
determinants of groundwater price. Different prices are also charged depending on the
types of contracts. In general, unit water price becomes higher when seller's physical
capital (total amount of investment) and human capital (schooling years) stocks are well
equipped. Hence, apart from the obvious effect from physical capacity, the results show
that human capital is also important, as it is in most of agrarian economies. Another
notable relationship is that an increase in amount of loan repayment (amount of
investment and animal holdings) may increase unit water price. Types of contracts also
significantly affect water price; the price becomes higher under sharecrop contract due to
a risk premium payment from buyers to sellers. This implies that some buyers have few
means but sharecrop contract for their income smoothing. Access to credit for this

purpose may be limited. As I mentioned, on the other hand, access to credit for the

50

purpose of irrigation investment may be relatively easy, thanks to the irrigation
development schemes. While dissemination of irrigation systems contributes to
productivity increase of both sellers and buyers, the sharecrop-buyers' welfare may not
improve as much as the others unless they have more options of income smoothing.
Further detailed investigation of the production environment of the sharecrop-buyers
would help guide policy makers in revising not only irrigation development schemes but

also other forms of credit programs.

51

Table 3-1: Ratio of water charge to irrigation cost and Return to Capital Investment in

different case studies

 

 

Location Ratio of Ratio of Return to
water water Capital
charge to charge to Investment
variable cost the sum of (%)
variable and
depreciation
cost

Case Studies Summarized by Shah (1993: 75) a

West Godavari, India 1.2 — 1.3

West UP. and Punjab, India 1.3 — 1.7

East and Central UP, India 1.7 — 2.0

North Kheda, India 1.89

Midnapur dist. West Bengal, India 2.3

Panchmahal dist. Gujarat, India 2.7 — 3.0

Madurai dist. T.N.; Karimanagar dist. 2.7 - 3.5

A.P., India

Case Study by Shah and Ballabh (1997) a

North Bihar, India 2.5 — 3.0 1.25 — 1.8
Case Study by Fajita and Hossain (1995) b

Barind tract, Bangladesh 2.59 2.00 69 %
My Study °

Madhya Pradesh, India 3.07 1.87 21 %

 

Notes: ‘ The unit is Rs. per hour of pumping.
b The unit is Taka per irrigated acre
° See Appendix for detail

52

Table 3-2: Price variations within and across villages by types of input goods

 

 

 

 

Dependent Variable
Seed price Urea price Super phosphate DAP price Unit water price
price
Village B dummy -0.394 0.400 0.250 0.250 -4.821
(1.04) (2.22)* (1.88) (0.63) (0.38)
Village C dummy -0.316 -0.005 -0. 100 -0.523 3.081
(1.00) (0.03) (0.74) (1.40) (0.25)
Village D dummy -l.138 -0.462 -0.250 -1.432 13.250
(3.21)" (2.62)‘ (1.84) (3 .28)" (1.05)
Village E dummy -0.686 -0.250 -0.037 -1 .139 5.741
(1.97) (1.47) (0.29) (2.77)* (0.47)
Village F dummy -0.560 0.361 -0.167 -0.917 24.259
(1.26) (1.67) (1.12) (2.10) (1.54)
High yield dummy 0.269
(1.48)
Constant 1.625 0.750 0.500 2.250 35.000
(5.64)" (4.9mm (4.13)” (6.31)" (3.14)"
Observations 36 34 25 23 38
R—squared 0.41 0.68 0.70 0.73 0.20
Adjusted R-squared 0.29 0.62 0.62 0.65 0.08
F statistics 3.36 1 1.65 8.76 9.03 1.60
[0.01] [0.00] [0.00] [0.00] [0.12L

 

Numbers in parentheses are absolute t-values. Numbers in brackets are p-values.
* significant at 10% level; ” significant at 5% level, *“ significant at 1% level

53

Table 3-3: Descriptive Statistics of Variables

 

 

 

Variable Mean Standard Deviation
Dependent Variable
Unit water price (kg. of wheat) 40.08 16.39
Explanatory Variables
Seller's Irrigation Technology and Agrarian Characteristics: h,
Amount of total investment ( 1000 kg. of wheat) 6.08 3.87
Area irrigated by others (acres) 2.71 2.97
Seller's last year animal holdings 3.13 1.66
Seller’s land holdings (acres) 9.63 9.95
Seller’s schooling years 4.32 4.29
Seller’s age 45.97 15.03
Buyer's Production Technology and Agrarian Characteristics: hb
Buyer’s last year animal holdings 2.26 1.81
Buyer’s land holdings (acres) 5.08 4.79
Buyer’s schooling years 5.58 4.99
Buyer’s age 44.42 11.01
Alternative water seller dummy 0.24 0.43
Controlling Variables
Caste matching dummy 0.61 0.50
Village B dummy 0.18 0.39
Village C dummy 0.29 0.46
Village D dummy 0.18 0.39
Village E dummy 0.24 0.43
Village F dummy 0.05 0.22
Sharecrop contract dummy 0.45 0.50

 

54

Table 3-4: OLS Regression Analysis of Determinants of Unit Water Price

Without sharecrop dummy

with sharecrop dummy

 

 

 

Model 1 Model 2 Model 3 Model 4
Seller '3 irrigation technologi and agrarian characteristics: h,
Amount of total investment 1.91 1 1.721 2.033 1.878
(2.06) (1.85) (2.44)‘ (2.22)‘
Seller’s last year animal holdings -5.274 -6.919 -6. 156 -7.402
(1.91) (2.27)* (2.47)* (2.68)’
Seller’s land holdings (acres) 1.399 0.523 0.791 0.134
(1.74) (0.49) (1.04) (0.14)
Seller’s schooling years 1.955 1.769 1.930 1.786
(2.10)‘ (1.90) (2.32)"‘ (2.12)"I
Seller’s a 0.322 0.343 0.398 0.410
(1.12) (1.20) (1.53) (1.58)
Area irrigated by others 1.321 1.037
(1.20) (1.04)
Buyer '3 production technology and agrarian characteristics: hb
Buyer’s last year animal holdings -0.612 0.062 -0.569 -0.042
(0.39) (0.04) (0.41) (0.03)
Buyer’s land holdings (acres) -0.022 0.213 0.222 0.394
(0.03) (0.27) (0.32) (0.55)
Buyer’s schooling years -0.037 0.039 0.300 0.343
(0.06) (0.06) (0.49) (0.56)
Buyer’s age 0.406 0.162 0.275 0.090
(1.40) (0.46) (1 .04) (0.28)
Alternative water seller dummy 8.913 8.690 9.180 8.991
(1.27) (1.25) (1.46) (1.43)
Controlling variable
Caste matching dummy -1.249 2.243 0.841 3.477
(0.20) (0.33) (0.15) (0.56)
Village B dummy 25.779 10.991 14.446 3.410
(0.94) (0.37) . (0.58) (0.13)
Village C dummy 40.447 28.534 22.307 13.874
(1.38) (0.93) (0.82) (0.49)
Village D dummy 48.571 29.925 29.676 15.995
(1.95) (1.03) (1.26) (0.59)
Village E dummy 23.627 12.596 -3.776 -11.047
(0.84) (0.43) (0.14) (0.39)
Village F dummy 25.296 22.721 32.010 29.648
(1.44) (1.29) (2.00) (1.84)
Sharecrop contract dummy 16.273 15.448
(2.49)‘ (2.35)"l
Constant -41.780 -22.260 -26.313 -1 1.772
(1.25) (0.61) (0.86) (0.35)
Joint F test 3.65 3.35 3.94 3.48
(Ho: h, = 0) [0.02] [0.02] [0.01] [0.02]
Joint F test 0.64 0.35 0.65 0.51
(Ho: h, = 0) [0.67] [0.87] [0.66] [0.76]
Joint F test 2.31 1.59 3.21 2.42
(Ho: all village dummies =0) [0.08] [0.20] [0.02] [0.07]
Observations 38 38 38 38
R-squared 0.61 0.63 0.70 0.71
Adjusted R-squared 0.31 0.32 0.45 0.45
F test 2.06 2.07 2.79 2.7
0.06 0.06 [0.02] [0.02]

 

Numbers in parentheses are absolute t-values. Numbers in brackets are p-values.
"‘ significant at 10% level; “ significant at 5% level, "* significant at 1% level

SUMMARY AND CONCLUSION

This paper explores the mechanisms of contract choice and price determination
under bilateral transaction between a owner of irrigation systems and a non-owner at
groundwater markets. I use a sample of water sellers and buyers in the six villages in
Madhya Pradesh in India for empirical analysis.

In our study field, we observe three types of contracts: sharecrop, fixed-rent, and
flat-charge contracts. However, the multinomial logit regression analysis in chapter 3
shows that there is no statistical difference in the factors associated with the choice
between flat-charge and fixed-rent contracts. This is probably due to the flexibility of
informal contracts which allows farmers to reschedule fixed-rent payments according to
the actual frequency of irrigation. Meanwhile, the results show that farmers significantly
distinguish sharecrop over non-sharecrop contract (i.e. fixed-rent and flat-charge
contract). I conduct both a logit regression analysis and a village-level qualitative
analysis on the choice of these two types. The analyses show that risk-sharing is the
factor underlying the choice of sharecrop contract; other factors such as limited liability,
managerial input sharing, screening and monitoring cost are not significant determinants.

This result is consistent with my finding in chapter 4, where OLS regression
results show that the unit water price becomes higher under sharecrop contract, implying
that the buyers pay a risk premium to the sellers when the risk is transferred from the
buyers to the sellers by sharecrop contract. The other significant factors affecting the unit
water price are the sellers' physical (irrigation systems) and human (education) capital.
The buyers' characteristics have no effect at all. As the existing literature suggests, this is

probably because the sellers sell "surplus" water afier meeting the needs of their field,

56

and because social norms does not allow the sellers to charge a high price for surplus. In
this case, regardless of the buyers' characteristics, only the seller's characteristics which
are related with the cost structure of water become significant. Thus, in contrast to the
conventional notion that a seller operates as a monopolistic exploiter, the market structure
in the study area rather seems to work in favor of the buyers in that they are charged no
greater than the cost of irrigation systems.

In sum, while the water prices charged to the buyers are not exploitative,
in a risky environment with limited means of income smoothing, the sharecrop contract is
chosen with a payment of a risk premium from the buyers to their sellers. Therefore,
there arises two concerns for sharecrop-buyers in terms of efficiency and equity. First,
the Marshallian inefficiency may exist in the sharecrop buyers' production as we see in
the sharecrop land tenants' production in India. Second, while the dissemination of
irrigation systems contributes to a productivity increase by all water users, the sharecrop
buyers' welfare may not improve as much as others since they pay a risk premium from
their inefficient outputs. If these conditions persist for the buyers, policy makers had
better consider how to increase access to means of income smoothing and also how to
reduce risk in agricultural production. For future research, detailed investigation of
production environments and outcomes of the sharecrop-water-buyers would help guide

policy makers in designing more detailed strategies to enhance efficiency and equity.

57

APPENDICES

APPENDIX A

REVIEW OF LITERATURE ON CONTRACT CHOICE AT LAND RENTAL
MARKETS

When a landlord decides not to cultivate his own land by himself, he may lease
out his land to a tenant by using either sharecrop or fixed-rent contracts, or he may offer
wage-labor contract to farm his own land. There are five primary microeconomic models
explaining contractual choices between a landlord and a landless in economic literature:
(1) risk-sharing, (2) limited liability, (3) screening, (4) double incentive, and (5)
Monitoring cost models.23 In general, a probability of a particular type of contract j (1' =
fixed-rent, sharecrop, fixed-wage) being chosen between a pair i of the landlord and the

worker is expressed as a function of four sets of variables:

Pi] = f (risk, wealth, ability, and transaction costs)

In this section I discuss how each model relates these variables to contract choice. Basic

paradigms and the list of the four sets of variables are summarized below.24

Risk-sharing Model
One of the views first presented in the literature is the recognition of contract as a

risk-sharing device under the environment of missing insurance markets (Cheung, 1969).

 

2’ Some papers consider only fixed-rent and sharecrop contracts.

2‘ I cover only basic models so that I can contrast models and see the differences. Moreover, because of the
purpose of this study, I will not explain all features of each model, but highlight the logic of the contract
choice. Hayami and Otsuka (1993) and Singh (1989) give us an intensive review of different theories of
contract.

59

Stiglitz (1974) and Newbery and Stiglitz (1979) complete this approach by introducing a
trade-off relationship between risk-sharing and work incentive. Introducing plausible
assumption that the landlord can not perfectly enforce contract regarding the tenant’s
labor effort, share contract may become the best institutional arrangement for risk-
sharing. Fixed-rent contract provide perfect incentives since all the output is obtained by
the tenant, while it provides no risk-sharing (the tenants shoulder all the risk); on the
other hand labors would shirk under fixed-wage contract given certain amount of wages
for sure when contract is unenforceable. Given this trade-off relationship, sharecrop
contract arises to share risks and also to provide incentives. Hence, not only the parties’
risk attitude and risk itself, but also the laborers’ reaction to incentives (i.e. share of
output) matter for contract choice

The hypothesis derived in this model is that assuming the landlord is always risk-
neutral the greater the landless’ the risk-aversion and the greater the risk, the closer the
optimal contract approximates fixed-wage contract. Besides, the greater the labor supply
elasticity, i.e. the more sensitive the worker is to incentives, the greater the share, i.e. the

closer to fixed-rent contract (Stiglitz, 1974, 1987).

Limited Liability Model

Limited liability may be a plausible reason to let some tenants choose sharecrop
contract and let others choose fixed-rent without assuming risk-averse tenants. The
intuitive logic is as follows. Due to nonexistence of contingent market in an agrarian
economy, liability of tenant’s payment is limited by their initial wealth. The landlord will

know that the default on fixed-rent may occur when the harvest fails and the tenant is

60

poor. One way to lessen the possibility of entire default is for the landlord to lower the
rent in bad states and raise it in good states, which is akin to sharecropping. Shetty
(1988) shows that the poor tenant will receive sharecrop contract given the assumption of
production uncertainty, limited liability, variation in tenants’ wealth, and non-
observability of labors’ effort.”

Note these models are incomplete in the sense that they exclude fixed-wage
contract from their spectrum of contracts. Nevertheless, among several useful
propositions derived in their theoretical works, the most important one for my research
purpose is that: for the tenant with small wealth, the share cropping contract is likely to
be chosen over fixed-rent contract given non-observability of labor effort as well as
uncertainty in production and limited liability.” As the model assumes risk-neutral

individuals, risk-sharing factors do not matter in this model.

Screening Model

The screening model represents another approach to explain the existence of a
variety of contracts between risk-neutral individuals under the environment of
unenforceable contract. A landlord, who cannot directly observe tenant’s true ability,
offers a menu of contracts which is designed such that tenants with low ability do not
mimic themselves as high ability tenants but they select a low profile contract by

themselves. A high ability tenant will prefer fixed-rent contract regardless of higher rent,

 

25 Along this line, Basu (1992) and Sengupta (1997) point out that non-obserbability of project-choice also
plays an important role in contract choice under limited liability. Under fixed-rent contract, the tenants
would prefer riskier but higher return projects even though such projects lower overall surplus. This is
because the tenant’s income does not fall below the outcome of bad states, but can keep all residual in good
sates being realized. Under the share contract the tenants’ income, on the other hand, moves in proportion
with the output realized, resulting in choosing the project that yields the highest joint surplus.

61

because he gets to keep his high marginal product entirely. But this amount of rent is too
high to accept for a low ability tenant, and thus the low ability person does not enter
market for high ability individuals. On the other hand the low ability tenant prefer fixed-
wage contract which guarantees specific amount of revenue to him regardless of his
(poor) performance. And the individual in between chooses share contract. Thus,
improvement in management ability increases the occurrence of fixed-rent contract and
decreases the occurrence of fixed-wage contract.

Hallagan(l978) and Newbery and Stiglitz (1979) independently show that the
landlord can design such a contract under some circumstances. However, this model is
criticized that the ability is generally revealed sooner or later especially in small rural
community. Hence, this model is claimed to be valid only when there are a perennial in-
rrrigration of fresh tenants in the community (Ray, 1998, ch 12; Singh, 1989; Eswaran
and Kotwal, 1985). An increase in the number of new immigrants, who are perceived
having low ability, has negative effect on fixed-rent but positive effect on fixed-wage

contract.

Double-incentive Model

All models mentioned so far assume perfectly unobservable labor efforts. This
model, however, assume several activities (including labor efforts) can be costly
monitored. Eswaran and Kotwal (1985) present a model in which wage contract is
chosen when a landlord can cheaply supervise hired labor, fixed-rent is chosen when a

landless can cheaply provide managerial inputs, and sharecrop is chosen when both are

 

2‘ This implication is consistent with the notion of ‘agricultural ladder’ which explains the landless farmers’
step-up from wage-labor to owner cultivator as they mature (Shetty, 1988; Ray, 1998, ch. 12).

62

expensive. They assume (1) both management and supervision are important in
production, (2) the landlord is more efficient at management while the tenant is more
efficient at supervision, because the landlord has better access to agricultural information
while the tenant is better at supervising his family labor, and (3) neither service is
unobtainable at the market. Thus, both landlord and tenant are required to apply their
inherent inputs. Since each party has different incentive regarding how much they apply
their effort, this model has double-incentive problem (Ray, 1998, ch. 12). Note that in
this model both parties are risk-neutral again, so risk-sharing does not enter as a
determinant.

Let me denote t,- as managerial input provided either by landlord (i = I) or by
tenant (i = t), and similarly s, as supervisory input. The idea of differential efficiency is
quantified by the parameter 7) where 0 S y, S 1. For example, the closer y, is to one, the
closer the landlord’s supervision efficiency is to the tenant’s (note that tenant’s
supervision efficiency is always one). Likewise, the closer y, to one, the more eflicient
the tenant’s management efficiency is. Given these assumptions, if fixed-wage contract
is chosen, production is realized by the combination of (t), ylsz), since all production
decisions are made by the landlord. The landlord is the residual claimant in this case. On
the other hand under fixed-rent contract, the tenant takes all production decision and is
the residual claimant. Hence production is realized by the combination of (Wt, s,). Under
share-cropping contract, each party jointly provide one of the unmarketed inputs. Thus,
they provide the factor who has the most advantage, namely t fiom landlord and s from

tenant, and production is realized by the combination of (t), s,), and output is shared

63

between them. Finally, the landlord compares the profit achieved under three alternative
contracts, and chooses the most profitable one.

Eswaran and Kotwal show which contact is chosen when 71, and y, vary.27 For
high values of y), the loss of efficiency is very small even if the landlord supervises the
production processes, and so the landlord chooses the fixed-wage contract. By contrast,
if the values of y, is high, the landlord is likely to offer fixed—rent contract because there is
small loss in tenant’s own management, and thus the landlord also can expect large
amount of rent from the tenant. However, if the production requires both management

and supervision, they should be jointly supplied under share-cropping contract.

Monitoring Cost Model

Another line of costly-monitoring approach is the one considering a trade-off
relationship between labor-effort monitoring cost and land-quality monitoring cost
(Roumasset and Uy, 1987; Datta et al., 1986; Alston et al., 1984). Basically they claim
that monitoring costs of labor-efforts tend to decrease as rewards become more closely
linked with efforts. For example, monitoring cost is greatest under fixed-wage contract
because the labor receives fixed amount regardless of his actual effort. Contrarily, the
costs are least under fixed-rent contract since the tenant is rewarded for his effort. They
claim that sharecropping contract occupy an intermediate ground in terms of monitoring
costs, so that the monitoring cost firnction is depicted as downward sloping curve in
Figure A-l in which horizontal axis, a, is share of output. On the other hand, the

monitoring cost function for land-quality has opposite SIOpe because they claim that it is

64

reasonable to assume that the tenant’s incentive to abuse land (and also landlord’s capital

embodied in his land) becomes larger as a get closes to one.28

Assuming the both functions are convex and independent as shown in Figure A-l,
the optimal a is determined at the point where total transaction costs are minimized.
Figure A-l shows the situation that the sharecrop contract is optimal. If the effort-
monitoring cost curve shifts upward through, for example, an increase of labor intensity,
and/or the quality monitoring cost curve shifts downward through a decrease of land
vulnerability, then fixed-rent contract becomes more likely to be chosen, and vise versa.

Allan and Lueck (1992) also consider additional transaction costs for physical
measurement and division of harvested crop because the tenant has an incentive to
underreport the harvest to the landlord under sharecrop contract. Under fixed-rent
contract, however, the landlord does not have to worry about this as the rent is paid by
cash in advance. Thus, the more difficult the harvest monitoring is, the less sharecrop
contract (more fixed-rent contract) is chosen.

It is better to include production uncertainty as an explanatory variable because
some of the papers with this approach obtain an opposite empirical result from the risk-
sharing model: the greater the production uncertainty, the more likely fixed-rent contract
is chosen (Allen and Lueck, 1992, 1995; Datta et al., 1986). Since they assume risk-
neutral tenants, they do not expect a contract as a risk-sharing device anymore.

Moreover, they conclude that an increase of uncertainty in production makes detection of

 

27 Since explicit analytical solution is not obtainable, Eswaran and Kotwal, assuming the Cobb-Douglas
production function, do numerical simulations and draw the figure.

8 They consider one period model. The tenant is implicitly assumed not to care about the effect of abuse
on renewal of the contract.

65

labor-cheating more difficult (upward shift of labor effort monitoring cost curve in Figure
A-l), resulting in choosing fixed-rent contract.29

Allen and Lueck criticize the assumption of risk-averse tenants, and support
transaction cost approach. However, as they mention in their paper, this critique may
apply only for the developed countries in which neither demography nor human capital
between landlord and tenant differs so largely.30 Note that this is the only variable I
expect the opposite sign from the alternative models. In developing countries,
assumption of risk-averse tenant is still plausible, while the assumption about monitoring
cost is also reasonable. Thus, effect of production uncertainty on contract choice may be
an empirical question. Hence, in this model contract choice is a function of not only

several kinds of monitoring costs but also production uncertainty.

 

29 Allen and Lueck (1992) likewise interpret that an increase of production uncertainty makes detection of
underreporting of the harvest which supposed to be shared.
3° Their empirical result rely on data from American Midwest.

66

Figure A-l: Transaction Cost and Optimal Contract Choice

Total Transaction cost

  
  

increase of decrease of land
labor intensity vulnerability

Labor-effort
monitoring cost

 

 

 

 

-. ._ .- Land-quality
monitoring
cost curve
all!
fixed-wage contract sharecrop contract fixed-rent
0‘ = 0 a = a* contract

":1

Source: Hayami and Otsuka, 1993 (modified by the author)

67

APPENDIX B

SUPPLEMENTARY TABLE FOR CHAPTER 2

Table B-1: Multinomial-logit Regression Analysis of Determinants of Contract Choice

(Constrained Models)
(The comparison group is sharecrop contract)

 

 

 

With Hired manager dummy Without Hired manager
' dummy
Fixed-rent Flat-charge Fixed-rent Flat-charge
Household Wealth
Buyer's animals 0.078 0.078 0.047 0.047
(0.25) (0.25) (0.15) (0.15)
Seller's animals -1.304 -1.304 -1.299 -1.299
(1.73) (1.73) (1.69) (1.69)
Buyer's land holdings 0.070 0.070 0.077 0.077
(0.47) (0.47) (0.50) (0.50)
Seller's land holdings 0.180 0.180 0.159 0.159
(1.42) (1.42) (1.26) (1.26)
Management Ability
Buyer's schooling -0.003 -0.003 -0.001 -0.001
(0.03) (0.03) (0.01) (0.01)
Seller's schooling -0.297 -0.297 -0.330 -0.330
(1.65) (1.65) (1.68) (1.68)
Buyer's age -0.076 -0.076 -0.057 -0.057
(1.31) (1.31) (0.93) (0.93)
Seller's age 0.018 0.018 0.016 0.016
(0.33) (0.33) (0.29) (0.29)
Factor related with Moral Hazard
Total investment -0.244 -0.244 -0.204 -0.204
(1.08) (1 .08) (0.90) (0.90)
Cast matching dummy 3.315 3.315 2.823 2.823
(1.44) (1.44) (1.23) (1.23)
Alternative seller dummy 3.443 3.443 3.461 3.461
(1.37) (1.37) (1.44) (1.44)
Manager dummy 1.44 1.44
(0.83) (0.83)
Constant 3.926 3.440 3.33 2.84
(0.99) (0.86) (0.83) (0.71)
Observations 38 38
)12 statistics 19.66 20.41
[0.05] [0.06]
Pesudo R2 0.25 0.25
3 Likelihood -30.25 -29.88

 

Numbers in parentheses are absolute value of z-statistics. Numbers in brackets are p-

values.

* significant at 10% level; ** significant at 5% level, “'* significant at 1% level

68

APPENDIX C

CALCULATION OF THE WATER-CHARGE-CST RATIO AND THE RATE OF
RETURN TO IRRIGATION INVESTMENT

Following the method used by Fujita and Hossain (1995), I calculate the water-
charge-cost ratio and the rate of return of inigation investment. Table A1 summarizes
the results for each of 45 samples used in my analysis. The average of the ratio of water
charge to operational cost weighted by area of irrigation (col. 4) is 3.43. The average of
the ratio of water charge to the sum of operational and depreciation costs is 2.15. The
average of the return of investment is 29.22 %.

Table C-l: The Water-charge-cost ratio and the rate of return to irrigation investment

 

 

Sarnplc (1) (2) (3) (4) (5) (6) (7)
Water Operational Capital Area Irrigated Ratio of Water Ratio of Water Capital Return
Charge/acre Cost/acre Investment/ by Buyer Charge to Charge to the (%) "

(R5) (R5,) ' acre (acre) Operational Sum of
(Rs.) b Cost ° Operational
and
Depreciation
Costs "
1 300 324 1364 2 0.93 0.76 -1.77
2 333 733 6000 1.5 0.45 0.32 -6.67
3 360 387 2500 5 0.93 0.70 -1.07
4 375 380 2545 3 0.99 0.74 -0.20
5 375 300 800 4 1.25 1.10 9.38
6 375 745 3857 5 0.50 0.40 -9.59
7 400 50 700 1 8.00 4.71 50.00
8 450 138 4500 2 3.27 1.24 6.94
9 480 521 3118 2.5 0.92 0.71 -1.33
10 500 589 1500 2 0.85 0.75 -5.93
11 500 311 4000 1 1.61 0.98 4.73
12 500 363 4267 2.5 1.38 0.87 3.20
13 600 311 4000 1 1.93 1.17 7.23
14 600 343 1333 3 1.75 1.47 19.29
15 625 672 2023 0.4 0.93 0.81 -2.35
16 625 672 2023 6 0.93 0.81 -2.35
17 750 725 3200 1 1.03 0.85 0.78
18 750 672 2023 0.5 1.12 0.97 3.83
19 800 311 4000 0.5 2.57 1.57 12.23
20 900 107 5333 2 8.44 2.41 14.88
21 1000 267 667 5 3.75 3.33 109.90
22 1000 226 8571 1 4.42 1.53 9.03
23 1000 766 9600 0.5 1.31 0.80 2.44
24 1000 343 1333 3 2.92 2.44 49.30
25 1042 166 1882 6 6.27 4.00 46.53
26 1067 226 8571 1.5 4.71 1.63 9.81
27 1067 226 8571 1.5 4.71 1.63 9.81
28 1083 188 2412 1.5 5.78 3.52 37.14

 

69

Table C-l (cont’d): The Water-charge-cost ratio and the rate of return to irrigation
investment

 

29 1167 107 5333 1.5 10.94 3.13 19.88
30 1500 574 1091 2.5 2.61 2.39 84.90
31 1500 305 2339 4 4.92 3.55 51.09
32 1500 166 1882 4 9.03 5.77 70.88
33 1500 223 2083 4 6.74 4.59 61.32
34 1500 107 5333 1 14.06 4.02 26.13
35 1500 672 2023 1.5 2.23 1.94 40.91
36 1600 398 3750 2 4.03 2.74 32.07
37 1666 574 1091 3 2.90 2.65 100.09
38 2125 690 6400 1 3.08 2.10 22.43
Area-Weighted Average 3.43 2.15 29.22

 

Notes: 3 Operational cost composed of such items as the cost of electricity, the cost of
repairs and necessary parts, and wage payments to labors. Wages are paid for
such work as excavating channels, operating pumps, and supervising buyers'
pump-use. A shadow wage (i.e. village wage) is used for cost calculation when
sellers do these works by themselves.

b Total investment in irrigation systems consists of the cost of well digging,
electric pump, and sprinkler systems.
:(5)=(1)/(2)

The depreciation cost of irrigation investment is estimated by the constant
amount method, assuming a life of 20 years of the system. The formula
becomes (6)=(1)/((2)+((3)/20)).

° (7)=((1)-(2))* 100/(3)

70

APPENDIX D

THE QUESTIONNAIRE

The Intensive Individual/Household Survey

FOR WELL OWNERS

District:

 

Village:

 

Name of Enurnerator:

 

1 Family
1.1 On the household head:

Name:

Date of Interview:

 

Sex: 1=male 2=female

Age:

 

Ethnic/Caste/Religious group:

 

 

 

 

 

Years of Schooling:

Married? 1. Yes ( years) 2. No
Occupation: 1. Farming 2. Non-farming(

In the past one year, did you have any diseases? 1. Yes 2. No

 

71

If 1. Yes, when and what were the diseases?

 

 

Have you heard about “global warming”? 1. Yes 2. No

1.2 On his/her children:

How many children do you have (including ones living separately)? Boys ( ) and
Girls ( )

How many of them are living separately? Boys ( ) and Girls ( )
How many children did the wife give birth to? Boys ( ) and Girls ( )
How many of them died before the age of five? Boys ( ) and Girls ( )

1.3 On the household (families sharing kitchen and food):

How large (how many household members) is your household?

 

How many of them work on farm? (including wage labor)

How many of them work off farm?

 

*The above classification includes someone who work both on and off farm. In
such a case, count twice in both categories.

 

How many married couples are there in your household?

72

2 Agriculture

Note: Ask about 1996/97 dry season

2.1 Land ownership

How large agricultural land did your household CULTIVATE (including the land rented
in) during dry season? ha

Irrigation of cultivated land

 

Area or Percentage Distance from the
irrigated source (m)

 

1. Well owned by yourself

 

2. Well owned by others

 

3. Common well/pond/tank

 

4. River/Stream

 

5. Large canal

 

 

6. Others( )

 

 

 

 

Distance: distance from the water source to your plot.

How large agricultural land does your household OWN? ha

How large of total owned land was irrigated in dry season ?___(ha or % of total)

How large of total owned land was rented OUT (cultivated by others) in dry season?
Irrigated:__(ha or %) Unirrigated:_(ha or %)

Source of irrigation water:

 

In addition to the owned land, how much is rented IN last year?
Irrigated: (ha or %) Unirrigated: (ha or %)

Source of irrigation water:

 

73

2.2 Tubewell

2.2.1 Tubewell Profile

How many tubewells do you have?:
If more than one, ask about the most important well on this questionnaire, and use
the supplementary questionnaires for the other wells (and the plots irrigated by

those wells).

 

 

 

Type: 1=Deep Tubewell 2=Shallow Tubewell 3=Other
Year of Installation: , New? 1=Yes 2=No,
Total Cost of Investment. for pump purchase Rs, for digging
___R8,
for pipe purchase Rs.

Payment: 1=Cash 2=Credit

 

 

 

 

In case of 2, how much? Rs., From where:

Unpaid: Rs.

Name of the maker: , Origin: , Horsepower:
hp

How large did THIS tubewell irrigate? (attention which year you are asking 1996/97 or

1997/98?)
For Own Cultivation: ha.
For Other Farmers’ cultivation: hg---> How many water buyers?:_

 

(visit these water buyers’ after finishing this household.)

74

2.2.2 Operation and Maintenance Cost

If you use your own labor for operation and maintenance, please fill in the days and the
numbers of people. Otherwise, please enter the amount of payment in Rs.

 

 

 

 

 

 

 

 

 

 

Cost Items Cost Does this cost include
the expenditure for
other farmers’
operation?

Fuel / Electricity Bill (Rs.) a = fuel, b= 1=Yes 2=No

electricity

Spare Parts (Rs.) N/A

Repair and Maintenance (Mechanical N/A

Charge) (Rs)

Well Transportation cost from house to field 1=Yes 2=No

(RS)

Driver’s / Linemen Salaries (Rs or days*#) 1=Yes 2=No

Canal digging (Rs or days*#) 1=Yes 2=No

Supervision Cost (supervision of buyers’ N/A

tubewell use) (Rs or days*#)

Miscellaneous (Rs) N/A

 

 

 

 

75

 

 

 

 

 

 

 

 

 

an :5
3: ac: >5
corn 5n onb
9%

RV

 

3c
6038
ofi «a
28 85
as

as

 

g
woo—ES
65 on
28 as
E

 

ca

“was:

so

 

 

 

 

 

 

 

 

 

 

956 as no 25 ea
Hood 33mg %onnouo no.5 05
\850 ..\a ASE no 32 no.5
6 E 5 S 8
mm H 8x38 05 3 Bow a no: 2: no 0:?» Son.
3 ” cum BE Eon.

Sancho? 28m

.3358 cougar; no mooofimmooo 05 :52 on
await, 05 E Boa =a 83:56 on. 652333 .8 832 swan—3 .xov oofion coo x8 .635, 2:: Bone commend ”v.08 Z
A363 C movie obs no» as» 05 Home 388 cm £682
.83 Batons: ~88 05 E mnoco =w Bonn xm< N682
:03 65 .3 @6335 $o_n acetone“ “won. 2: Sons 674 :6on
asaso as 835 53369.. :3
8888a 3538?. S.

76

 

8.3 use as

 

 

 

 

 

 

 

 

 

 

.893 038 om:
“one.“ 033 8:
Spain 033 8:
98:2 @853
.5 6a .5
Etc 288} is»
browns.» auburn 93o 53:5: mEootn when eon—SE
.3 £8 05 on mots Hon—3 Snood. 2583 «ED
6: a: 9.: a:

 

 

 

.noeo some Con Esofiw on: oEZn 8:58 20:38: 2: t 96: 9 :5: Bob 5:38 05 E Enoch :38 5 En

v

as 63.;

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Aw Aw Aw E E
<25 as: can: as: as: my: use: mac 63% as:
corn ba onb worn EU onb born by onb coin En onb corn in onb
oEoumon oEoumon 25:32 awn—Eon uoN2E0n n20
ANAL 2-2; 2: as: 2.3 A:

 

 

 

3:8 63a:

77

Table: Labor

 

Household Labor Hired Labor

 

activitie which # of people # of days # of # of days daily
month?
5 people wage“

 

ploughi

ng

 

manure
applicati
on

 

seeding

 

weeding

 

harvesti

n8

 

 

 

 

 

 

 

 

 

Which month?: In which month was the activity carried out? ‘
*If the wage is not pay per day basis, please specify the payment scheme below.

 

 

2.2.3 Water Management
If you use common water sources (common well/tank or canal) for another plot, do your
household members participate in water management activities?

1. Yes (what kind? and how much

 

Rs/ day)

2. No

78

2.2.4 Dry Season without Water
How large did you cultivate UNIRRIGATED area during LAST DRY SEASON (1997)?_

ha ( % of area rented IN %), Crop:

 

Total Output: kg Price/kg: Rs (if sold)

2.2.5 Agriculture during Rainy Season

How large did you cultivated LAST RAINY SEASON (1997)? ha ( % of area
rented IN__%)

What was the most important RAINY-SEASON CROP LAST YEAR (1997)?

Area Planted: ha ( % of area rented IN %)

High Yield Variety? 1=Yes 2=No

Total Output: kg Price: Rs (if sold)

2.2.6 High Yield Variety
When your household adopted high yield varieties, what was the most important source
of information?
1. extension service
2. radio and/or TV programs
3. newspaper and/or magazines
4. neighbors
5. dealers/merchants of agricultural inputs

6. others:

 

79

3 Forest
3.1 Forest Use

Does your household use forest for the following purposes?

Timber collection 1=Yes 2=No
Fuelwood collection 1=Yes 2=No
Fodder collection 1=Yes 2=No
Grazing 1=Yes 2=No
Grass collection 1=Yes 2=No
NTFP (non-timber forest products) collection 1=Yes 2=No

3.2 Forest Management Activities

Does your household members participate in forest management activities funded cy
PUBLIC SECTOR ?

1 = Joint Forest Management 2 = other government work 3 = none

If 1 or 2, fill in the table.

 

 

 

 

Activities (circle all # of people Total days of Total Revenue
applicable numbers) from your working (sum as wage in kind
in 1997 household of all
participants)
1 2 3 4 5 6 (wage rate
Rs/day)

 

 

 

 

 

1=Plantation, 2=Trench digging, 3=Pruning, 4=Thinning, 5=Forest watching, 6=Other
Does your household members participate in forest management activities OTHER
THAN the one mentioned above?

1=Yes (who organizes? ) 2=No

If yes, fill in the table.

80

 

 

 

 

 

Activities (circle all # of people Total days of Revenue
applicable numbers) from your working (sum as wage in kind
in 1997 household of all
participants)
1 2 3 4 5 6 (wage rate
Rs/day)

 

 

 

 

 

1=Plantation, 2=Trench digging, 3=Pruning, 4=Thinning, 5=Forest watching, 6=Other

3.3 Fuel

How often in a week does your household usually collect fuelwood?

 

times

Who and how many people usually goes to the forest to collect fuelwood?

 

 

 

How much (kg) does your household collect fuelwood on average in one collection?

kg

 

How many hours does your household spend collecting fiJelwood in one collection?

hours
Does your household use cow dung for cooking? 1. Yes ( %)
2. No
Does your household use kerosene for cooking? 1. Yes (___
__%) 2. No
Does your household use agricultural residues for cooking? 1. Yes (—
____%) 2. No

*Percentage against total energy consumption for cooking.

81

3.4 Grazing and Livestock

3.4.1 Grazing

 

Animal

current # of
animals

% of grazing

# of animals 10
years ago

% of grazing
10 years ago

 

Oxen

 

Cow

 

Buffalo

CS

 

Sheep

 

 

Goats

 

 

 

 

 

Who takes care of grazing?

1=household members

2=neighbors

3=grazers (specialists)

4=other

If 1=household member,

Who?

 

 

If 2 or 3, how much do you pay? (

__l

, How many people?

 

82

, How many hours/day?

 

3.4.2. Livestock Transaction

 

transa kind of animal numbe unit price
ction r (Rs)

purpose of
transaction

 

1

 

2
Sales in 1997

 

3

 

 

 

 

Purchase in 1997

 

 

 

 

 

 

 

 

4. Credit

Have you ever borrowed money from others?

 

 

 

 

 

1=Yes 2=No
most recent
If yes: When?
How much?
From who
For what purpose?
Interest?
If no: Why? 1=Not needed 2=Not available

83

second recent

 

 

 

 

 

3=Expensive

 

5. Household Income

Note: Ask about household income LAST YEAR (1997).

 

Income Source

Amount
(RS)

activity and wage
rate

 

farming (from sales of ag.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

products) N/A
Agn culture working for other
wage rate
household’s plot Rs/day
sales of forest products N/ A
Forest
working for J FM etc. wage rate
Rs/day
your own business activity
Non-farm activities in activity
the village
working for other’s wage
rate Rs/da
y
your own business activity
Migration act1v1ty
working for others wage
rate____Rs/d
ay
money from non- from
Remittance whom

 

household members

 

 

 

 

84

 

6 Household Assets

6.1 Asset Composition

 

Item

current qty

qty 10 years ago

 

Tractor

 

Cart

 

Automobile

 

Bike

 

Bicycle

 

TV

 

 

Radio

 

 

 

85

 

7 Health and Nutrition

7.1 Nutrition
How many days in a week do your household members usually take beans/pulse?_

days
How many days in a week do your household members usually take vegetables?_

days

 

How many days in a week do your household members usually take meat?
days

What meat is it:

 

Does your household purchase milk? 1. Yes 2. No
Does your household sell milk? 1. Yes 2. No
How many days in a week do your household members take dairy products (milk, cheese,
etc.)

in milking season? days per week for months” per a
year
How many days in a week do your household members take dairy products (milk, cheese,
etc.)

in non-milking season? days per week for __

months" per a year
*Total should be 12 months. Milking season is a period (a few months or a whole year)

when one can milk cows continuously, but not necessarily every day. Non-milking
season is a period when no milk is available from the household’s own animals.

86

7.2 Health
7.2.1 Drinking Water
What is your household’s major drinking water sources?
1. pond, lake, or dam
2. river, or spring
3. common well, or borehole
4. common hand pump
5. private well, or borehole
6. private hand pump
7. rainwater stored in tank
8. tap water from running water (is it disinfected? 1. yes 2. no)
9. buying from water seller

How ofien does your household collect drinking water in a day?

 

times
Who and how many people usually collects drinking water?

people

 

How many hours does your household spend collecting water in one collection?_

hours

87

7.2.2 Diseases

In the past one year, did your household members have any diseases for which he/she had
to go to health center/clinic/hospital? Please list all the cases.

 

 

 

 

 

 

 

(1) Who (2) In (2) Name (3) Where (4) (5) Total (6) How
(sex, age) which of disease did he/she Admitted expendit did you
month? or go? to the ure manage to
Househol symptom Type and hospital? pay the
(1 head? location If yes, expense?
how many
days?
1
2
3
4
5

 

 

 

 

 

 

 

(1) If he/she is the household head, please write so.

(3) Where did he/she go: answer the type of medical facility, such as health center, clinic,
hospital, and place where it is located (name of the city/town).
(4) Admitted to the hospital: If he/she had in-hospital care, answer “yes”. Otherwise,

“no”. If yes, answer how many days he/she stayed in the hospital.

(6) Self-financed or borrowed. If self-financed, then how did the household do it? (e.g.
saving, selling cows, selling wheat, and so on). If borrowed, then from whom did the
household borrow? (e.g. neighbors, informal money lender, formal bank, and so on).

88

 

FOR N ON-O WNERS
The diflerence between owner ’3 question and non-owner ’s question is only section 2.2

(question on tubewell) which is shown below. The rest of the questions are the same.

2.2 Irrigation (about the plot questioned in Agriculture section)

From which owner did you buy water?

 

(specify the well if the owner has more than one: )

Did you pay for water?

1=fiee
2=charged -> How much did you pay in total to irrigate the plot questioned
above? Rs (if pecuniary payment)

 

 

 

kg of (if in kind )

If 2=charged, how did you pay?

 

 

 

 

 

 

l=fixed rent per season (Rs . for ha)

2=crop sharing ( % of the total harvest)

3=mixture of 1 and 2 (Rs for ha plus % of
the harvest)

4=charge per hectare per application (Rs )

5=charge per hectare per season (Rs )

6=charge per hour of pump-use (Rs per hour)

7=other( )

 

How often did you apply water to the plot you are asked above?

times

89

How much did you apply water at each application?

 

 

 

 

1St application ha (or hour of pumping)
2"d application fihg (or hour 0f pumping)
3rd application ha (or hour of pumping)

 

 

(more? write answer below)

Operational Cost Paid by Yourself

1 =non

2=fuel (Rs )

3=Driver’s / Linemen salaries (Rs ) or ( days by yourself)

4=transport (Rs ) or ( days by yourself)

5=chanel digging (Rs ) or ( days by yourself)

6=other (Rs ) or ( days by yourself)

Did you have another plot irrigated by the same well?

1=Yes (how large? ha)

Opportunity of other activities

Choose the one best alternative opportunities from the following if the current well owner

refused to sell water to you. (Note this is a hypothetical question to the household who

actually used water last dry season.)

1=buying water from a different well owner (from whom?

how far km)

90

 

2=buying water from other water source (type of irrigation?

 

 

 

ha)

 

 

 

how far km)

3=farming without water (crop? how large?
4=non-farm family business (what kind? )
5=non-farm labor work in the village (what kind? wage rate?—
Rs/day)
6=seasonal migration (what kind? wage rate?__
Rs/day)

=other (what kind? wage rate? Rs/day)

 

91

BIBLIOGRAPHY

92

Allen, Douglas and Dean Lueck. 1992. “Contract Choice in Modern Agriculture: Cash
Rent Versus Cropshare.” Journal of Law & Economics 35(2) pp. 397-426.

Allen, Douglas and Dean Lueck. 1995. “Risk Preferences and the Economics of
Contracts.” American Economic Review 85(2) pp. 447-451.

Alston, Lee J., Samar K. Datta and Jeffrey B. Nugent. 1984. “Tenancy Choice in a
Competitive Framework with Transactions Costs.” Journal of Political Economy
92(6) pp. 1121-1133.

Bagachi, Kathakali S. 1995. Irrigation in India: History and Potentials of Social
Management, New Delhi: Upalabdhi.

Bardhan, Pranab K. 1984. Land, Labor, and Rural Poverty — Essays in Development
Economics, Oxford University Press: Delhi.

Basu, Kaushik. 1992. “Limited Liability and the Existence of Share Tenancy.” Journal of
Development Economics 38 pp. 203-220.

Bell, Clive, 1977. "Alternative Theories of Sharecropping: Some Tests Using Evidence
from Northeast India." Journal of Development Studies 13 pp. 317 - 346.

Bell, C. 1989. “A Comparison of Principal-Agent and Bargaining Solutions: The case of
Tenancy Contracts.” In P. K. Bardhan ed., The Economic Theory of Agrarian
Institutions, Clarendon Press pp. 73-92.

Bell, C. and Minhas Zusman. 1976. “A Bargaining Approach to Cropsharing Contracts.”
American Economic Review 66(4) pp578-88.

Bliss, C. J., and N. H. Stern. 1982. Palanpur: The Economy of an Indian Village. New
York: Clarendon Press.

Campbell, Donald. 1995. “Design and Operation of Smallholder Irrigation in South
Asia.” World Bank Technical Paper No. 256, Washington D. C., The World
Bank.

Cheung, Steven N. S. 1969. The Theory of Share Tenancy, University of Chicago Press.
Datta, Samar K., Donald J. O’Hara, and Jeffiey B. Nugent. 1986. “Choice of Agricultural
Tenancy in the Presence of Transaction Costs.” Land Economics 62(2) pp. 145-

158.

Dasgupta, Partha. 1993. An Inquiry into Well-Being and Destitution, Oxford: Clarendon
Press.

93

Dhawan, B. D. 1988. Irrigation in India ’s Agricultural Development: Productivity,
Stability, Equity, New Delhi: Sage Publications

Eswaran, Mukesh and Ashok Kotwal. 1985. “A Theory of Contractual Structure in
Agriculture.” American Economic Review 75(3) pp. 352-367.

Frisvold, George B. and Margriet F. Casswell. 1995. “A Bargaining Model for Water
Transfers Including Water Quality.” In Ariel Dinar and Edna Tusakloeman (eds.)
Water Quantity/Quality Management and Conflict Resolution: Institutions,
Processes, and Economic Analysis, London: Praeger.

Fujita, K. and F. Hossain. 1995. “Role of the Groundwater Market in Agricultural
Development and Income Distribution: A Case Study in a Northwest Bangladesh
Village.” Developing Economies 33(4) pp. 442-463

Hayami, Yujiro and Keijiro Otsuka. 1993. The Economics of Contract Choice: An
Agrarian Perspective, Oxford: Clarendon Press.

Ishikawa, Shigeru, 1975. "Peasant Families and the Agrarian Community in the Process
of Economic Development." In Lloyd Reynolds (eds.), Agriculture in
Development Theory, Yale University Press, pp. 451 - 496.

J anakarajan, S. 1992. “Interlinked Transactions and the Market for Water in the Agrarian
Economy of a Tamilnadu Village.” In Themes in Development Economics:
Essays in Honor of Malcolm Adiseshiah, ed S. Subramanian, Oxford University
Press.

Kahnert, Friedrich and Gilbert Levine, 1994. “Groundwater Irrigation and the Rural Poor:
Options for Development in the Gangeic Basin.” A World Bank Symposium.
Washington D. C.

Meinzen-Dick, Ruth, 1997. “Groundwater Markets in Pakistan: Participation and
Productivity.” Research Report No. 105, Washington D. C., International Food
Research Institute.

Nash, J. R. 1950. “The Bargaining Problem.” Econometrica 18, pp. 155-62.

Newbery, David M. G., and Joseph E. Stiglitz. 1979. “Sharecropping, Risk-sharing, and
the Importance of Imperfect Information.” In J. A. Roumasset, J. M. Boussard,
and 1. Singh (eds.), Risk, Uncertainty, and Agricultural Development, Ch. 17.
New York: Agricultural development Council.

Otsuka, Keijiro, Masao Kikuchi, and Yujiro Hayami, 1986. "Community and Market in

Contract Choice: J eepney in the Philippines', " Economic Development and
Cultural Change, 34(2), pp. 279 - 98.

94

Pant, N. 1991. “Ground Water Issues in Eastern India.” In Future Directions for Indian
Irrigation: Research and Policy Issues, ed. R. Meinzen-Dick and M. Svendsen.
Washington D. C.: International Food Policy Research Institute.

Pant, N. 1992. New Trend in Indian Irrigation, New Delhi: Ashish Publishing House.

Rao, C. H. Hanumantha. 1971. “Uncertainty, Entrepreneurship, and Sharecropping in
India.” Journal of Political Economy 79(3) pp 578-595.

Ray, Debraj. 1998. Development Economics, Princeton University Press.

Robertson, A. F. 1982. “Abusa: The Structural History of an Economic Contract.”
Journal of Development Studies 18(4) pp. 447-478.

Rosegrant, Mark W., 1997. “Water Resources in the Twenty-First Century: Challenges
and Implications for Action.” Food, Agriculture, and the Environment Discussion
Paper No. 20, Washington D. C.. International Food Research Institute.

Rosenzweig, Mark R. 1988. “Risk, Private Information, and the Family.” American
Economic Review 78(2) pp. 245-250.

Rosenzweig, Mark R. and Kenneth I. Wolpin, 1993. "Credit Market Constraints,
Consumption Smoothing, and the Accumulation of Durable Production Assets in

Low-Income Counties: Investments in Bullocks in India." Journal of Political
Economy, 101(2), pp. 223 - 244.

Roumasset, James and Marilou Uy. 1987. “Agency Costs and the Agricultural Firm.”
Land Economics 63(3) pp. 290-302.

Salanie, Bernard. 1997. The Economics of Contracts, The MIT Press.

Saleth, R. Maria, 1998. "Water Markets in India: Economic and Institutional Aspects." In
Markets for Water: Political and Performance. K. William Easter, Mark W.
Rosegrant, and Ariel Dinar (eds.) Massachusetts: Kluwer Academic Publishers.

Shaban, Radwan Ali, 1987. "Testing between Competing Models of Sharecropping."
Journal of Political Economy 95(5) pp. 893 - 920.

Shah, Tushaar, 1993. Gropund Water Markets and Irrigation Development, Bombay:
Oxford University Press.

Shah, Tushaar and Vishwa Ballabh. 1997 “Water Markets in North Bihar: Six Village

Studies in Muzaffarpur District.” Economic and Political Weekly 27(12) pp. A183
- A190.

95

Sharma, Naresh and Jean Dreze. 1996. “Sharecropping in a North Indian Village.”
Journal of Development Studies 33(1) pp. 1-39.

Sengupta, Kunal. 1997. “Limited Liability, Moral Hazard and Share Tenancy.” Journal
of Development Economics 52 pp. 393-407.

Shetty, Sudhir. 1988. “Limited Liability, Wealth Differences and Tenancy Contracts in
Agrarian Economics.” Journal of Development Economics 29 pp. 1-22.

Singh, Nirvikar, 1989. “Theories of Sharecropping.” In P. K. Bardhan ed., The Economic
Theory of Agrarian Institutions, Clarendon Press pp. 33-72.

Stiglitz, Joseph E., 1974. “Incentives and Risk Sharing in Sharecropping.” The Review of
Economic Studies 41(2) pp. 219-255.

Stiglitz, Joseph E, 1987. “Sharecropping.” in The New Paragrave: A Dictionary of
Economic Thought and Doctrine, ed. J. Eatwell, M. Milgate, and P Newman ,
MacMillan: London. vol. 3 pp. 320-23

Strauss, John and Duncan Thomas, 1998. "Health, Nutrition, and Economic
Development." Journal of Economic Literature, 36 pp. 766 - 817.

Tunali, Insan. 1993. “Choice of Contracts in Turkish Agriculture. ”Economic
Development and Cultural Change 42(1) pp. 67-86

Walker, Thomas S. and James G. Ryan. 1990. Village and Household Economics in
India 's Semi-arid Tropics. The Johns Hopkins University Press.

96