University of Southern California Dissertations and Theses

Essays on agrarian institutions

(USC Thesis Other)

Essays on agrarian institutions

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ESSAYS ON AGRARIAN INSTITUTIONS by Sripad Motiram A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements of the Degree DOCTOR OF PHILOSOPHY (ECONOMICS) August 2002 Copyright 2002 Sripad Motiram Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3094361 UMI UMI Microform 3094361 Copyright 2003 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA The Graduate School U niversity Park LOS ANGELES, CALIFORNIA 900894695 This dissertation, w ritten b y MDTffcA-M __ U nder th e direction o f hj..?... D issertation Com m ittee, and approved b y a ll its m em bers, has been p resen ted to and accepted b y The Graduate School, in p a rtia l fulfillm ent o f requirem ents fo r th e degree o f DOCTOR OF PHILOSOPHY Dean o f Graduate Studies D ate DISSER TA TION COM M ITTEE^ _______ Chairperson Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS I would like to thank all the members of my thesis committee for their advice and support. I would like to thank my chairs Prof. James Robinson and Prof. Jeffery Nugent for their regular guidance, without which my dissertation would not have seen the light of the day. It would be difficult for me to express my gratitude in words, but I would like to thank Vamsi for his continuous feedback and support during my Ph.D., especially during the fieldwork stage. I would like to thank USC and the Haynes Foundation for providing me with fellowships, without which it would not have been possible for me to go to India and conduct fieldwork. For his comments on Chapter 1,1 thank Prof. Herbert David at USC. A preliminary version of Chapter 1 was presented at USC Graduate Students Workshop in September 1999, the NEUDC Conference at Harvard University in October 1999 and the Cecele and Michael Birnkrant seminar in Economic Development at USC in November 1999.1 would like to thank all the participants of the above for their comments, especially Prof. Oriana Bandiera, who was the discussant for my paper at the NEUDC conference. For their hospitality and co-operation during my fieldwork, I would like to thank all the villagers of Gangadevipalli, Machapur and Kothamolgara. Without their support, my fieldwork would not have been possible. I would like to specially thank all the laborers and employers who took time away from their work and gave thoughtful answers to my questions. Several people, both academics and non- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iii academics, helped me during the course of my fieldwork. I take this opportunity to thank all of them. Space permits me to name only a few - Mr. K. Rajamouli, Mr. Butchanna, Mr. Dhan Raji Reddy, Mr. Joji Reddy, Mr. Sardar Beg, Mr. Madhu, Mr. Murthy, Prof. D. Narasimha Reddy of University of Hyderabad, the Village Resurvey Group in Hyderabad and Warangal, Prof. Samuel Bowles of UMASS Amherst, Mr. Venku Reddy of PRDIS, Mr. Vijay Mahajan and Mr. Y.V. Raghunath Reddy of BASIX. A preliminary version of Chapter II was presented in October 2001 at USC, in the Graduate Students Workshop and in the Cecele and Michael Birnkrant seminar in Economic Development. I would like to thank all the participants, whose comments greatly improved the Chapter. I presented some of my ideas on Telangana at the Annual Conference on South Asia at University of Wisconsin at Madison in October 1999.1 thank all the participants for their comments, which helped me clarify my ideas before I undertook my fieldwork. I have received and continue to receive enormous support from my family, Vamsi and his family, and I would like to thank them for this support. Prof. James Robinson and his family were very generous hosts on my several trips to Berkeley and I take this opportunity to thank them. I would also like to thank several friends at USC (especially the Bonsallo/Estrella community, the students who started their Ph.D. programs in economics, along with me in 1996, and the students in Prof. Nugent’s regular meetings), who have provided me with motivation and support at various stages in my Ph.D. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iv TABLE OF CONTENTS ACKNOWLEDGEMENTS ii LIST OF TABLES vii LIST OF FIGURES ix ABSTRACT X INTRODUCTION 1 1 . Description of the Essays 1 2. Common Themes 4 2.1 Approach towards Institutions 4 2.2 Agrarian Power 5 2.3 Imperfections in Markets 9 2.4 Strategic Behavior 9 CHAPTER I: INTERLINKING AND COLLUSION 11 1. Introduction 11 2. Model 18 2.1 Collusion in the Credit Market 19 2.2 Collusion in the Tenancy Market 21 2.3 Interlinkage 23 3. The Model with Frictions 25 4. Conclusions 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V CHAPTER II: CONTRACTUAL CONDITIONS OF ATTACHED LABOR: RESULTS FROM AN INDIAN SURVEY 36 1 . Description of the field-work region 37 2. Literature Survey and Gaps Addressed 43 2.1 Literature Survey of Empirical Work on Attached Labor 43 2.2 Gaps Addressed 54 3 Data Analysis and Results 57 3 .1 Methodology and Description of the Data 57 3.2 Details of the Institution 62 3.3 Analysis and Inferences 63 3.3.1 Credit Relations and the Determinants of the Choice of Attached Labor 63 3.3.2 Tasks Performed, Remuneration and Bargaining Power 68 3.3.3 Default on the Attached Labor Contract 74 3.3.4 Mobility 77 3.3.5 Collective Bargaining and Collusion 83 3.3.5.1 Findings 83 3.3.5.2 Analysis and Inferences 88 3.3.6 Some Other Remarks on the Status of Attached Laborers 90 4 Conclusions 91 CHAPTER 3: A MODEL OF TIED-LABOR, EMPLOYER RELATIONS 95 1. Introduction 95 2. Literature Survey 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vi 3. A Simple Model 110 3.1. Description of the Economy 111 3.1.1 Time and Seasonality 111 3.1.2 Laborers and Labor Contracts 111 3.1.3 Landlords 113 3.1.4 Timing and Sequence of Events 115 3.2 Determination of the Equilibrium 118 3.3 Results 119 4. Extensions 130 4.1 Alternative Source of Credit 130 4.2 Credit Offer to Retain the Tied-laborer 135 4.2.1 The Laborer’s Problem 136 4.2.2 The Employer’s Problem 136 5 Conclusions 141 BIBLIOGRAPHY 143 APPENDIX I QUESTIONNAIRE ADDRESSED TO ATTACHED LABORERS 150 APPENDIX II QUESTIONNAIRE ADDRESSED TO EMPLOYERS 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vii LIST OF TABLES Table 1: Details of Warangal and Mahbubnagar Districts 42 Table 2: Details of the Field Work Villages 44 Table 3: Some Important Variables 59 Table 4: Descriptive Statistics for Some Variables 59 Table 5: Statistics for Dummy Variables 60 Table 6: Reasons for Entering Attached Labor 64 Table 7: Details of Borrowing by Attached Laborers 64 Table 8: LOGIT Analysis of the Probability of a Household Having At Least One Attached Laborer in the Age Group 13-75 in Warangal 66 Table 9: LOGIT Analysis of the Probability of a Household Having at least One Attached Laborer in the Age Group 10-75 in Warangal and Mahbubnagar 66 Table 10: Number of Agricultural Laborers Performing Various Tasks 69 Table 11: Linear Regression Analysis of the Determinants of Remuneration of Attached Laborers in Gangadevipalli 70 Table 12: Linear Regression Analysis of the Determinants of Remuneration of Attached Laborers in Gangadevipalli and Machapur 71 Table 13: Linear Regression Analysis of the Determinants of Remuneration of Attached Laborers in Warangal and Mahbubnagar 71 Table 14: LOGIT Analysis of the Probability that an Attached Laborer Moves from the Current Employer Next Year in Gangadevipalli 79 Table 15: LOGIT Analysis of the Probability that an Attached Laborer Moves from the Current Employer Next Year in Gangadevipalli and Machapur 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 16: Average Debt to Employers and Mobility of Attached Laborers in Warangal 80 Table 17: LOGIT Analysis of the Probability that an Attached Laborer Works for an Employer for More than 1 Year (Continuously) in Gangadevipalli 82 Table 18 : LOGIT Analysis of the Probability that an Attached Laborer Works for an Employer for More than 1 Year (Continuously) in Gangadevipalli and Machapur 82 Table 19: LOGIT Analysis of the Probability that an Attached Laborer Works for an Employer for More than 1 Year (Continuously) in Warangal and Mahbubnagar 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES Figure 1 : Location of Andhra Pradesh State in India 38 Figure 2: The Telangana Region in Andhra Pradesh 39 Figure 3: A Schematic Representation of Gangadevipalli and Machapur 60 Figure 4: Participation Constraint of the Laborer 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT My dissertation deals with agrarian institutions in developing countries and comprises of three essays. In the first essay, I suggest a new rationale for the existence of interlinked contracts in agriculture. Using the framework of an infinitely repeated game with discounting, I show how offering interlinked contracts can help the dominant parties to collude in cases where collusion is not possible with non interlinked contracts. This occurs either because interlinkage pools incentive constraints across markets, or because it affects the incentives of agents to accept deviating contracts. I illustrate these mechanisms by considering the case of interlinkage between markets for credit and share tenancy. The second essay, based on my fieldwork in the state of Andhra Pradesh in South India, deals with attached labor contracts in agriculture, and explores certain issues that have not been adequately dealt with in previous contributions. I investigate the reasons for laborers to enter into these contracts, and analyze the factors that impact their bargaining power (as measured by remuneration) and their mobility (ability to move away from current employers to other employers or to other occupations). I show that the likelihood that a household has an attached laborer depends upon the level of indebtedness, total borrowing, and availability of other income. I also show that while bargaining power depends upon caste, overall indebtedness, and some other individual/household characteristics, mobility depends upon the level of indebtedness to the current employer. I also analyze default on the attached labor contract and collective bargaining among attached laborers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In the third essay, I develop a game-theoretic model of tied labor that formalizes the mobility of laborers, strategic interaction of employers, and implication of credit markets for the above. These are issues that have not been addressed adequately in the theoretical literature. Using this model, I explain some findings from my fieldwork and from other studies. I show that the likelihood that a laborer moves away depends negatively upon indebtedness to the employer. I also derive interesting comparative static predictions about mobility of tied-laborers and about the behavior of employers in case they can make credit offers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 INTRODUCTION This dissertation deals with agrarian institutions in developing countries, institutions being defined as, “social rules, conventions, and other elements of the structural framework of social interaction.”1 The focus of the three essays that comprise this dissertation is on two institutions that are found in many developing countries. Apart from the broad area of analysis, there are certain themes that recur in these essays. In what follows, I will briefly explain the issues explored in these essays and also highlight these common themes. 1. Description of the Essays The first essay tries to understand collusion in the agrarian economy, a feature that has been pointed out by social scientists belonging to various disciplines, in different parts of the world, but not formalized adequately in the literature. In this essay I argue that collusion is related to interlinking of markets, another feature that characterizes agriculture in many developing countries. An interlinked contract is one in which two or more interdependent exchanges are simultaneously agreed upon. Using the framework of an infinitely repeated game, I show that collusion can occur with interlinked contracts in situations where it cannot occur with non interlinked contracts. W hile, I consider interlinking of land and credit markets to illustrate this, the argument can easily be generalized to other forms of interlinking. 1 This definition is from Bardhan and Udry (1999), pp. 217. 2 This definition is from Basu (1997), pp. 281. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 Apart from formalizing collusion, this essay also throws light on strategic aspects of interlinking, an issue that has received relatively less attention in the literature on interlinking. The second essay deals with attached labor contracts and is based upon fieldwork conducted by me in the Telangana region in Andhra Pradesh State of India. These are contracts in which laborers work for one particular employer for long periods of time and these should be contrasted with “Casual Labor” contracts where laborers are hired on a daily basis. I focus on annual attached labor contracts and explore the factors that determine the choice of these contracts and the bargaining power of attached laborers. I also try to understand the ability of attached laborers to leave their employers (either to other employers or to occupations other than tied-labor) and the determinants of the duration that they spend with their employers. This analysis cannot resolve methodological debates on freedom (although it is intimately linked to the issue of freedom), however, it does give insights into the functioning of this institution and into the nature of changes that have occurred. I also explore the issues of collective bargaining and collusion among attached laborers and collusion among employers, to throw light on how these processes work in some real-life contexts. The third essay formalizes the phenomenon of tied-laborers3 moving from one employer to another or from tied-labor to other occupations. This is a 3 As I point out later, in the empirical literature, especially on India, the term “attached labor” is much more common than other terms like “tied-labor,” “permanent labor,” “regular farm labor” etc, which Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 phenomenon that I observed in my fieldwork, and that several people who have studied other contexts have observed. A consequence of the above is that the duration that a laborer spends with a particular employer comes down. This is an important phenomenon because it reflects to a certain extent the erosion of patronage4 in the relationship between tied-laborers and employers. This process is intimately linked to strategic behavior on the part of both employers and laborers, and the functioning of credit markets. Despite the wide empirical support and the importance pointed above, this issue has not received attention in the theoretical literature on tied-labor. Hence, I formalize the above using the framework of search and matching and incorporate strategic behavior and credit markets. Using this model, I derive some interesting comparative static predictions and predictions regarding strategic behavior on the part of employers. As I mentioned above, apart from the fact that all these essays deal with agrarian institutions, there are some themes that are common to them. I will briefly discuss these below. refer to similar arrangements. In the theoretical literature, the term tied-labor is more common. Hence, I will use the term attached labor in the second essay and the term tied-labor in the third essay. 4 1 am using the term “patronage” in the sense in which it is used in the literature on agrarian relations, to capture certain features - provision of subsistence guarantees, display of loyalty, display of generosity etc. (on this issue, see Breman (1974) and Hardiman (1996)). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 2. Common Themes 2.1 Approach towards Institutions In all these essays, I take the position that institutions should not be taken as given, but the rationale for their existence should be explained. I try to explain institutions by looking at the purposes that they serve. This approach I believe does provide insights, even though it has certain problems.5 1 also take the position that while trying to understand the economic rationale of institutions one cannot ignore approaches and insights provided by other social sciences. In these essays I have used the work of people who have worked in various disciplines - history, social anthropology, sociology etc. I have done so both to understand how these institutions can be conceptualized and also to obtain concrete details from particular contexts. These two aspects of the methodological position - search for an economic rationale and an emphasis on an interdisciplinary approach are quite common in the literature that deals with agrarian institutions (e.g. see Bardhan (1989), Stiglitz (1989)). The above approach is most evident in my first essay, which deals with interlinking and collusion. I try to provide a rationale for interlinked contracts by arguing that these contracts exist because they serve a particular function - facilitating collusion among the dominant parties in an agrarian economy. The impetus for offering interlinked contracts comes from the dominant players (landlord-moneylenders) and they offer these contracts because these facilitate 5 There are of course many valid critiques of a “functionalist” methodology. For a discussion on this issue in the specific context of agrarian institutions, see Bardhan (1989) pp. 1-17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 collusion and thereby increase the surplus that they can obtain. In my second essay on attached labor contracts in Telangana, while I try to understand various aspects of these contracts, I also try to understand why these contracts exist in the first place. I try to argue that attached labor contracts exist because they serve a purpose both for the laborers entering into these contracts and for employers hiring these laborers. In the third essay, to make the analysis simple, I do not explicitly model the co existence of tied-labor and casual labor contracts. However, this model can be extended to deal with this issue. 2.2 Agrarian Power All these essays deal with some form of agrarian power. Since “power” is a term that has a wide variety of meanings, I would like to clarify the sense in which I use this term. While several economists have remarked that power plays an important role in the agrarian economy, only a few of them have tried to specify what power means and formalize it. There is almost unanimous agreement among the above that conceptualizing and formalizing power is a very difficult task and developing a single model that can bring out all its manifestations is probably not possible.6 1 have hence attempted to present a few ways in which a particular institution facilitates the exercise of power. In the first essay on interlinking and collusion, I use two notions of power. The first is the notion popularized by Michael Taylor (see Taylor (1982)) according 6 For some issues involved in the conceptualization of power in economics, see Bardhan (1991). For some references on agrarian power, see Desai et. al. (1984). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 to which, “ A has power over B if A can affect the incentives facing B in such a way that it is rational for B to do something he would not have otherwise chosen to do.’M affects the incentives of B mainly through the offer of a reward or the threat of a penalty, or some combination of a threat and an offer. This way of defining power has the advantages of allowing the study of power even when transactions are voluntary (as most economic transactions are) and helping us to quantify power (e.g. on the lines suggested by Harsanyi7 ). The second notion of power is the more standard one, which has been used by economists both implicitly and explicitly for a long time. According to this, the amount of power A has over B depends upon the amount of surplus that A can get from B. Thus a perfectly price discriminating monopolist or a monopolist who can offer bundled deals has more power over the consumer than an ordinary monopolist, who in turn, has more power than a perfectly competitive seller. In a duopolistic setting, we can model a leader as having more power than the follower. I will present the details of the models used for exploring interlinking and collusion later, but I will briefly explain how these notions of power will be used. First, the landlord-moneylender who can offer interlinked contracts has more power 7 See Harsanyi (1976) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 over the peasants than the pure moneylender or the pure landlord because he8 can extract more surplus from the peasant. Second, and more interestingly, in the case where there is one landlord-moneylender who can offer interlinked contracts, I show that by doing so, he can “force” the other moneylender to collude with him on the credit market. In other words, by offering the interlinked contract the landlord- moneylender can affect the incentives of the other moneylender in such a way that he will not defect on the credit market (something that he would have done in case the landlord-moneylender would have offered non-interlinked contracts). In this sense, the landlord-moneylender who can offer interlinked contracts exercises power over the moneylender who can only offer non-interlinked credit contracts. Similarly, in my third essay, which deals with tied-laborer and employer relations, I show how the ability of the employer to make “take-it-or-leave-it” credit offers confers power to him over the tied-laborer because by making this offer, he can affect the probability that the laborer will stay with him. In the second essay, in my analysis of collective bargaining among attached laborers in Warangal district, I show how a “Smithian” concept of power can be used to illustrate the conflict between employers and tied-laborers. Adam Smith argues that wealth confers power because those who are wealthy can sustain themselves longer in a conflict (apart from other reasons). He points out the following about the conflict between masters and workmen, 8 1 will use the pronoun “he” throughout the dissertation to refer to “he or she.” This is because in many cases the people (e.g. attached laborers, landlords) I am referring to are men. The substitution of “he or she” can be done wherever appropriate. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 It is not, however difficult to foresee, which of the two parties must, upon all occasions, have the advantage in the dispute, and force the other into compliance with their terms. The masters, being fewer can combine much more easily; and the law, besides, authorizes, or at least does not prohibit their combination, while it prohibits those of the workmen. We have no acts of parliament against combining to lower the price of work; but many against combining to raise it. In all such disputes, the masters can hold out much longer. A landlord, a farmer, a master manufacturer, or merchant, though they did not employ a single workman, could generally live a year or two upon the stocks they have already acquired. Many workers could not subsist a week, few could subsist a month, and scarce any a year without employment. In the long run, the worker may be as necessary to the master as the master is to him, but the necessity is not so immediate.9 Using the above idea, I show that the employers exercise power over the tied- laborers because the tied-laborers cannot last long in a conflict with them. This is further compounded by the fact that the indebtedness of the tied-laborers weakens them in this conflict. While it can be argued that employers have more power than the tied- laborers, it has become very clear from the work of several people both in agrarian contexts and other contexts, that this one-dimensional view of power is at best incomplete because the exercise of power co-exists with resistance to it.1 0 As pointed out by James Scott (see Scott (1985)) this resistance could manifest itself in subtle forms, as “Weapons of the Weak.” I have tried to throw light on some of these processes in my second essay. I believe that the attempts of tied-laborers to bargain collectively (even when they are unsuccessful), their desire to move from one 9 Smith (1776), pp. 169 1 0 On this issue, the seminal work is of course that of French philosopher Michael Foucault (see Foucault (1991)). Also see Scott (1990), in the context of agrarian studies an interesting reference on this issue is Hardiman (1996). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 employer to another, or to leave tied-labor, and their tendency to show disloyalty (by not reporting to work, or by defaulting altogether) can be understood in these terms. 2.3 Imperfections in Markets In all these essays, I deal with imperfections in markets. In the first essay, I consider the case where there are frictions in the land market. In the third essay, I consider a situation where there are imperfections in both credit markets and land markets. Imperfections in credit markets are modeled by assuming that there are limits on the amount that the laborer can borrow, whereas imperfections in land markets are modeled by assuming that every laborer who wants to work with a landlord cannot do so, and vice-versa. The reason for focusing on imperfections is that there is enough evidence to suggest that markets in agrarian economies are characterized by such imperfections, which are reflected in the form of transaction costs, information costs etc. 2.4 Strategic Behavior In all these essays, I deal with “Strategic Behavior,” i.e. I deal with situations where agents take into account the consequences of their actions and the actions of others, and act “strategically.” They do not behave like agents who live in a perfectly competitive world. There are two reasons for this. First, there is evidence to believe, based upon my own fieldwork and work by others from other contexts, that strategic considerations play an important role in the agrarian economy. Second, information and transaction costs and other imperfections in markets, which are very common in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 rural settings, imply that even when there are a large number of agents (landlords, laborers etc.) it is more realistic to work with models of imperfect competition.1 1 Once, we work with models of imperfect competition (e.g. Oligopoly) strategic behavior assumes importance. The focus on collusion in the first essay and to a certain extent in the second essay is dictated by similar considerations. In this context, it is worthwhile to point out the observation of Stiglitz (see Stiglitz (1989)) that, Not only may imperfect information result in imperfect competition, but the long-term relationships (and relatively high barriers to exit) that characterize village life in many LDC s are particularly conducive to establishing collusive outcomes. 1 1 On this issue, see Stiglitz (1989) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 CHAPTER I INTERLINKING AND COLLUSION 1. Introduction In this chapter, I study the issue of collusion in the agrarian economy, an issue that has received relatively little attention in the development literature. This is surprising because social scientists belonging to several disciplines have pointed out the existence of collusive behavior in the agrarian economies of several parts of the world. In his study of Western India, David Hardiman (see Hardiman (1996), pp. 142) points out that Even here, the peasants seemed to be at a disadvantage. Haats tend to be dominated by petty Baniyas and other traders who worked together to keep purchase prices depressed. Similarly in his study of rubber plantations in Malaya, Wharton (see Wharton (1962)) observes that, In certain rural areas, additional power factors may be found, though probably less frequently. Other factors are ...Lastly, the cultural factor, either among buyers fostering collusive behavior or among sellers causing them to sell on particularistic or traditional basis rather than on price basis. In their study of southern U.S. after emancipation, Ransom and Sutch (see Ransom and Sutch (1977) pp. 147) discovered that, The close social ties between planters and merchants in many areas also suggest that competition between nearby merchants might be avoided by tacit or explicit collusion. In areas where several merchants shared a territory, only one or two merchants needed to be party to a noncompetitive agreement. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 According to the Montgomery Daily Advertiser in 1868, some Southerners felt that merchants colluded for ‘fixing the value of their wares’. While the above examples deal with collusion among traders and trader-creditors, there is enough evidence of other kinds of collusion. For example, Lacey (see Lacey (1981) pp. 163) observes that full-time servants in the 1930 s in South Africa, had limited options even within their own districts. Not only would a farmer be reluctant to poach his neighbour’s servants, but the farmers together tacitly agreed about keeping wages down at the same low level. Some other instances of collusion are pointed out by Bhadhuri (1973), Sarap (1987), and Rudra (1984).1 In fact collusion can be seen as a way in which agrarian power can be exercised.2 Much of this apparent ability to collude might be explicable in terms of cultural, caste and kinship factors. However, the fact that this has occurred in such a wide variety of settings raises the question of whether there are any structural features of agrarian economies that facilitate collusion.3 Though many authors have studied various forms of contracting and market institutions in the agrarian economy, these have rarely been examined from the point of view of their effect on the ability of agents to collude. In this chapter, I explore whether one 11 found from my fieldwork in Gangadevipalli village in Warangal district (in the state of Andhra Pradesh in India) that many tied-laborers believe that employers collude in setting their salaries. 2 On this issue, see Rao (1999), who argues that Agrarian power is exercised by individuals and sometimes by tacit or open coalitions. It is exercised in both personalised and institutionalised relationships; it's instrumentalities range from the sheer pressure of labourer's survival imperative to the long arm of law and order; it is bound up with ideologies of local patronage but has increasingly resorted to the trans-local politics of caste and class. 3 1 am not, however, implying that collusion is much more common in agrarian economies as compared to industrial economies. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 salient feature of many such economies, namely the interlinking of various “markets,” may facilitate collusion. There are two reasons why this exercise could be insightful. First, there is at least some evidence to suggest that interlinking could have something to do with collusion. In his study of sugarcane production in Gorakhpur,4 during the late colonial period, Shahid Amin (see Amin (1983)) shows how the sugar mills, which could not directly collude, used an intermediary to do so. Rather than bidding for the sugarcane from the peasants directly, these mills used an intermediary to get sugarcane at pre-determined prices. This intermediary happened to be a “powerful man,” a landlord-moneylender-trader, who used this position to make the peasants deliver sugarcane at prices determined by him and prevented them from getting any benefit from competition between the mills. Similarly, Sarap (see Sarap (1991), pp. 133) in his study of the Sambalpur district in the state of Orissa in India points out that - Moreover, the tenant’s own landlord is not the main source of borrowing for the linked tenants. It is noteworthy that from whichever informal source the linked tenant households have borrowed, they have had to pay high rates of interest. This signifies that there is some sort of implicit collusion among the landlord-creditors Another example is from Hardiman’s study cited above. It is interesting to quote Hardiman at length, to see how the merchant-usurers used their position vis-a-vis the peasants (see Hardiman (1996), pp. 282) 4 A region in the north Indian state of Uttar Pradesh. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 For example, there was a large increase in the area under cotton in Bhavnagar state in Saurashtra during the first two decades of the twentieth century. Marketing was however monopolized by Baniya Sahukars, who forced their clients to hand over their cotton in repayment of debts and recorded very low prices in their account books so as to prevent any diminution in their debts. During the boom in cotton prices at the end of First World War, these Baniyas united to force their clients to consent to even harsher terms. The second reason why examining the relation between collusion and interlinking could be interesting is that very few studies of interlinking have looked at the “strategic” aspects of interlinking i.e. the relationship between interlinking and competition. Most formal models of interlinking have assumed a monopolist, who can offer interlinked contracts (e.g. Bardhan (1984) pp. 86-91, Gangopadhyay and Sengupta (1987)). In light of the above, I explore the relationship between interlinking and collusion using the standard framework of an infinitely repeated game with discounting. I will explain the structure of the game later, but I will briefly present the results and the intuition. To fix ideas, I concentrate on two markets, the credit market and the market for share tenancy. For these two markets, I consider the cases where non-interlinked and interlinked contracts are offered and examine the conditions under which it is possible for competitors to collude. The first result is that if two agents (whom we call “moneylenders”) compete in interlinked contracts on both markets simultaneously, then this can allow them to collude in both credit and share tenancy markets, while if they had not interlinked, they could have colluded in neither market. This is due to the fact that interlinkage pools incentive Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 constraints. The second result stems from seeing that, when there are market frictions, interlinkage may aid collusion by making peasants reluctant to accept deviating offers. Consider the credit market with two moneylenders attempting to collude. If one of them interlinks into the tenancy market, then if the non-interlinked moneylender attempts to undercut he must induce peasants to accept a deviating contract. To accept this better contract in the credit market, a peasant with an interlinked contract must also give up his tenancy contract. This is costly if there are frictions in the tenancy market. Thus interlinking by one moneylender into a market where there are frictions can allow collusion to take place in the credit market where otherwise it would not have been possible. There are broadly two strands in the literature on the theory of interlinking. The first highlights the efficiency aspect of interlinking i.e. its focus is on showing how interlinking can be seen as an efficient, albeit second-best outcome in the presence of moral hazard, transaction costs etc. The second highlights how interlinking can be used by the dominant parties to consolidate their power and extract surplus that cannot be extracted using one market alone.5 Some examples in the first class of models are Braverman and Stiglitz (1982), Bardhan (1984, Ch. 6), Kotwal (1985), Datta, Nugent, Tishler and Wang (1988)) etc. Braverman and Stiglitz analyze a tenancy and credit interlinked contract in a situation characterized by moral hazard. They show that the landlord, by changing the terms and amount of the loan offered to the tenant, can induce the tenant to work harder. By doing so, the 5 This classification of the literature on interlinking is based on Bardhan (1989). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 landlord can also induce the tenant to undertake projects that are more desirable to the landlord, i.e. adopt techniques that offer higher mean return, but that have higher variance (in the absence of proper incentives, the tenant might adopt techniques that are “too safe” from the landlord’s point of view). Bardhan (1984, Ch 6) considers a situation where interlinking of consumption credit and labor markets, saves transaction costs and ensures double coincidence of wants. There is a slack season, when there is no work and production and hence no income for the laborer, and a peak season when there is work and income. The laborer borrows from the creditor- employer in the slack season, with a promise to work for the employer in the peak season. This arrangement is advantageous for the laborer because credit is available only from the employer-creditor (since the laborer cannot offer collateral) and by entering into this arrangement, the employer assures himself labor in the peak season.6 This contribution falls into the second strand and is related to a number of other contributions. Within the development literature, Basu and Bell (see Basu and Bell (1991)) use the framework of a fragmented duopoly to show how offering an interlinked contract could create barriers to entry. Ray and Sengupta (see Ray and Sengupta (1989)) look at the pattern of competition that is predicted by the theory of interlinkage and allude to the relationship between interlinking and collusion (as does Bardhan (1989)), without however, formally exploring it. Within the industrial 6 A very interesting example of interlinking of labour and credit from a non-agrarian context is provided by Datta and Nugent (see Datta and Nugent (1985)). This example deals with pearling industry in Bahrain and interlinking essentially serves the purpose of reducing transaction costs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 organization literature, part of our work is related to that of Bernheim and Whinston (see Bernheim and Whinston (1990)), who use to a repeated game to show how firms may be able to collude more effectively if they compete simultaneously on more than one market. One of the contributions of this analysis is to show that these results have important implications for interlinkage. The first result on interlinking and collusion is an application of the idea developed by Bernheim and Whinston, to the agrarian economy. An interesting issue raised in the context of Bhaduri’s model (see Bhadhuri (1973)) of technological stagnation concerns the extra power that is conferred to the dominant agent by offering interlinked contracts. A valid criticism of Bhaduri’s model is that it does not provide a convincing answer to the above question (on this issue, see Braverman and Stiglitz (1982) and Basu (1997)). The models developed below clearly show the extra power conferred to the dominant agent by the ability to offer interlinked contracts. In both the models below, the ability to offer interlinked contracts allows the dominant agents to collude under conditions where such collusion is not possible with non-interlinked contracts. Thus, the dominant agents can extract more surplus from the peasants by offering interlinked contracts as compared to the case where they can only offer non-interlinked contracts. The remaining part of the chapter is organized into three sections. Section 2 develops a model of interlinking without market frictions. Section 3 then extends this analysis by introducing frictions into the tenancy market, and Section 4 concludes. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 2. The Model We will consider an infinite horizon economy in discrete time. There is a mass iV of identical pea-sants and initially two “moneylenders.” It is assumed that the peasants attempt to maximize the discounted sum of consumption of a numeraire good, where dp is the discount rate of a peasant. The moneylenders are risk neutral and they maximize their lifetime profits. Si and 6 2 are the discount rates of the two moneylenders. There are two markets, one for share tenancy and one for credit. A share tenancy contract involves being given a piece of land from which output/ can be produced. The contract specifies the share a, e [0,l]of the output that the tenant retains, with the landowner getting a share (I- at ). It is assumed that/is a strictly increasing and concave function of the amount of credit that a tenant takes, denoted it with f(0)=0. Thus to be productive, a share tenant must also enter the credit market. For ease of exposition, land and labor are not explicitly introduced as inputs t o / and the only argument is the amount of credit. However, it is obvious that these can be explicitly introduced and since these are inelastically supplied, this does not change the results of the model. A contract in the credit market specifies an interest rate Rt at which the agent can borrow (with the quantity being determined residually by the tenant). Moneylenders have access to an unlimited amount of credit at the rate r. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 2.1 Collusion in the Credit Market I will first consider the situation where each moneylender just operates in the credit market and takes as given, the outcomes in the share tenancy market. The moneylenders attempt to collude. Let, Xcj and X c 2 be the exogenous market shares of the two moneylenders in the credit market, where Xc 1 +X C2 =1. In this game, there are two players - moneylender 1 and moneylender 2, who compete on the rates of interest, in a manner similar to a standard Bertrand game, where firms compete on prices. When the moneylenders are successful in their attempt to collude, they share the market in the proportions X° 1 and X c2 and offer the same profit maximizing interest rate (which is determined as shown below). In case, they are unsuccessful in their attempt to collude, their market shares are still Xcj and Xc2, but they both offer the competitive interest rate (which is the opportunity cost of funds, r ). In case one of the moneylenders offers a defecting contract (when the moneylenders are attempting to collude) he gets the whole market. Given the opportunity cost for funds, r, the profits of moneylender j=l,2 in period t are, n ; =i(at,Rt)(R ,-r)X ;N where the quantity of the loan, i(at,Rt) is chosen optimally by a peasant and therefore satisfies the first-order condition af(i<) =Rt. Clearly, di/da>0 and di/dR<0 .Since I restrict attention to stationary equilibria I will drop time subscripts from now on. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 Given a, the best collusive agreement for the moneylenders involves charging the monopoly interest rate Rm which solves the maximization problem, R m = argmaxK{z'(G:,i?)|7?-r]} To consider the conditions under which collusion is feasible, I restrict attention to stationary pure strategy subgame perfect equilibria of the discounted infinitely repeated game between the moneylenders. Following Abreu (1988), I also restrict attention to the best equilibria for the moneylenders within this class. To do so, I examine what outcomes can be supported using optimal trigger strategy punishments, which here involves Bertrand reversion (Rt=r for all subsequent dates) and implies that both moneylenders make zero profits along the punishment path. Let a* be the share on the tenancy market that the moneylenders take as given. At the interest rate Rm , collusion is possible for moneylender j if, -ON (i) This inequality says that collusion is sustainable, if the profits from a market share of Xc j are sufficiently large that it is not worthwhile to offer a slightly lower interest rate and grab the whole market for one period and then get zero profits in the future. This simplifies to, < 5 , > l - X C j. Thus collusion at Rm is feasible if both moneylenders have discount factors sufficiently high compared to their market shares. The larger a particular agent’s market share, the lower are the deviation Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 profits relative to the collusive profits and the less the temptation to cheat. Thus the higher is X °h the greater the range of < 5 , for which collusion can be sustained. I define critical values for the discount rates, 8 x = 1 - A \ and 8 2 = 1 - X2 To make the analysis interesting, I now make the following assumption on the parameters. Assumption \ : 8 X> 8 X and S 2 < S 2 Assumption 1 says that while moneylender 1 has no incentive to cheat, moneylender 2 does. The assumption therefore implies that collusion is infeasible in the credit market. 2.2 Collusion in the Tenancy M arket Consider the two moneylenders (now also landlords) attempting to collude in the market for share tenancy, ignoring the fact that they also operate in the credit market. Let Xs j be the exogenous market share of landlord j=l,2 in the share tenancy market and let Xs i + XS 2 = \ 1 ■ I assume that both landlords h av e# plots of land and can hence hire all the peasants. The structure of the game is similar to the one on the credit market. There are two landlords who compete on the shares offered to the peasants, in a manner similar to firms competing on prices in a standard Bertrand game. When the landlords are 7 It is not necessary for the results that the two moneylenders share the whole of the tenancy market. We could easily allow for the case where (X s 2 + Xs 2) < l,and the only difference is that the expressions are slightly more complex. We can assume that the incentive constraints of any other agents who may be initially in the tenancy market are satisfied and do not pose a threat to collusion. For simplicity, we assume that previous landlords drop out of the market. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 successful in their attempt to collude, they share the market in the proportions Xs j and X s2 and offer the same profit maximizing share (which is determined as shown below). In case, they are unsuccessful in their attempt to collude, their market shares are still Xsi and Xs2, but now they both offer competitive shares (which is 1, corresponding to the peasant getting the whole produce8 and the landlords making zero profit). In case one of the landlords offers a defecting contract (when the landlords are attempting to collude) he gets the whole market. I again focus on the best collusive equilibria for the moneylenders, which again implies Bertrand reversion with ay =7 along the punishment path. Let R* be the rate of interest on the credit market that they take as given. Collusion is possible on the share tenancy market at the rate a* iff, (2) This has a similar interpretation to equation (1) and simplifies to Sj> (1- Xs j ).I now make the following assumption on the parameters. Again, I first irst defi critical values for the discount rates, 8 X = (1 - X \ ) and 8 2 = (1 - X2 ). Assumption 2:8X < 8 and 8 2 > 8 Assumption 2 says that the situation with respect to collusion for the moneylenders in the tenancy market is the reverse of the situation in the credit market. In the credit 8 The assumption that the competitive solution corresponds to peasants getting the whole produce, is not required, and is made for ease of exposition. If there is an alternative to leasing out the land (say self-cultivation) then the competitive outcome involves both landlords setting shares (say a <1) that give them the same profit as what they would get from the alternative. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 market it was moneylender 2 whose incentive constraint was violated. In the tenancy market it is the other way round. Moneylender 2 finds it optimal to collude while moneylender 1 wishes to defect. Intuitively, moneylender l ’s market share is just too low in the tenancy market for collusion to be worthwhile. Assumptions 1 and 2 together imply that Sx e (Sx , Sx) and S2 e (< 5 2, S 2) 2.3 Interlinkage The assumption that the moneylenders treat the two markets as separate is of course unrealistic. If they do, Assumptions 1 and 2 together imply that collusion will be feasible in neither market. In this case competition generates the outcomes. Proposition 1: If Assumptions 1 and 2 hold, and if the two moneylenders offer simple contracts, then collusion is not feasible and the unique subgame perfect equilibrium involves the peasants getting all the rents with R=r and a=J. I now show that interlinkage may change the results of Proposition 1. The difference between interlinked and simple contracts is that in the case of an interlinked contract, a bundle of credit and share tenancy contracts is offered. This in turn implies that when an agent defects from a collusive agreement, he offers a better bundle of credit and tenancy contracts. The effect of this is that deviation profits involve summing the profits from deviation in the credit and tenancy markets. If we examine collusion with interlinked contracts, we have to simultaneously optimize over a and R. Call the solution to this optimization problem (which we assume to be unique) a* and R*. These values solve the problem, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 max{i(a, R)[R - r] + (1 - a)f(i(a, R))} {a,R} subject to a participation constraint for the peasants. For simplicity, we normalize the reservation utility of peasants to zero, so that this constraint will always be slack. a* and R* are implicitly defined by the first-order conditions, - ^ - [ * - r + ( l - a ) / ' ( i ) ] - / ( i ) = 0 da and, oR Collusion is possible with the interlinked contract if, (1 - a ) f ( i ( a \ R ') ) X ; N + i( a \R ') [ R '- L]X;N 1-S, (1 - a ) f ( i { a , R ') ) N + i(a',R')[R'-r]N (3) or if the following equations simultaneously hold, s )i(a ’R > [R -d + O -V X i-g )/(/(« )) (5) Ka ,R*)[R* - r] + (1 - a*)/(/(«*, R *)) Define critical levels of these discount rates S j such that these inequalities hold as equalities. Note that S j < 1 .We now have the following lemma (recalling that (5 , > 8 X while d2 < 8 X ), Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 Lemma 1: 8 1 < 8 l and 8 2 < 8 2 Proof: To see that < 5 “ 1 < 8 l , use the definition of 8 1 from (4), multiply through both sides by - r ] + (l-a*)/(z'(a*,i?*))and recall the definition of£, to cancel the terms in (1 - Xx 1 )(1 - a *)/(/' (a * , R *)) . 8 1 < < 5,, then implies that, ( \ - X ‘)i(a \R ')[R '-r]< {\-lxs)i{a ,/?*)[**-/;] which follows from Assumptions 1 and 2.The result 8 2 < 8 2 is established in a similar way from (5) and Assumptions 1 and 2. Q.E.D The main result of this section is that, despite the fact that under Assumptions 1 and 2, collusion would not be feasible if the two moneylenders attempted to collude in either market separately, it is feasible when they interlink. This is because interlinkage has the effect of pooling the incentive constraints (1) and (2). Proposition 2: With interlinked contracts even if Assumptions 1 and 2 hold, collusion is feasible in both markets if8 ; > 8 j. Moreover, if8 ] e (8 ,,8 ,) and A 8 2 e (S2, S2) then collusion is feasible with interlinked contracts when it would not have been so with simple contracts. 3. The Model with Frictions The model of the previous section showed that interlinkage can facilitate collusion when both moneylenders can interlink. In this section I show that even if only one moneylender can interlink, this can still facilitate collusion in the case where he can interlink into a market with frictions. I will also show that collusion Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 could occur even if one of the moneylenders offers both interlinked and simple (non- interlinked) contracts. I introduce frictions into the tenancy market in a simple way by assuming that in any period not all agents who wish to transact can do so. Thus in any period, there are peasants who would like to become share tenants, but cannot find a landlord with vacant land, and simultaneously there are landlords who would like to hire a tenant but cannot find one. I model this with a standard matching technology (see for example, Pissarides (2000 a)) and let m denote the total number of matches in any period, m is a function of the number of peasants searching, and the number of vacant plots of land that are available. Since I wish to focus on steady-states, I assume that in each period, previous matches between tenants and landowners break up with probability s. I also assume that a landlord who offers interlinked contracts can withdraw both land and credit contracts, if a peasant withdraws from the credit contract. I will first consider the situation where both interlinked and simple contracts are offered. I assume that the timing of events is as shown below - (i) Peasants and landlords search for matches. Some of them find matches, and others don’t. (ii) Those peasants who match with the landlord, who can offer interlinked deals, get interlinked deals. Those peasants who find matches with Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 (iii) landlords who cannot offer interlinked deals, obtain credit from one of the moneylenders. (iv) One of the moneylenders offers defecting offers. (v) Those peasants who get defecting offers, and who are holding simple contracts, accept the defecting offers. Those peasants who get defecting offers, and who are holding interlinked contracts, search for new matches while holding the current match. If they find matches, they accept the defecting offers and move to the new landlord. From the above, it is clear that I assume search by two kinds of peasants - by those who don’t have matches, and by those who are matched with interlinked contracts. I assume that only peasants with interlinked contracts can search, and not peasants with simple contracts. This assumption simplifies the model because we need not consider search by peasants who are matched with landlords offering simple contracts. However, this assumption does not change the results of the model because peasants who are holding simple contracts do not face the possibility of losing these, and also do not gain from searching while they are holding these contracts. I also assume that search by a peasant holding an interlinked contract is “on the job.” This assumption makes the model fairly simplified, and also allows in an interesting way for agents who wish to deviate from a collusive agreement to make deviating offers and attract away a proportion of the peasants of the other moneylender. The reason I assume that search is “on the job” is that this implies that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 the model is always in steady-state. If peasants who were matched had to quit to search for another tenancy contract (when they accepted a deviating credit contract), then this would introduce complicated transitory dynamics along the punishment path, which are not important for the points we wish to make. I assume that there are M landlords, who have L plots of land each, where L>N. Hence each landlord can hire all the iV peasants. Out of these M landlords, there is one, viz. moneylender 1, who also operates in the credit market (and can hence offer interlinked contracts). Moneylender 1 has a discount rate of 3] and the remaining landlords have a discount rate of 3. 1 also assume that the market shares of the moneylenders are the same in the credit market even when one of them offers interlinked contracts. There are two kinds of peasants who borrow from moneylender 1- those who are matched with him on the land market and those who are matched with other landlords. I will assume that both these kinds of peasants get the same credit contracts, i.e. moneylender 1 cannot offer different interest rates to the two kinds of peasants who borrow from him. This is an assumption about the nature of the credit market and it simplifies the model considerably. This is because if moneylender 1 offers different interest rates to these two kinds of peasants, then the other moneylender has to make different defecting offers to these two kinds of peasants. The assumption allows us to focus on the main result without considering the above issues. Since the interest rates offered by moneylenderl to all peasants are Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 the same, I also assume that when the moneylender 1 makes the share tenancy offer to the peasants matched with him, he takes the interest rate, R as given. Let V” and V“ be the expected discounted values to a peasant of having a match and being unmatched respectively. These values satisfy the following recursive relationships, V m = qf(i) - Ri + Sp [(1 - s)Vm + sVu ] Vu =w + S J — V m + ( I ) V U] - p L u u w is the reservation utility of the peasant, a is the share, R is the interest rate, and i is the amount that is borrowed. Here m/U is the probability that an unmatched peasant finds a match, where U is the number of peasants looking for matches. As we noted earlier, we model m to be a function of the number of peasants looking for matches and the number of vacant plots. In particular, m(U,P) depends at date t on U, the number of unmatched peasants and P, the number of unmatched plots. If there are U unmatched peasants, the total number of peasants who have tenancy contracts is (N- U), and the number of vacant plots is given by P=ML-(N-U). As is standard, I assume that m has positive marginal products, negative second derivatives, and is homogeneous of degree one. In addition to these values, there is a flow equation which determines the evolution ofUt. UM =Ut -m(Ut,P,) + s[N - U t], In steady-state, this difference equation becomes, m(U,P) = s[N-U] (6) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 The above equation determines the number of peasants who are unemployed, if, the number of peasants who are matched, (N-lf), and the number of vacant plots, P* =ML-(N-U*). The number of vacant plots with the landlords who can offer non- interlinked contracts and with the landlord who can offer interlinked contracts are Let Wsm and Wsu be the value to a landlord offering non-interlinked (simple) contracts of having a matched plot and an unmatched plot respectively. These satisfy the following equations, where -k is the cost associated with having an unfilled vacant plot while looking for a tenant and m/ML is the probability that the plot will be matched (there are L plots for each landlord). Similarly, let W,m and W,u be the values to a landlord, who offers interlinked contracts, of having a matched and an unmatched plot. These satisfy the following equations, The formal definition of the equilibrium is tedious and is not required for our purpose, so I will skip it and sketch how it is determined. It is worthwhile to point ~ P ( M - 1) P P = — -------- and — respectively. M M W f = (1 - a) fit) + £[(1 - s)w ; + sw ; I (7) (8) Wtm = (1 - a)f(i) + S [(1 - s)Wf + sW; ] (9) ( 10) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 out that this model falls into the class of “Wage Posting” models where firms post offers and workers who find matches with these firms either accept these or search again. In the other important class of models in the literature on matching, wage offers are determined by Nash Bargaining (on this issue, see Pissarides (2000 b)).The equilibrium is derived in the following manner. The landlords who offer simple contracts maximize Wsm taking R as given, subject to the constraint that Vm > yu jhg ian(jiord who offers interlinked contracts maximizes W,m taking R as given, subject to the constraint that V m > V u. The rate of interest, R is determined in the game between the two moneylenders, where the moneylenders take the shares on the tenancy market as given. In equilibrium, all the values taken as given are the actual values. Let (a,R*) be the values of the share and the rate of interest for the non-interlinked case. We can similarly derive the equilibrium when only simple contracts are offered. I now consider the effects of these frictions on collusion. I begin by assuming that Assumption 1 holds and again let X}c and X2 C represent the market shares in the credit market. In this case, collusion is not possible in the credit market with simple contracts only. Notice that only matched peasants desire to transact in the credit market. Now consider the effect of moneylender 1 interlinking into the tenancy market. I show that this can allow collusion in the credit market by making peasants reluctant to accept deviating offers by moneylender 2 (who was the impediment to a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 collusive agreement under Assumption 1). Collusion is now feasible in the credit market for moneylender 2 if, K a ' . R ' W - t W i N - i r ) ^ i(a*,R*)[R*-r](A2c(N-U *) + \ P)) (11) M M The above inequality has the following interpretation. The left hand side is the profit from collusion in the credit market given that the total number of peasants wanting loans in steady state is (N-lf) and given the market shares X/ and . The right hand side is the profit from deviation. If moneylender 2, offers an undercutting contract then he keeps his peasants A2c(N-U *), but due to market fictions, only captures a proportion of the peasants of his competitor, moneylender 1. The y L ^ competitor initially has /I,c (N - U*) peasants, out of which — ------------ have M interlinked deals. The peasants who have interlinked deals search on the job and m(—— of them find a new match and accept the defecting contract with M moneylender 2. The remaining peasants do not find a new match and therefore stay with moneylender 1. We can re-write the inequality in a useful way by defining Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 X 2 = (1 - — + m ( ( N ' U * ) /M ’ p ) - ) We have t0 note that X <1 This is 2 M (N-U ) m((N-U* )/M,P) because— ------------ — < 1, since this is the proportion of peasants, who get (N-U ) /M defecting offers and who find matches. Then (11) implies, (12) 2 X 2 From (12), we can define a critical value for the discount factor, S2 .It is then immediate that the following holds, Lemma 2 S2 <S2 X c X ° Proof: Since JG<7, - X - > X i ° .This implies that(l — —) < (1 - X2 ). The result X 2 X 2 follows from the definitions of S2 and S2. Proposition 3: With interlinked contracts even if Assumption 1 holds, collusion is feasible in the credit market if S2 > S2 .Moreover, if S2 e (S2, S2) then collusion is feasible with interlinked contracts when it would not have been so with simple contracts. The intuition for this result is straightforward. By interlinking into the tenancy market, moneylender 1 makes it more costly for his clients to accept deviating deals from moneylender 2. This is because in order to accept a better credit deal, a peasant has to search to find a new tenancy contract. Since the tenancy market Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 has frictions, only a proportion of potentially deviating peasants, who search, find new matches and quit. The effect of this is to reduce the expected deviation profits for moneylender 2 and therefore to increase the size of the part of the parameter space over which both moneylenders can collude. 4. Conclusions This chapter has provided a new rationale for the existence of interlinking. I have shown that interlinked deals could be offered because they facilitate collusion. In particular, I have isolated two mechanisms that function in this way. Firstly, interlinkage has the effect of pooling incentive constraints across markets. This can allow collusion with interlinked contracts when collusion with simple contracts would not have been feasible. Second, interlinkage affects the incentives of peasants to accept deviating contracts and thus reduces the deviation profits of agents considering deviating. While the models that I have considered deal with the interlinking of sharecropping and credit contracts, we can easily extend the analysis to other kinds of interlinking - product-credit, labor-credit etc. Also, while the models above deal with a situation where tenancy contracts are sharecropping contracts, similar results can be obtained if the contracts are fixed rent contracts. By modeling collusion in the agrarian economy, I have thrown light on an aspect of the agrarian economy that has been explored very little in the development literature. By focusing on the strategic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 aspects of interlinking, I have been able to derive significant insights into an aspect of the theory of interlinking that has been relatively neglected in the literature. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 CHAPTER II CONTRACTUAL CONDITIONS OF ATTACHED LABOR: RESULTS FROM AN INDIAN SURVEY In this chapter, I describe results and analysis from field-work on attached labor conducted in the Telangana region of Andhra Pradesh state in South India. This chapter is organized in the following manner. The first section gives a brief description of the villages where the field-work was conducted and the second section discusses the literature on attached labor and the gaps addressed by my work. The third section presents the institutional framework of attached labor in the villages, the data collected, and results from an econometric analysis. The fourth section concludes. Before I proceed further, I will make a terminological clarification. In the theoretical literature, the terms “tied labor,” “attached labor,” and “permanent labor” are used interchangeably to refer to a situation where a laborer works on a long-term basis for a particular employer. In the empirical literature, especially on India, the term “attached labor” is more common although the above terms are also used. These terms are used in contrast to “Casual Labor” which refers to a situation where a laborer works on a short-term, usually daily basis for an employer, and does not do so on a regular basis for one particular employer. In this study, I will use the term Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 “attached labor,” but I will refer to the terms used in previous literature. I will also clarify in greater detail the precise nature of the institution under study. 1. Description of the field-work region Field-work was conducted by me in the period 2000-01 in three villages - Gangadevipalli (G), Machapur (M), and Kothamolgara (K). While G and M are in Warangal district, K is in Mahbubnagar district. All these districts are in the state of Andhra Pradesh in South India. The state of Andhra Pradesh is one of the 29 states in India. It came into existence in 1951 and consists of 23 districts, which are spread across three regions - Telangana, Coastal Andhra, and Rayalaseema. The location of Andhra Pradesh and the Telangana region are shown in Figures 1 and 2. As we can see from Figure 2, both the districts where field-work was conducted are in the Telangana region. The Coastal Andhra region comprises of the following nine districts - Srikakulam, Vizianagaram, Vishakapatnam, East Godavari, West Godavari, Krishna, Guntur, Prakasam and Nellore. The Rayalaseema region comprises of the districts of Kurnool, Cuddapah, Anantapur and Chittoor. The language spoken by majority of the people in the state of Andhra Pradesh is Telugu, which is also my native language. While the language spoken in all the above regions is Telugu, there are slight differences in the dialect that is spoken. Also, these regions have very different historical backgrounds. The Telangana region was part of the State of Hyderabad before Indian independence (in 1947), while Coastal Andhra and Rayalaseema were directly under Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 British rule. Under British rule, India consisted o f - (a) The British administered provinces of India, called British India (b) “princely India,” or those states governed by princes and kings, which were under indirect rule. At the time of Indian independence, there were 565 such states and Hyderabad was the largest. Figure 1: Location of Andhra Pradesh State in India Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 2: The Telangana Region in Andhra Pradesh Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 The Hyderabad State had an area of 82,000 sq. miles and a population of 18.6 million. It consisted of three linguistic regions - Telangana (9 districts), Marathwada (5 districts) and Karnataka (3 districts).1 The languages spoken in these regions are Telugu, Marathi, and Kannada respectively.2 The Telangana region has witnessed several agrarian movements and uprisings both prior to Indian independence and after it. During 1946-51, it witnessed a large-scale peasant mobilization, referred to as “The Telangana Armed Struggle,” which is one of the largest mobilizations of its kind in the world.3 Since 1968, it has witnessed the “Naxalite” movement, which is a Maoist movement inspired by peasant mobilization which started in the Naxalbari village in West Bengal in 1968.4 The Telangana region is underdeveloped, dry and has seen enormous amount of distress in recent years, especially in 1997-98. All the above factors make it an interesting region to study agrarian institutions.5 The villages where field-work was conducted are representative of villages in their respective districts in terms of the crops grown, agricultural practices, 1 These figures are from K.Lalitha et. al. (1989). 2 For a brief history of the region that formed a part of the Hyderabad State, see Qureshi (1937). 3 Moore (see Moore (1966)) points out that this is the second largest peasant mobilization in Asia. Good references on the Telangana Armed Struggle are Gour (1973), Sundarayya (1972). For an interesting account of the struggle from the perspective of women who participated in it, see K. Lalithaet. al.(1989). 4 Good references on this are Dasgupta (1974). 5 For an analysis of the various aspects of this region, see Motiram (1999), Rao.S (1999), and Vakulabharanam (1999). For a detailed analysis of development of this region, including macroeconomic performance, see Vakulabharanam (2003). This study also focuses on some aspects of the agrarian economy of the same villages. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 irrigation, social structure, and livelihood of the majority of the people. Gangadevipalli and Machapur villages are in Geesukonda mandal6 of Warangal district and Kothamolgara village is present in Boothpur mandal of Mahbubnagar district. Some important data about the districts is presented in Table 1. Table 2 presents some important summary data about these villages. As we can observe Mahbubnagar is much less irrigated and receives much less rainfall than Warangal. In fact, Mahbubnagar is a drought prone district and receives much less rainfall than the normal average rainfall in the state (which is 925 mm). We can also note that on many indicators, Mahbubnagar is much less developed than Warangal. From the crops grown, we can see that the proportion of commercial crops (cotton, maize, chillies, and turmeric) is higher in Warangal. In Warangal there is very little agriculture for household consumption as compared to Mahbubnagar. The choice of field work in these two districts is dictated by the fact that this will give us an opportunity for comparing attached laborers in two contexts from the same region, one being more developed, commercialized, and irrigated than the other. As many eminent social scientists have pointed out, caste still plays a very important role in the Indian society in general and in the agrarian economy in particular. Excellent references on the Indian caste system are Srinivas (1962) and Gupta (1991). For our purpose, it is sufficient and important to point out that there 6 The administrative hierarchy is state, district, mandal, and gram panchayat (which is at the village level). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 Table 1: Details of Warangal and Mahbubnagar Districts7 Unit Warangal Mahbubnagar Demographics (1991 Census) Area Sq.Km. 12,846 18,432 Population ‘000 2818.83 3077.05 Urban Population ‘000 546.62 342.19 Rural Population ‘000 2272.21 2734.86 Urbanization % 19.39 11.12 Agriculture Workers as a % of total population (1991) % 47.78 50.20 Workers in Agriculture and Allied Activities as % 75.55 83.40 % of main workers (1991) Per-Capita Food grain production (1997-98) Kg- 117 110 Gross irrigated area as % of gross cropped area % 52.02 23.82 (1997-98) Normal Rainfall per year (1999-2000) mm 1049 754 Per capita bank credit to agriculture (94-95) Rs. 425 423.23 Rice (1997-98) % Area 17.49 13.2 Jowar (1997-98) % Area 5.83 22.67 Cotton (1997-98) % Area 22.65 8.32 Turmeric (1997-98) % Area 1.35 Negligible Maize (1997-98) % Area 10.31 3.01 Arhar(1997-98) % Area 2.69 5.16 Chillies (1997-98) % Area 5.16 1.15 Sesame (1997-98) % Area 4.26 0.14 Some Indicators Literacy (1991) % 39.3 29.58 Male (1991) % 51.68 40.80 Female (1991) % 26.08 18.03 Post offices per 100,000 persons (1996-97) No. 27 27.46 Bank branches per 100,000 persons (94-95) No. 6.67 6.24 7 Source : www.andhrapradesh.com - The official website of the State Government of Andhra Pradesh. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 are people belonging to the following caste groups in the villages - Forward Castes (FC), Other Backward Castes (OBC) and Scheduled Castes (SC). The Scheduled Castes are ex-untouchable castes and people belonging to these castes are also referred to as Harijans or Dalits.8 It is generally agreed that Scheduled Castes are lowest in the ritual hierarchy, as this hierarchy is enforced in a particular region. A convincing argument made by some sociologists (see Gupta (1991)) is that different caste groups have their own hierarchies, but a particular hierarchy might be enforced in a particular context. There are also people belonging to the Scheduled Tribes (ST) who can be loosely described, for want of a better term, as people belonging to indigenous groups. In Gangadevipalli and Kothamolgara, these are migrants from neighboring villages. In the section on data analysis and results, where I deal with the specific issues investigated, I will give some more details about the villages. 2. Literature Survey and Gaps Addressed 2.1 Literature Survey of Empirical Work on Attached Labor It is very difficult and unnecessary for our purpose to summarize all the empirical literature on tied-labor, given the fact that this institution has been subject to analysis by social scientists belonging to various disciplines, in many geographical and historical contexts. Instead, since the field-work region is in India, we will focus on the literature that exists on various parts of India. The purpose of the survey is three 8 The term “Harijan” was coined by Mahatma Gandhi during the Indian independence movement and it means “children of God.” Dalit is a more recent term and it means “oppressed.” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 Table 2: Details of the Field Work Villages Gangadevipalli Machapur Kothamolgara Type of Household9 Large Farmer 7 1 32 Medium Farmer 30 23 68 Small Farmer 56 138 168 Marginal Farmer 66 197 279 Landless 82 194 6 Crops grown Cotton, Maize, Paddy, Turmeric, Chillies, Vegetables Cotton, Maize, Paddy, Turmeric, Chillies, Vegetables Castor, Millets, Cotton, Jowar, Cotton Seed Castes present Mostly OBC OBC and SC FC, OBC and SC Sources of Irrigation1 0 Tube Well, Tube Well, Bore Well, Rainfall and Tank Rainfall and Tank Rainfall fold. It provides a description of this institution in various parts of India, and helps us to get a contrast between agrarian relations in various parts of the country. It will help us to identify certain aspects of the institution that are relatively less subject to analysis and as a consequence less well understood. Most importantly, the survey will help us to contrast the results that we have obtained with the results obtained by other people from various parts of the country and thereby help us to identify our specific interventions into the empirical literature. In the literature in economics, the most well known village surveys are of course the ones conducted by Bardhan and Rudra in 1975-76 and in 1979. In the first 9 The Agricultural Census of Andhra Pradesh defines Marginal, Small. Medium, Large households as those with landholdings in the ranges 0-2.5 acres, 2.5 acres-5 acres, 5 acres-10 acres, greater than 10 acres respectively. 1 0 In Mahbubnagar district, most of the wells have dried up and there is very little rainfall. In Gangadevipalli and Machapur villages, there is a Tank but it provides water only to a few fields. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 one, based on a random sample of more than 300 villages in east India (in the states of West Bengal, Bihar, Orissa and Uttar Pradesh) they found that debt-bondagen is very uncommon (2 % in West Bengal, 14 % in Bihar, and 11 % in Uttar Pradesh). However, voluntary labor-tying arrangements were very common if one includes under these arrangements people involved in annual labor contracts, contracts for a season, for a specified number of days, for a specific operation. In the second survey, based on 110 randomly selected villages in West Bengal, they looked at a spectrum of labor contracts ranging from “totally unattached laborers” to “totally attached laborers,” the intermediate ones being referred to as “semi-attached laborers.” They consider three types of semi-attached laborers - Type 1, Type 2 and Type 3. Type 1 laborers are attached to an employer for part of a year, but for the major part of the year they have the freedom to work for other employers. Type 2 laborers are obliged to work for the employer whenever called, for a stipulated number of days in a stipulated period, and finally Type 3 laborers are obliged to work for the employer whenever called, for an unstipulated number of days over an unstipulated period. Apart from the above three types, they also identify another type of workers - Kirshens1 2 , who are very similar to totally attached workers, but who get paid in terms of a share of the produce. The salient findings of this survey are the following 1 1 Debt-bondage being defined as a situation where bonded laborers are obliged to work exclusively for a single employer for an indefinite period until some loan taken in the past is repaid; 1 2 These are different from sharecroppers because they work under the supervision of the employer, on the employer’s land, using the employer’s means of production. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 - (i) the importance of attached and semi-attached laborers is much more in highly advanced villages as compared to the villages that are not advanced (ii) the basis of attachment is either the allotment of land or current consumption loans, and there is not a single case of labor attachment due to hereditary or outstanding long-term debt (iii) the duration of association between a laborer and his employer can be and is frequently much longer than the duration stipulated in the contract or agreements1 3 (iv) there were only a very few cases of fully-attached laborers working for people who were not their employers (when their masters did not have any work). In the case of semi-attached laborers of types 2 and 3, who had the opportunity to work for other employers, a considerable proportion of them worked for the same employer1 4 (v) taking consumption loans from employers is a common practice. Overall, Bardhan and Rudra conclude that while there is a relationship of dependence of the attached laborers on the employers, and there is unequal bargaining power between the employer and the attached laborers, there is no evidence to suggest that there is any extra-economic coercion. In another survey from eastern India, Sarap (see Sarap (1991)) looked at six villages in the state of Orissa, three from the wet area and the remaining three from the dry area. His major findings are that, (i) there are two forms of attached laborers 1 3 This is because in most cases, the agreement is renewed after the contract period. Bardhan and Rudra interpret this as a consequence of both parties feeling that the arrangement is convenient for them. ^Bardhan and Rudra interpret this to mean that while there is economic dependence of the laborer on the employer, both the employers and the laborers prefer to work with the same opposite party rather than changing parties too often, and this does not imply any extra-economic coercion. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 - guti and kuthiya. While the former refers to agricultural laborers who work in the fields of employers, the latter refers to young herdsmen who take care of the cattle of the employers. There are three forms of remuneration to the guti - (a) remuneration on the traditional pattern which consists of both monthly and annual payment in kind (b) monthly payment in kind (c) monthly payment in cash only. Some landlords also allot plots of land and bullocks to these laborers for cultivation as an incentive. On the contrary, kuthiyas are paid only in kind - paddy, food at the employer’s place, and clothes (ii) the employer provides some amount of cash or paddy at the beginning of employment. This amount is interest free and is given to cement the relationship. The employer charges interest on this loan in case the laborer violates the contract mid-season. In the above cases, the payment is monthly after work for that particular month, so the interest charged on the loan acts as a disincentive for the laborer. This kind of default is very rare also because of the personalized nature of the relationship between employers and laborers and because the laborers are worried about the impact that this has upon their ability to get employment from other people in the village (iii) laborers circulate among the various employers, and this trend has received an impetus in the recent times. Also, the tenure of the contract has shortened. The reasons are two-fold, coming from both the laborers and the employers. The laborers want to shorten the tenure because they want freedom of work and also want to exploit the opportunities prevalent in other jobs like construction, brick making etc. However, the more important reason for the above Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 trend is due to the employers. The employers want to find docile and cheaper laborers who have the same commitment and ability to work. They can find such laborers among migrants who come to this region in the peak season (iv) the labor market in general, is segmented on the lines of caste, tribe, kinship, gender, local/migrant etc. and this reflects a lack of collusion or collective bargaining among the labor households. While not so well known in the literature in economics, one of the most well known debates in recent times, on the nature of agricultural labor in the Indian context, is the one between Jodhka and Brass (see Jodhka (1994), (1995), and (1996), Brass (1990), (1995), (1996a), and (1996b)). This debate is based on the field-work conducted by both these authors in the state of Haryana in North India, and has been carried out in the larger context of the debate on free and unfree labor. It is not necessary for me to intervene into this highly contested terrain because I am concerned with attached labor only, on which there is largely agreement between the two.1 5 Moreover, as pointed out by Rao (see Rao (1999)) this debate is largely methodological and interpretive and the distinction between free and unfree labor is useful only as a tool for historical analysis, and is difficult to sustain on methodological grounds. I agree with Rao that, 1 5 The only differences between the two in this regard are that while Brass finds evidence for increasing unfreedom (or what he refers to as d e p r o l e t a r i a n i s a t i o n ) , Jodhka finds evidence for countertendencies - laborers’ ability to change employers, their search for alternative sources of employer-creditors, and their increasing assertiveness. Also while Brass finds evidence for heritability of debt, Jodhka finds no evidence of this. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 the concepts of free and unfree employment relations, even when they have been consistently defined, do not lie on a continuum because they are based on the incommensurable categories of negative freedom and self- determination. Nor can any clear separation be made between labor subject to “non-economic” coercion versus labor subject exclusively to “economic” coercion.1 6 The important findings of these authors on attached laborers are as follows1 7 (i) they enter into the arrangements voluntarily when they are “compelled” by the need for credit (ii) they work for longer hours than debt-free unattached workers for the same or lower wage rates (iii) their higher rate of exploitation is due to their dependence on the employer for credit, coupled with the restrictions he places on their ability to participate freely on the open labor market. Jodhka also finds the following about attached laborers, (i) They are more indebted than their counterpart casual laborers, they have more difficult working conditions and most of them strongly dislike working as attached laborers (ii) Unless there was a serious compulsion, rarely did any laborer want to work as an attached laborer. The compulsions always arose out of a need for credit, required in most cases, for weddings in the family (iii) The average number of years a laborer works as an attached laborer seems to be coming down. Attached laborers try to move from one employer to another. This is a consequence of growing tensions between farmers and laborers, some of which had 1 6 While Rao’s argument is based on a Marxist perspective, Basu (see Basu (1994), pp. 11-12) makes a similar argument based on a neo-classical/triadic analysis. Basu argues that, Under such circumstances it is no longer clear what it means to say that a person is free to quit. Hence, Thomer’s definitions (about free labor) seem to run into difficulty. But what this demonstrates most clearly is that we are on a terrain that is conceptually hard. 1 7 We adopt this summarization from Rao (1999). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 resulted in physical violence. If the laborer has debts with the current employer, then the way this movement is accomplished is by finding an employer who can lend enough money so that with the cash advance, the current employer’s debt can be paid-off. The laborers also tried to use other sources of credit like petty moneylenders/shopkeepers, and adopted several strategies for minimizing debts to their employers.1 8 Overall the picture of attached-labor that emerges from Jodhka’s study is one of an antagonistic relationship between the laborers and employers, with the laborers using all the means at their disposal to get out of attached labor. Jodhka’s findings contradict Bhalla’s (see Bhalla (1976)) findings from Haryana regarding the caste composition of attached and casual laborers. While Bhalla found that attached laborers and casual laborers came mostly from marginalized landowning households and menial castes respectively, Jodhka found that both attached and casual laborers came from the menial castes. Finally, we will look at two surveys in the ICRISAT villages of Aurepalli and Dokur in Andhra Pradesh. These surveys are important for us because these are from the same region (although the ICRISAT villages have a different socioeconomic structure from the villages in the Warangal district). Also, one of the villages in our analysis is also in the same district (Mahbubnagar) as these ICRISAT villages. The 1 8 The strategies found by Jodhka are - (i) borrowing from the Integrated Rural Development Program (IRDP) (an institutional lending scheme, where laborers can borrow to buy an asset) and using the amount to repay the debt, instead of buying the asset, (ii) depending upon wife’s casual labor income for daily expenses of the family and minimizing borrowings from the employer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 first survey was conducted in the 1970 and 1980s1 9 (for details, see Binswanger et. al. (1984)) and the second survey was conducted in 1991-92 (see Krishniah (1998)). Both these surveys focused on Regular Farm Laborers (or Regular Farm Servants) who were defined by Binswanger et. al. as any laborer whose labor contract exceeds three months and by Krishniah as a person who agrees to work with an employer on regular basis, for a fixed sum of remuneration paid either at the end of the period of work, or in the beginning or on some mutually agreed upon point of time during the period. While the rainfall and soil conditions in Aurepalli and Dokur are similar, and neither of the two have a major source of labor demand (like an industrial center or a government project) nearby, Binswanger et. al. pointed out the following differences between Aurepalli and Dokur - (i) Dokur is more irrigated than Aurepalli and a substantial portion of the irrigated area is double-cropped. As a result agricultural demand is greater and more evenly distributed in Dokur as compared to Aurepalli. (ii) From Dokur many unskilled laborers migrate temporarily to long distances to work on government projects, but there is no such practice in Aurepalli. The main findings of Binswanger et. al. are (i) In Aurepalli, only workers from the two lowest caste groups are RFS s and none of the members of the first two caste groups 1 9 This survey included six villages from Andhra Pradesh, Maharastra and Madhya Pradesh. We will focus on the surveys in Andhra Pradesh. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 participate in the hired labor market2 0 , whereas in Dokur all caste groups participate in the hired labor market and only people belonging to the first two caste groups have not become RFS s. (ii) In Aurepalli, most of the RFS enter this arrangement because they need credit. Most of these loans are needed for ceremonial expenses like weddings, but there are also consumption loans and medical loans. In Dokur, the main reason for the laborers into entering this arrangement is the advance, (iii) In Aurepalli, a regression on the monthly remuneration indicates that the following variables are significant - interest rate for loans, the dummy for cowherd (indicating whether the RFS is a cowherd or an agricultural laborer), the dummy for whether the laborer belongs to two caste groups (Kurmas and Malas) which are low in the hierarchy. The RFS are drawn from three castes - Malas, Madigas and Kurmas (Kurmas are herdsmen, whereas Malas and Madigas are ex-untouchable castes), and the significance of this dummy also represents the fact that laborers belonging to these two castes get higher remuneration than those belonging to the Madiga caste. This is consistent with the prevalent notion among the villagers that Madigas are weak, lazy, and untrustworthy - something not borne by the facts. The fact that personal characteristics like age and experience do not have a significant impact on the remuneration is consistent with wage fixing, which is reflected by the fact that while the wages for irregular labor have fluctuated, the wages for RFS s have remained stable for many years. While a strike in 1977 was successful in pushing the 2 0 Binswanger et. al. identified seven caste groups, which they ranked hierarchically. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 wage, another one in 1978 failed to do so (iv) In Dokur, where the system of temporary migration is present, this system acts as a substitute for the RFS system. There are relatively very few plowmen in Dokur and most RFS are herdsmen. The households that send herdsmen are typically the same that send out temporary migrants. The typical history of an individual in these families is to start working between the ages 10 to 15 and work till the age of 16 to 18, and then to work as a daily farm laborer supplemented by temporary migration to obtain project work during the dry season, (v) Due to the advances provided by the system of temporary migration, unlike in Aurepalli, there is no need for marriage loans from employers (vi) There is monopsony control of the employers in Aurepalli because of the absence of local and distant opportunities and because of strong caste ideology. Krishniah (see Krishniah (1998)) looked at the same villages in the 1990s. Two significant changes that have occurred in the late eighties in Aurepalli are the increase in area under cotton crop and the increase in non-farm occupations (like toddy tapping, sheep rearing). Cotton is a commercialized crop that requires large amounts of labor over a long harvesting period and the regular application of pesticides. The increase in the acreage under cotton crop has led to an increase in demand for RFL and an increase in non-farm alternatives has led to the tightening of the labor market. Both these factors have led to an increase in the remuneration of RFL s in Aurepalli. While in 1980, the payment in cash and kind to RFL s was higher in Dokur as compared to those in Aurepalli, it is the opposite case in 1991-92. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 The other main findings are, (i) A significant proportion of RFL s in Aurepalli are RFL s because they need assured employment and income, whereas in Dokur, this is because they need credit, (ii) While individual characteristics like age, marital status, landholding category, Per Capita Income in the household, are significant in explaining the choice of RFL in Aurepalli, in Dokur household characteristics like Overdue amount and Irrigated Operated Land are significant in explaining the same. The reason for this is that the motivation for RFL as pointed out above is different in the two villages, (iii) The remuneration in Aurepalli significantly depends upon the marital status, per capita operated land, Per Capita Income, landholding capacity, Probability of Unemployment. The remuneration in Dokur significantly depends upon the literacy status, years of association with the employer, irrigated operated land, amount borrowed from the employer. Those factors that increase the bargaining power of the laborer increase the remuneration and vice-versa, (iii) The association between laborers and employers in Aurepalli significantly depends upon the marital status, Per Capita Operated Land and remuneration. In Dokur, the same is true for the Per Capita Income. 2.2 Gaps Addressed My empirical study will complement and provide a contrast with empirical studies from other parts of India. There are three significant contributions that this empirical study will make to the burgeoning empirical literature on agrarian institutions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 (a) Contractual Forms. Terms and the impact of debt and other socioeconomic factors: I will investigate the various forms taken by the contract in the field-work region and the implications of debt, need for credit, caste and other socio-economic factors for the terms of the contract. In particular, I will investigate the factors that influence the bargaining power of the laborers entering into this arrangement. I will also look at the mobility of the laborers, i.e. the ability of the laborers to move from one employer to another and to move away from attached labor, and the factors that influence the actual duration that the laborer spends with an employer. In other words, I will systematically explore the socio-economic factors that impact the “freedom” of the laborer. Based upon this analysis, I will be able to provide several insights into this institution, some of which have been derived from other contexts and some of which are new. I will consider some variables that have not been considered in past studies, e.g. history of default, variables that capture the specific nature of tasks that some laborers do etc. (b) Default on the labor contract: Another issue that has not been systematically explored is the incidence of default on attached labor contracts. In my empirical analysis, I will explore the following aspects of default - (i) the institutional mechanisms to take care of default (ii) the impact of default history on the laborers’ bargaining power and their ability to re Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 enter the labor market (iii) the implications of the possibility of default for collective bargaining. (c) Collective Bargaining and Collusion: It has become increasingly evident in recent times that factors like worker solidarity, class-consciousness, notions of fairness etc. play a very important role in explaining the actual form that various contracts take in the agrarian context. For example, in a study of casual labor in Palanpur, Mukherjee (see Mukherjee (1999)) found that most of the standard theories of wage determination had very little explanatory power in explaining the wage rates. However, when one looks at empirical work on agrarian institutions, one can observe that there are not many studies examining these issues, although there are studies that have pointed out instances of collective bargaining, collusion etc. (e.g. see Binswanger et. al. (1984)). Bardhan and Rudra (1980) is an exception in this regard. The empirical analysis that I undertake, systematically explores the issue of collusion and collective bargaining. Based upon my analysis, I can derive some conclusions which are of analytical importance, and which can provide the basis for modeling collective bargaining. It is worthwhile to point out that there is very little theoretical work on collective bargaining, given the importance of this phenomenon. Moreover, the existing theoretical models focus on casual laborers and not on tied-laborers. Osmani (1990) formalizes collusion among laborers using the framework of an infinitely repeated non-cooperative game with Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 common knowledge. Using this model, he shows how involuntary unemployment can exist in equilibrium. Datt (1996) explores collusion on the side of both the laborers and the employers using the framework of an asymmetric two-party Nash Bargaining game. Using this framework, he shows the existence of involuntary unemployment, and the existence of gender disparities. 3 Data Analysis and Results 3.1 Methodology and Description of the Data Data was collected from attached laborers and their employers by surveying them. The questionnaires used are given in Appendixes I and II. The data collected from attached laborers was on various aspects of the contract - credit, labor, collective bargaining etc. The data collected from employers was on the reason for employing attached labor, credit, collective bargaining etc. I interviewed the laborers and employers after establishing rapport with the people in the village. This process involved living in the village for long periods of time, and having discussions with various people in the village (farmers, casual laborers, representatives of attached laborers, headmen of the villages etc.) on various issues about the activities in the village. These discussions also form a basis for some of the conclusions reached below. I ensured that proper methodological precautions were taken in the process of collecting data. Laborers were not interviewed in front of their employers; it was Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 made clear to the laborers, employers, and other people that their responses would be kept confidential and not revealed to anyone. The surveys and many important discussions were conducted after sufficient rapport was built with the villagers so that there was a reasonable amount of trust that was built between the author and the people in the village.2 1 The field-work conducted was done with the intention to return at a later point to resurvey these villages and also to conduct repeated visits and sustained follow-up in the future. It is well known that such surveys are very useful, but are very rare (on this issue, see Lanjouw and Stern (1998)). The data set on attached laborers includes 62 laborers and 25 employers. A list of important variables and their descriptions is presented in Table 3. Some descriptive statistics about attached laborers are presented in Table 4 and Table 5. In the analysis of the determinants of attached labor, I will also use data on some landless, marginal, small and medium farmers. As I pointed out earlier, the village of Gangadevipalli does not contain any people belonging to the Scheduled Castes, while people belonging to these castes are present in the other villages. It is also interesting to note that while these people live in a separate “colony” cut-off from the rest of the village in Machapur, they live inside the village (although in houses grouped together) in Kothamolgara. 2 1 For some of the methodological problems involved in interviewing laborers, see Brass (1999) pp. 14. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 Table 3: Some Important Variables Variable Description AGE Age in years HHSIZE Size of the Household DEPRATE Dependents as a proportion of the size of the household TOTLAND Total Land owned by the household PERCAPL Per-Capita land owned by the household TOTDBT Total Debt of the household EMPDBT Debt of the household to the employer TOTBOR Total Borrowings of the household EMPBOR Borrowings of the household from the employer REMUN Remuneration SALARY Salary CONTYRS Years the laborer has been working continuously with the employer EXP Experience in years DUR 1 if the laborer has worked for the employer for more than one year SWENY 1 if the laborer has left the employer next year SCORST 1 if the laborer belongs to a Scheduled Caste or Scheduled Tribe AGRLAB 1 if the laborer is primarily an agricultural attached laborer DISTDUM 1 if the district is Mahbubnagar MIG 1 if the laborer is a migrant from another village PASTDEF 1 if the laborer has defaulted in the past BULLOCKS 1 if the laborer owns Bullocks SEED 1 if the laborer works on a cotton seed farm Table 4: Descriptive Statistics for Some Variables Variable Mean Std. Dev. Min Max AGE 32.09677 15.7498 10 75 HHSIZE 4.66129 1.837171 2 12 DEPRATE .2974334 .210774 0 .714285 TOTLAND .9699597 1.830946 0 10 PERCAPL .1938332 .3775028 0 2.5 TOTDBT 24572.58 16061.93 0 62500 EMPDBT 4487.581 7064.8 0 40000 TOTBOR 7129.032 8443.61 0 30000 EMPBOR 2709.677 4778.322 0 30000 REMUN 10030.81 3810.89 2050 15360 SALARY 9540.323 3988.363 1000 15000 CONTYRS 1.354839 1.765765 0 9 EXP 12.74194 12.61269 0 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 Table 5: Statistics for Dummy Variables Variable Number of cases where the variable takes value 1 DUR 32 SWENY 20 SCORST 22 AGRLAB 40 DISTDUM 17 MIG 12 PASTDEF 6z l BULLOCKS 3 SEED 2 Figure 3: A Schematic Representation of Gangadevipalli and Machapur SC W G M N W - Warangal, G - Gangadevipalli, M - Machapur, N - Narsampet, SC - SC Colony 2 2 There was one laborer in the sample in Warangal, who defaulted the year 2000. One laborer in Mahbubnagar was working for an employer because the previous laborer had defaulted. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 A schematic description of the location of the Gangadevipalli village and the SC colony in Machapur is shown above in Figure 3. Both Gangadevipalli and Machapur villages lie on a road that connects Warangal town to Narsampet, another smaller town. Some laborers belonging to this SC colony work for employers in Gangadevipalli village both as attached laborers and casual laborers. The sample of laborers from Warangal district consists of people from Gangadevipalli village, people from the SC colony in Machapur. There are a few migrant attached laborers in Gangadevipalli and Kothamolgara villages. These laborers are originally from a different village, but live in and work for employers in these villages. There are also some shepherds who live in Gangadevipalli village, but who work for employers in Geesukonda, a neighboring village. Most of the attached laborers in Warangal district are landless laborers, marginal farmers. There are only two attached laborers who are from small farmer households and two-thirds are from landless households. In Mahbubnagar, attached laborers are largely from landless, marginal and small farmer households. There is one laborer each from a medium farmer and a large farmer household, but in both these cases they had absolutely no irrigated land. My analysis and survey is of laborers involved in annual attached labor contracts. In Warangal, attached labor contracts of a shorter duration are very rare, while in Mahbubnagar there are some laborers who are involved in such contracts. However, in both these districts, there are laborers who do particular tasks for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 employers and work regularly for a particular employer. I do not look at these arrangements. In other words, I look at a subset of Regular Farm Laborers, by focusing on laborers involved in annual contracts only. The reason for this is that by doing so, presumably, we can focus on the lowest stratum of the agrarian economy. 3.2 Details of the Institution I will perform a detailed econometric analysis of some aspects of this institution. But, before doing so, I will try to give a description of the institutional framework of attached labor as it exists in the two districts. There are some differences between the conditions of attached laborers in these two districts and also in the way these contracts are structured and negotiated. However, there are several common features. I will first describe the common features here and in the portion that deals with the econometric analysis, I will describe the differences. In both these districts, these laborers are referred to as jeethagallu, which literally translates to as “people who receive a salary,” jeetham being the term for salary. The duration of the contract is one year, starting on the festival Ugadi,2 3 which marks the beginning of the Telugu year. The contract is from one Ugadi to the next one. At the end of a year, the contract is either renewed by the laborer and employer or not. There are two kinds of attached laborers, the first meant to take care of agricultural operations (we will refer to these as agricultural laborers) and the 2 3 The calendar followed is a lunar calendar and this festival occurs on different dates every year according to the Western Gregorian calendar. In the year 2001, Ugadi was on March 25. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 second meant to take care of the cattle - buffaloes, cows or sheep (we will refer to these as shepherds).2 4 In the sample, 40 laborers are primarily agricultural laborers and 22 are primarily shepherds. Sometimes the laborers who are primarily shepherds also do a few agricultural operations for a few days. All the attached laborers are men whose ages are in the range 10-70 years.2 5 While a laborer works with his employer, he cannot work as a casual laborer, or for another employer without the permission of his employer. The employer can however ask the laborer to work for another person and take wages from that person. The laborer’s compensation is composed of a monetary portion and a non-monetary portion. The monetary portion is a cash advance referred to as jeetham. The non-monetary portion could be footwear, blanket, food provided by the employer, clothes etc. Not all employers provide the above non-monetary items, although almost everyone provides footwear. While the salary, which is a cash advance, can itself be interpreted as a form of interest free credit, laborers also borrow from their employers, in which case, there is an explicit interest charged on this borrowing. 3.3 Analysis and Inferences 3.3.1 Credit Relations and the Determinants of the Choice of Attached Labor Table 6 shows the reasons stated by the attached laborers for their choice of the 2 4 Sarap (1991), Binswanger et. al. (1984), and Krishniali (1998) also report two kinds of attached laborers. 2 5 Since the laborers are all men, we will use the pronoun “He” to refer to the laborer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 attached labor contract in Mahboobnagar and Warangal districts. From the table, it is clear that access to credit and the need for a regular and dependable source of income are two reasons for the choice of an attached labor contract. Much of this need for credit arises from the need to pay interest on outstanding debt or a need to repay 26 outstanding debt. Since the salary is a cash advance, it serves this purpose. Table 6: Reasons for Entering Attached Labor Warangal Mahbubnagar (1) Access to Credit 21 1 1 (2) Regular/Dependable Income 7 2 (3) Both (1) and (2) 17 4 Table 7: Details of Borrowing by Attached Laborers Warangal Mahbubnagar Average amount borrowed from employers (Rs.) 2644.444 2882.353 Average amount borrowed from non-employers (Rs.) 3422.222 7058.823 Average amount borrowed from all sources (Rs.) 6066.666 9941.177 Average amount borrowed from employers as a percentage of the total amount borrowed (Rs.) 51.52 % 58.82 % Average (modal) interest rate charged by employers 30 %2 / 18% Average (modal) interest rate charged by non-employers 30% 36% Number of laborers who borrowed some amount as a percentage of the total number of laborers 73.33 % 70.59 % Number of laborers who did not borrow from their employers, as a percentage of the total number of laborers who borrowed some amount 27.27 % 41.18% Total Debt (Rs.) 27031.11 18064.71 Debt to employers (Rs.) 4486 4491.765 26 We have to note that in most cases, institutional credit is not available to these laborers. 2 7 About 50 % of the employers charged an interest rate of 2.5 % per month or 30 % per annum. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 Table 7 gives some details about credit. From the table, it is clear that on the average the total amount borrowed and the amount borrowed from the employers is higher in Mahbubnagar as compared to Warangal. This reflects the fact that the salary in Warangal is higher than that in Mahbubnagar. It is also clear from the table that the average interest rate charged in Warangal is higher than that in Mahbubnagar. Another interesting observation that we can make is that on the average laborers in Warangal district borrow much less than their counterparts in Mahbubnagar district. This is because of the fact that the average cash advance in Warangal for both agricultural laborers and shepherds is higher than the same in Mahbubnagar, and this can serve the purpose of credit. Another interesting observation from the above table is that the percentage of laborers who did not borrow from their employers is much higher in Mahbubnagar than in Warangal. From this, we can infer that there is more reliance on non-employers for credit in Mahbubnagar as compared to Warangal. In Warangal, laborers on the average don’t receive credit at a discount from their employers, whereas this seems to be the case in Mahbubnagar. Interest free credit is very uncommon in both the districts, with only one laborer in Warangal and two laborers in Mahbubnagar receiving such credit. Overall, we can say that the attached laborers operate in an environment which is characterized by credit rationing and hence they try to obtain credit from all the sources available to them. To understand the determinants of choice of attached labor, we look at landless or marginal farmer households in Warangal and landless, marginal, small Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 and medium households in Mahbubnagar. We will consider the probability that the family has at least one member working as an attached laborer. We perform the following regressions - (i) for the data from Warangal district, (ii) for the data from Warangal and Mahbubnagar districts. The results from the regressions are as shown in Table 8 and Table 9. Table 8: LOGIT Analysis of the Probability of a Household Having At Least One Attached Laborer in the Age Group 13-75 in Warangal Dependent Variable: ATTLAB otherwise. = 1 if a family has at least one attached laborer and 0 Explanatory Variable Estimated Coefficient Standard Error PERCAPL -1.830 1.451 TOTDBT* 0.0000417 0.0000235 TOTBOR** -0.0000801 0.000035 NONAGR -0.653 1.119 Likelihood Ratio Chi2 (4) = 10.68, Prob > Chi2 = 0.0304 Number of Observations = 57 * Significant at 3 % significance level. * * Significant at 2 % significance level Table 9: LOGIT Analysis of the Probability of a Household Having at least One Attached Laborer in the Age Group 10-75 in Warangal and Mahbubnagar Dependent Variable: ATTLAB otherwise = 1 if a family has at least one attached laborer and 0 Explanatory Variable Estimated Coefficient Standard Error PERCAPL 0.732 0.989 TOTDBT 0.0000173 0.0000187 TOTBOR -0.0000282 0.0000313 NONAGR* -1.915 0.981 BULLOCKS** -2.978 0.859 Likelihood Ratio Chi2 (5) = 25.54, Prob > Chi2 = 0.0001 Number of Observations = 82 * Significant at 5 % significance level. ** Significant at 5 % significance level. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 Before we analyze the results from Warangal district, we have to note that none of families that have an attached laborer have bullocks, whereas 41 % of the other households do. When we observe the results from the data from Warangal district, we can note the following. The value of total debt plays a significant role in the likelihood of a family having an attached laborer. This is quite consistent with the observation that payment of interest on existing debt or repayment of debt is one of the main reasons for many people to become attached laborers. An interesting result concerns the total amount borrowed in the year, which is significant and negative implying that the higher the need for borrowing in the year, the more the likelihood of a household not having any attached laborers. On the face of it, this appears counter-intuitive because the need for credit is one of the main motivations for people entering attached labor. However, this paradox can be resolved if we note that the cash advance itself is used by the laborers as credit and due to this the attached laborers need to borrow less. We have to also note that landed households in general need to borrow for production purposes. The landless, small and marginal households that do not have attached laborers have on the average, more land than the households that have an attached laborer. The significant and positive coefficient on total borrowings reflects this. Now, we will analyze the data from Warangal and Mahbubnagar districts. Unlike in Warangal district, we can note that in Mahbubnagar district, households having attached laborers have bullocks. From the results, it is clear that if a family Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 owns bullocks, it is less likely to have a member who is an attached laborer. This is quite intuitive, and is similar to the finding of Bardhan from the data on West Bengal (see Bardhan (1984)). The coefficient on NONAGR is significant implying that the presence of a non-agricultural source of income would make the household less likely to have an attached laborer. This is quite intuitive because if a household has a non-agricultural source of employment, then it has another source of income, and hence it does not need to have an attached laborer. The sign on the coefficients for total debt and total amount borrowed are the same as before, however the coefficients are not significant. 3.3.2 Tasks Performed, Remuneration and Bargaining Power As we pointed out earlier, there are two kinds of attached laborers - those who are primarily agricultural laborers and those who are primarily shepherds. In both the districts, the laborers have similar tasks. The shepherds take the cattle for grazing in the village commons and also take care of the cattle in case the cattle fall sick. The number of attached laborers performing various agricultural tasks is as shown in Table 10. Some of the shepherds also perform agricultural tasks for a few days. As we can see from the table, all the agricultural laborers are involved in physically demanding tasks like ploughing, applying fertilizer, applying pesticides, watering the fields etc. There are only a few of them who take the crop to the market, and as we can observe there are only three laborers who sell the crop in the market. A considerable number of them do domestic work for the employers, which is not Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 physically demanding. It usually involves filling drinking water for the employers from the tap and collecting milk from the buffaloes. It is interesting to note that in Warangal district, all the laborers except one supervise casual labor, and in Mahbubnagar district half the laborers supervise casual labor. Agricultural attached laborers work for long hours, especially in the peak season. This is more so in Warangal where there is more work because of commercial crops like cotton, chillies etc. In this regard, the field work region is not an exception and this has been found in other parts of India. For example Datt (see Datt (1996) p. 69) points out in his study from Papda village in Madhya Pradesh state in India that, “the laborers pointed out that working as a harwaha (attached laborer) meant doing a 24 hour duty.” Table 10: Number of Agricultural Laborers Performing Various Tasks Task Warangal Mahbubnagar Ploughing 24 14 Applying Fertilizers 28 1 3 Spraying Pesticides 27 12 Sowing 19 8 Harvesting 27 9 Watering the field 28 14 Supervising Casual Labor 24 7 Taking the crop to market 9 5 Selling the crop 3 1 Domestic Work 16 9 As we pointed out earlier, laborers receive a cash advance and non-monetary items - footwear, clothes, blanket, food etc. There is only one laborer who has been given a place to stay. He happened to be a migrant belonging to the same caste as the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 employer. There are also very few laborers (8 %) who are provided food by the employers. To understand the determinants of remuneration, we compute the remuneration in the following manner. We add the monetary value of the non monetary assets to the salary and arrive at a monetary value for the total remuneration. Many of these non-monetary items (e.g. footwear) are given along with the cash advance. However some (e.g. clothes) are given during the year. Although there is a monthly interest rate on loans, we will not discount these non monetary items because these are very small compared to the cash advance. The results from the regressions on remuneration are shown in Tables 11, 12, and 13. Table 11: Linear Regression Analysis of the Determinants of Remuneration of Attached Laborers in Gangadevipalli Dependent Variable: REMUN = Annual Remuneration in Rupees (1) Variable Est. Coef. t-stat (2) Variable Est. Coef. t-stat INTERCEPT** 10094 6.74 INTERCEPT** 10038 6.50 AGE 29.334 1.08 AGE 30.025 1.00 PERCAPL 2860.922 1.31 PERCAPL 2747.514 1.16 DEPRATE 1503.759 0.56 DEPRATE 1157.236 0.41 TOTDBT* -0.045 -1.78 TOTDBT* -0.050 -1.77 AGRLAB** 3011.442 3.65 EMPDBT 0.047 0.54 PASTDEF 518.745 0.44 AGRLAB** 3048.761 3.53 (1)R2= 0.448 (2) R2 = 0.455 Number of Observations = 33 * Represents a variable significant at 10 % significance level ** Represents a variable significant at 5 % significance level Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 Table 12: Linear Regression Analysis of the Determinants of Remuneration of Attached Laborers in Gangadevipalli and Machapur Dependent Variable: REMUN = Annual Remuneration in Rupees (1) Variable Est. Coef. t-stat (2) Variable Est. Coef. t-stat INTERCEPT** 9134.945 7.56 INTERCEPT** 9005.972 7.40 AGE* 39.127 1.77 AGE* 41.284 1.79 PERCAPL** 4096.121 2.00 PERCAPL* 4006.198 1.91 DEPRATE 887.867 0.45 DEPRATE 903.266 0.45 SCORST** -2922.634 -3.68 SCORST** -2855.081 -3.52 TOTDBT -0.025 -1.17 TOTDBT -0.036 -1.53 AGRLAB** 3513.859 4.70 EMPDBT 0.061 1.25 PASTDEF 627.512 0.56 AGRLAB** 3444.569 4.51 (1)R2= 0.528 (2) R2 = 0.550 Number of Observations = 45 * Represents a variable significant at 10 % significance level ** Represents a variable significant at 5 % significance level Table 13: Linear Regression Analysis of the Determinants of Remuneration of Attached Laborers in Warangal and Mahbubnagar Dependent Variable: REMUN = Annual Remuneration in Rupees (1) Variable Est. Coef. t-stat (2) Variable Est. Coef. t-stat INTERCEPT** 9181.579 8.71 INTERCEPT** 9085.342 8.97 AGE** 49.44 2.71 AGE** 50.66 2.85 DEPRATE 935.925 0.65 DEPRATE 968.958 0.70 PERCAPL 424.432 0.48 PERCAPL 414.681 0.49 SCORST** -2343.28 -3.51 SCORST** -2294.733 -3.58 TOTDBT* -0.033 -1.73 TOTDBT* -0.033 -1.71 AGRLAB** 3387.481 5.40 EMPDBT 0.045 1.09 DISTDUM** -6025.858 -8.03 SEED** 3866.683 2.40 AGRLAB** 3108.155 5.02 DISTDUM** -6432.904 -8.67 (1)R2= 0.711 (2) R2 - 0.743 Number of Observations = 62 * Represents a variable significant at 10 % significance level ** Represents a variable significant at 5 % significance level Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 We will first consider the results from Gangadevipalli, and interpret the coefficients that are significant. An interesting result is that the coefficient on total debt is significant and negative. This implies that the higher the total debt, the lower is the bargaining power of the laborer. This also means that the information flows in the village regarding debt are efficient - households which are heavily indebted are known to be heavily indebted. This is also consistent with the observation of the author. This is a result that has not been obtained in other studies like Binswanger et. al.(1984), Krishniah (1998) and Bardhan (1984). The coefficient of AGRLAB is significant and positive reflecting the fact that attached laborers who are primarily agricultural laborers get higher wages than those who are primarily shepherds. A similar result is obtained by Binswanger et. al.(1984). We will now look at variables that are not significant and interpret this. When we observe the coefficient on the debt to the current employer, we notice that it is not significant. This is an interesting result because this would imply that the debt to the current employer does not significantly impact the bargaining power of the laborer vis-a-vis the employer. Another interesting result concerns the coefficient of the variable PASTDEF which is 1 if the laborer has defaulted in the past. The fact that this coefficient is not significant indicates that whether a laborer has defaulted in the past or not does not have any bearing on his remuneration28. The coefficients on the variables PERCAPL and DEPRATE are not significant and this has two interesting 2 8 The punishment to the laborer, in case he defaults, is discussed in detail later. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 implications. The first is that ownership of land does not significantly increase the bargaining power of the laborer. The second implies that the results do not conform to the prediction of the efficiency theory of wages which would argue that the employers would pay higher remuneration to laborers belonging to households with less per-capita landholdings or with higher dependency ratio (on this see Bardhan (1984)). We will now consider the combined data from Gangadevipalli and Machapur villages. The coefficient on the variable SCORST (indicating whether the laborer belongs to the ex-untouchable caste/scheduled tribes or not) is negative and significant. This finding assumes importance in light of the fact that laborers belonging to SC caste or Scheduled Tribes do not differ from other laborers in terms of the tasks that they perform. This is similar to the result obtained by Bardhan (see Bardhan (1984)) and Binswanger et. al. (see Binswanger et. al. (1984)) and indicates the fact that there is discrimination on the basis of caste/tribe. It also indicates the importance of caste ideology for the economy. Based upon his analysis in West Bengal, Bardhan found that people belonging to ex-untouchable castes get lower monthly wages, whereas Binswanger et. al. found that among the Regular Farm Servants in Aurepalli those belonging to two caste groups get higher than the third caste group. The coefficient on AGE is significant and positive indicating that the remuneration for attached laborers increases with age. We have to note that very young or old attached laborers are predominantly shepherds. This is the reason why Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 the variable AGRLAB captures the nonlinearity involved and there is no reason for a nonlinear term in age. The coefficient on PERCAPL is significant and positive indicating that ownership of land increases the bargaining power, since land can act as an outside option. When we look at the combined data from Warangal and Mahbubnagar districts, we can observe that the coefficients of SCORST, TOTDBT, AGRLAB, AGE, SEED and DISTDUM are significant. The interpretation for the significance and signs of SCORST, TOTDBT, AGRLAB, and AGE is similar to those from Warangal district. The positive coefficient on SEED represents the higher salaries paid to attached laborers involved in cotton seed farming, which requires certain special tasks. An interesting result is that the coefficient of DISTDUM, which is the dummy variable for the district and also the dummy for development (and for less rainfall), is significant and negative indicating that the remuneration in the less developed district is lower controlling for other factors. This is quite intuitive. It is worthwhile to point out that in Mahbubnagar district there is less demand for attached laborers. This is also compatible with the fact that there is higher demand for and hence higher remuneration for attached laborers in regions where there is more advanced and commercialized agriculture. 3.3.3 Default on the Attached Labor Contract To understand the impact of default on the attached labor contract, the laborers were asked whether in the past they had defaulted on an attached labor contract either due Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 to illness, or due to a conflict with the employer or due to any other reason. They were also asked whether they contemplate the possibility of default when they enter into the contract and about the possibilities of re-employment in case of default. I did not rely solely on the accounts of the laborers and cross-checked with employers and other people in the village. In Warangal, there are six laborers who have a history of default (including one who defaulted in 2000-01) whereas in Mahbubnagar, there is only one laborer who has a history of default. In Mahbubnagar, the laborers reported that there had also been cases of default in the past. Also, one of the laborers was working for his employer after the original laborer had defaulted. There is an institutional mechanism to take care of default. In case a laborer defaults, he has to return all the salary to the employer. The employer might in some cases give consideration to the period for which the attached laborer worked and take back only the cash advance for the portion that the laborer did not work for. In the case of migrant laborers, some person in the village (usually a relative) provides third party guarantee. In case the laborer defaults, the person who provides third party guarantee ensures that the cash advance is repaid. We have already looked at the impact of default history when we analyzed the determinants of remuneration. We will look at a few cases of default to understand the consequences of default. Case 1: Two years ago (1999), CS worked for ER as an attacas an attached la salary of Rs. 14,000. After working for six months, he fell sick and couldn’t work. ER Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 demanded the whole salary back and CS had to repay it after borrowing the amount from others. Case 2: In the year 2000, B, a migrant laborer, worked for 6 months for EA for a salary of Rs. 13,000. After working for six months, he left. BW, who is a relative of B and who has been working in the village for the last village for 5 years, ensured that the amount was returned. Case 3: In the year 2000, NR worked for PL for a salary of Rs. 10, 000. After working for 3 months, he left because he did not want to continue as attached laborer. He had to repay an amount Rs. 7500, which was the equivalent of 9 months of the salary. In some extreme cases, the laborers were forced to pay interest on the cash advance. Case 4: In 1999, NV defaulted because of ill health and he paid back the cash advance and interest on the salary. Case 5: In 1999, GG worked for BI for a salary of 10, 000. After 6 months he fell ill and couldn’t work He repaid the whole salary and interest on the salary. Some other authors, who have studied agriculture in other parts of India, have pointed out similar arrangements to take care of default. For example, Brass (Brass (1999), pp. 81) points out in his analysis of agriculture in Punjab that, The most important form of permanent worker employed in agriculture during the early twentieth century, however was the siri. Its subjects were full-time labourers, who worked for rich peasants, and were also recruited through the mechanism of cash advances... Such labourers were not Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 permitted to leave a creditor-employer without first repaying the whole cash advance together with 50 % interest. Most of the laborers in both Warangal and Mahbubnagar believed that if a laborer defaults then he can work as an attached laborer with another employer in the same village or in another village. He can also work as a casual laborer with the same employer or any other employer in the same village or in another village. While the above description deals with default on the attached labor contract, there are also situations where laborers do not show up for work for a few days. The institutional mechanism in the villages to deal with this is to deduct an amount of the modal casual wage2 9 (which was Rs. 50 in the years 1999 and 2000) for each day the laborer doesn’t show up. The prevailing practice is that the employer keeps an account for the number of days absent and at the end of the year charges the laborer for the total amount owed. It is worthwhile to point out here that in Warangal district agricultural attached laborers are entitled to 12 holidays per year (one for every month) whereas those in Mahbubnagar are not. 3.3.4 Mobility As we pointed out earlier, an important issue in the debate on agrarian relations in the Indian context is the “freedom” of the laborers - their ability to move from one employer from another, their ability to move away from attached-labor, the amount of time they spend with a particular employeroyer bef(.Uy move. Even if we agree 2 9 The casual wage fluctuates during the agricultural year. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 that the debate on freedom and unfreedom can only be inconclusive because of the incommensurability of these notions, these are interesting empirical issues to investigate and as pointed out by Rao, are of utility in characterizing change. In order to understand this phenomenon better, we perform two sets of regressions. Regressions on the ability to move away from the employer: . 30 • In the last few years, there has been a new arrangement called bayana in the villages in Warangal district. An employer has an arrangement with a laborer promising that the laborer will work for him next year. The employer pays the laborer Rs. 1000 as a part of the guarantee. In case the laborer refuses to work the next year, the laborer pays the employer twice the amount, Rs.2000. In case the employer does not hire the laborer, he forfeits the amount of Rs. 1000. In most cases, the salary and other aspects of the contract are agreed upon at the time of bayana, and in some cases the employer tells the laborer that these will be decided later after looking at the “going” salaries for the next year. This arrangement is absent in Mahbubnagar. This arrangement according to both laborers and employers has arisen in order for the employers to assure themselves of attached laborers for the next year. In the survey, we have data from Warangal district on attached laborers who moved to other employers or moved away from attached labor. In Waraln Warangalict 50% of the laborers continued with the same employer next year, 30% moved to another 3 0 The literal meaning of this term is “guarantee.” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 Table 14: LOGIT Analysis of the Probability that an Attached Laborer Moves from the Current Employer Next Year in Gangadevipalli Dependent Variable: SWENY = 1 if the laborer moves from the current employer next year. (1) Variable Est. Coef. Std. Error (2) Variable Est. Coef. Std. Error TOTDBT 0.0000506 0.0000375 AGE -0.131 0.0821 EMPDBT* -0.0003139 0.0001469_ DEPRATE 15.829 10.945 TOTBOR 0.000176 0.0002151 PERCAPL -7.833 5.895 EMPBOR 0.0001495 0.0003219 MIG** 5.123 2.966 TOTDBT 0.0001786 0.0001128 EMPDBT*** -0.0007771 0.0004636 TOTBOR 0.0008916 0.0006499 EMPBOR -0.0002124 0.0007305 AGRLAB 0.592 1.792 (1) Likelihood' (2) Likelihood Number of Obs * Significant at ** Significant a *** Significant Ratio Chi2 (8) = 14.24 Prob > Chi2= 0.0066 Ratio Chi2 (8) = 27.43 Prob > Chi2= 0.0012 ervations = 31 3 % significance level it 8 % significance level, at 9% significance level. Table 15: LOGIT Analysis of the Probability that an Attached Laborer Moves from the Current Employer Next Year in Gangadevipalli and Machapur Dependent Variable: SWENY = 1 if the laborer moves from the current employer next year. (1) Variable Est. Coef. Std. Error (2) Variable Est. Coef. Std. Error TOTDBT 0.0000463 0.0000329 AGE 0.009755 0.0295 EMPDBT* -0.0003113 0.0001262 DEPRATE 1.3753 2.772 TOTBOR 0.0000684 0.0000768 PERCAPL -4.5498 4.126 EMPBOR 0.0002842 0.0001891 MIG 1.5450 1.43 SCORST 1.4131 1.383 TOTDBT 0.0000596 0.0000388 EMPDBT** -0.000287 0.0001389 TOTBOR 0.0000117 0.0001035 EMPBOR 0.000376 0.0002783 AGRLAB 0.5432 1.8636 (1) Likelihood! (2) Likelihood Number of Obs * Significant at ** Significant Ratio Chi2 (8) = 18.79 Prob > Chi2= 0.00 Ratio Chi2 (8) = 21.26 Prob > Chi2= 0.00 ervations = 40 1 % significance level it 4 % significance level. 1 6 1 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 Table 16: Average Debt to Employers and Mobility of Attached Laborers in Warangal Laborers who stayed with the current employer Rs. 7540 Laborers who moved to another employer Rs. 2422.5 Laborers who moved away from attached labor Rs. 250 employer and 20 % moved away from attached labor. The results from the regressions for villages in Warangal district are summarized in Tables 14 and 15.3 1 It is clear from the results that what matters for the ability to move is the debt to the employer. The reason for this is that when the laborer moves from the employer, he has to pay-off a considerable portion of the debt of the current employer. In case he moves to another employer, he does this by using the salary from the new employer and any money he can borrow. In case he moves away from attached labor, he repays the employer by borrowing. Table 16 presents the average debt to the employers for three kinds of laborers - (i) those who continued with their employers, (ii) those who moved to other employers, and (iii) those who moved away from attached labor. We can note that the average debt to the employers is highest for case (i) and lowest for case (iii). We can also note that in the third case, these debts are quite small. In Gangadevipalli, the coefficient on MIG is significant indicating that the likelihood of moving away from the current employer is higher if the laborer is a migrant. This is fairly intuitive and does not need much explanation. Two other interesting findings relate to caste. Belonging to SC or ST does not have 3 1 Regressions for Warangal district focusing only on tied-laborers who moved to a new employer yielded similar results. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 any bearing on a laborer’s ability to move, which is interesting in light of the earlier finding that people belonging to this caste have lower bargaining power. Regression on the duration of time a laborer spends with an employer: In Warangal, the longest period that any laborer was working continuously for an employer is 9 years; in Mahbubnagar the corresponding figure is 4 years. In Warangal district, 40% of the laborers were working for the employer for the first time, they had moved to their current employers from either another employer or from casual labor. The remaining 60 % continued with the employer from previous years. In Mahbubnagar district, the corresponding figures are 47 % and 53% respectively. In the whole sample, 43% of the laborers were working for their employers for the first time. To understand the determinants of the duration of employment with a particular employer, we will perform a regression on the variable DUR (which takes a value of 0 if this is the first year and 1 if this is not). The results from the regression are shown in Tables 17, 18, and 19. It is clear that the higher the debt to the current employer (EMPDBT), the more likely it is that the laborer will work for longer than a year for the employer. In a sense, the higher the debt the more likely it is that the laborer will work longer for the current employer. We can also note that the sign of the coefficient of EMPBOR is negative and significant. This has an interesting interpretation. We have to note that when DUR=0, this implies that the laborer has moved from another employer or from casual labor to the current employer this year. The negative and significant Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82 coefficient on EMPBOR indicates that the higher the credit from the employer, the more likely it is that the laborer will move to the current employer from another employer or from casual labor. This is quite intuitive because the need for credit is one of the important motivations for entering into this arrangement. Table 17: LOGIT Analysis of the Probability that an Attached Laborer Works for an Employer for More than 1 Year (Continuously) in Gangadevipalli Dependent Variable: DUR = 1 if the laborer has been working for the employer for longer than 1 year. Variable Estimated Coefficient. Standard Error TOTDBT -3.03 *10'6 0.000036 EMPDBT* 0.0005019 0.0002191 TOTBOR 0.0001085 0.0001233 EMPBOR -0.0003727 0.0002404 Likelihood Ratio Chi2 Number of Observation * Represents a variable (4) = 20.70 Prob > Chi2 = 0.0004 s = 33. significant at 2 % significance level. Table 18: LOGIT Analysis of the Probability that an Attached Laborer Works for an Employer for More than 1 Year (Continuously) in Gangadevipalli and Machapur Dependent Variable: DUR = 1 if the laborer has been working for the employer for longer than 1 year. Variable Estimated Coefficient. Standard Error SCORST -0.746 0.755 TOTDBT 0.000185 0.0000235 EMPDBT* 0.0001101 0.0000594 TOTBOR 0.0000356 0.0000616 EMPBOR** -0.0003067 0.0001418 Likelihood Ratio Chi2 Number of Observation * Represents a variable * * Represents a variabl (4) = 12.41 Prob > Chi2 = 0.0296 s = 45 significant at 6 % significance level, e significant at 3 % significance level. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 Table 19: LOGIT Analysis of the Probability that an Attached Laborer Works for an Employer for More than 1 Year (Continuously) in Warangal and Mahbubnagar Dependent Variable: DUR = 1 if the laborer has been working for the employer for longer than 1 year. Variable Estimated Coefficient. Standard Error SCORST -0.754 0.6315 TOTDBT -0.0000106 0.0000206 EMPDBT* 0.0001156 0.0000514 TOTBOR 0.000047 0.0000433 EMPBOR** -0.0003184 0.0001169 DISTDUM -0.3263 0.712 Likelihood Ratio Chi5 1 Number o f Observatior * Significant at 2 % sig ** Significant at 1 % s 6) = 15.87 Prob > Chi2 = 0.0145 is = 62 nificance level, gnificance level. In fact, as we would expect, the average amount borrowed by the laborers who continued with the current employers (DUR=1) is Rs. 1514, which is much lower than the same for laborers who moved to their employers from casual labor or from another employer, which is Rs. 4423. The coefficient on SCORST is not significant, and this has the same interpretation as it had in the regression for the ability to move away from the current employer. 3.3.5 Collective Bargaining and Collusion 3.3.5.1 Findings There are no formal bargaining arrangements in the three villages. However in all the villages, there are informal arrangements. The informal arrangements involve meetings between laborers and employers to decide wages, discussions among Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 laborers regarding the salaries that they are receiving, discussions among employers regarding the salaries that they are paying. There was also an attempt around 1999 Ugadi among the attached laborers in Gangadevipalli village in Warangal district to form a union and increase the salaries which were in the range 12,000-13,000 to Rs. 18,000. They also went on a strike. In this attempt, the laborers in this village were joined by laborers from some other neighboring villages. This attempt failed to form a union and the strike did not last beyond a few days. The laborers did not get the increase that they demanded, but the salaries did increase as a result of this attempt to Rs. 14, 500. In Mahbubnagar, in the year 1998, there was meeting among some attached laborers and employers where they agreed upon a salary of Rs. 5000 for agricultural laborers and an increase of Rs. 500 every year. We will first consider the Warangal district. The findings from Warangal district are, (1) Most attached laborers felt that it was difficult for them to indulge in collective bargaining because of various reasons. The reasons they cited are, (a) They have to be continuously with the employers, at their beck and call. (b) Since attached laborers are indebted, employers can ask them to pay their debts and leave. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 (c) Attached laborers have no status, because they have to depend on their employers for everything. They also can’t afford to jeopardize their long-term relationship with their employers. (d) It is difficult to find a leader among the laborers because they cannot spare their time for this, as they have to work all the time for their employers. (2) Most attached laborers felt that employers collude to set the terms of these contracts (3) Most attached laborers felt that employers prevent collective action on the part of attached laborers. (4) Most attached laborers felt that the unionization attempt did not work because of pressure from employers. The employers demanded that the laborers either accept their terms, or pay-off debts and leave. The employers also wanted laborers to provide guarantees for those who default (especially migrant attached laborers). (5) Most employers said that they discuss among themselves regarding salaries that they should pay to attached laborers, although some of them said that this tendency has come down in the recent years. Almost all employers also said that the laborers discuss among themselves and try to collude among themselves. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 (6) Regarding the strike, many employers (including the key ones) reported that they had a meeting where they decided that they cannot agree to the demands of the attached laborers. They asked the attached laborers to repay their debts and leave. Since attached laborers couldn’t find alternative sources of credit and employment they had to concede. These are the statements of some of the key employers - “When the tied-laborers tried to form a union, we also tried to form a union. We felt that their demands were unreasonable. So, we asked them to pay-off their debts and leave, if they don’t accept our terms.” “We wanted them to provide guarantees in case of default or in case laborers don’t report to work. We decided that we cannot give the salary that they wanted. We told them that if they wanted higher salaries then they should pay-off their debts and leave. We were confident that they would come back, although we were not sure. They couldn’t raise credit and find work. They couldn’t maintain solidarity among themselves.” When we look at the findings from Mahbubnagar district, we can observe the following, (1) The reasons cited by attached laborers for lack of formal collective bargaining (a) Lack of unity on the part of attached laborers. (b) General insecurity on the part of attached laborers and a fear that the employers might mount a reprisal. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 (c) Indebtedness of the attached laborers, which makes them weaker in terms of their ability to bargain. Although there is no formal bargaining arrangement for casual laborers, some attached laborers felt that it is easier for casual laborers to bargain collectively because they have higher demand than attached laborers. (2) 53 % of the attached laborers interviewed felt that employers collude among themselves, while 24 % felt that this is not the case. The remaining said that they did not know. (3) Most of the attached laborers felt that employers did not prevent collective action on the part of the laborers. (4) There have been no strikes in the recent past, but as we pointed out earlier, there was an agreement among some employers and laborers to fix salaries and salary increases. However, this agreement was only for the salary and salary increase and not for the remuneration (since the non-monetary items offered are different). It is quite interesting to note the importance of caste ideology for this arrangement as the employers involved were all members belonging to Reddy caste (a Forward Caste), which is the dominant landholding caste in the village. If we observe the salaries of the agricultural laborers working for Reddy caste employers in our sample, we can note that except one laborer, all the rest get a salary of 6000. One of the Reddy employers remarked that, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 “We try to fix salaries by having meetings, however there are some employers, e.g. S, who do not stick to the agreement and they pay more.” 3.3.5.2 Analysis and Inferences An interesting observation that one can make from the above findings is that both in Warangal and in Mahbubnagar, most of the laborers were able to understand and respond to the question on whether there is collusion among employers. This is in contrast to the study of casual labor by Mukherjee (see Mukherjee (1998)) in Palanpur village, where she found that most laborers could not make sense of this question. Mukherjee concludes that this could probably be because there is no explicit collusion among the employers. We can interpret our findings from Warangal and Mahbubnagar to mean that laborers are aware of the possibility of explicit collusion and that they have seen some evidence of it. In Gangadevipalli, we can note that individual characteristics like AGE, PASTDEF are not significant, and among household characteristics only TOTDBT is significant. This coupled with the statements of the laborers and employers indicates that at least to a certain extent, collusion exists and can explain the prevailing terms of attached labor contracts. There are two analytical issues raised by the unionization attempt in Gangadevipalli, which will be useful in formalizing collllectiining in the agrarian economy, (1) Effectiveness of collective bargaining depends upon indebtedness. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 (2) Effectiveness of collective bargaining depends upon the ability of some laborers to default. Moreover, there is evidence from other parts of India about such unsuccessful instances of collective bargaining. For example, Brass (see Brass (1999), pp. 98) who studied three villages in Haryana state in North India found the following, Local Casual Laborers also engaged in worker organization and mobilization that culminated in the withdrawal of labour (boycotts, strikes, picketing the village workplace against migrants. The response by landholders to this worker resistance and militancy took the multiple form of recruiting migrant labour, the threat to recall outstanding loans or cash advances, and the prevention of access to fodder for cattle. Breman (1993) (see Breman (1993) p. 214) observed from his study in Gandevigam village in Gujarat state in India that, Another recent phenomenon is that the Dublas may go on strike. Some years ago the farm servants of Gandevigam stopped work for a few hours when they were denied the right to pick wood in the gardens and the landlords refused to raise their wages in compensation. This action failed. The Anavils fetched laborers from other villages and, to demonstrate their fairness went to work themselves. The case pointed out by Brass is particularly interesting because it illustrates how indebtedness can hinder the effectiveness of collective bargaining and is similar to our observations in Gangadevipalli village. Our finding from Mahbubnagar points to an interesting situation where collusion could occur over a subset of employers where this subset could be based upon caste. In a situation characterized by credit rationing and low demand for attached labor, this collusive agreement could be enforced. This provides another Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 interesting analytical issue that can be used in formalizing collusion in the agrarian economy. 3.3.6 Some Other Remarks on the Status of Attached Laborers 74 % of the attached laborers felt that if they did not need access to credit, they would not be involved in attached labor. Both in Warangal and in Mahbubnagar, there was a perception among attached agricultural laborers that attached labor is something avoidable due to the hard physical work involved and the fact that it involves being at the employers’ beck and call. One of them remarked that they have no status, and in the words of another, “they are so low that nobody wants to invite them to weddings in the village.” In Mahbubnagar one attached laborer reported that he was physically abused by the employer and another said that his employer asked him to sweep the house, which he did not, because he found it demeaning. The same laborer told that he wanted to default, but he was afraid of physical violence in case he did so. It is also instructive to get the perspective of some people who did not choose to become attached laborers and chose to be casual laborers instead. There are several reasons cited for this, including “subjection” that has to be displayed to the employers, excessive physical labor involved in attached labor etc. A casual laborer CR told the author that, I am an educated man. I cannot take abuse from anyone. If I become an attached ltached laborer m(.Ul have to take orders from employers. I have to take abuse from them. This is the reason why I did not choose to become an attached laborer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 Another laborer, KK said that he was not working as an attached laborer because “attached labor involves work from 4 A.M to 9 P.M. It is physically enervating.” Open conflict and confrontation between attached laborers and employers is very rare. However, relations between them seem to be antagonistic and except in a few cases, also devoid of any element of patronage. In this regard, the situation of the attached laborers is similar to their counterparts in some other parts of the country, e.g. in Haryana, as pointed out by Jodhka and Brass in their above mentioned studies. 4 Conclusions Overall we can conclude that while some of the results that we have obtained in our analysis are similar to the results obtained from other studies, there are several that are different. We have also systematically investigated certain issues that have hitherto been subject to relatively little analysis in the literature on attached labor. In doing so, we have shed light on some processes that have not been systematically investigated in the past. The fact that we have obtained certain results that are different from other studies reflects the specificities of the institution in the context that we have studied. More importantly, our analysis makes a case for more micro studies so that various institutions can be studied in different contexts, which will in turn lead to a better conceptualization of the agrarian economy. It also provides a cautionary note on generalizing based upon the study of a particular context. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 In what follows, I will summarize the findings from the above analysis and also point out certain methodological and analytical issues involved in conceptualizing attached laborers. (1) Need for credit and indebtedness play a very important role in the lives of attached laborers. Their choice of attached labor is due to their need for credit (arising out of past debt or otherwise) and a need for regular and dependable income. Also, overall indebtedness reduces their bargaining power in the market, thereby reducing the remuneration that they receive. (2) Attached laborers try to change employers, they do not work for one particular employer for long periods of time. This is a departure from the past although this trend started a long time ago. This reflects a change of the institution from a patron-client relationship to a contractual one. Apart from moving from one employer to another employer, some attached laborers also move to casual labor. (3) This ability of laborers to move away from their current employers depends crucially on their indebtedness to the current employer. The higher the debt to the current employer, the more unlikely it is for the laborers to move away. The amount of time the laborers spend with employers also crucially depends upon this debt - the higher the debt, the longer the stay with a particular employer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 (4) Among attached laborers, those belonging to certain castes viz., Scheduled Castes and Tribes (SC/ST s) receive less remuneration. This is a reflection of the fact that caste ideology is important for the functioning of the agrarian economy. It also reflects the fact that there is discrimination on the basis of visible and identifiable characteristics in the agrarian economy. This is similar to discrimination present in some industrial economies based upon characteristics like gender, race etc. (5) There are institutional mechanisms for taking care of default on the attached labor contract. An attached laborer who defaults on the contract has to return the whole cash advance or the portion of the cash advance that covers the period for which work was not done. While default on the labor contract is uncommon, it is not uncommon to find attached laborers who do not show up for work for a few days in the year. There is an institutional mechanism to take care of this, which is to deduct an amount equal to the modal casual wage for each day that the laborer does not show up. History of default does not affect either the bargaining power of the laborers or their ability to re-enter the market. After default, laborers can work as casual laborers or as attached laborers for other employers. (6) While there are no formal bargaining arrangements among attached laborers, there are informal bargaining arrangements. The ability to bargain collectively and to form formal bargaining arrangements is hindered by Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 various factors, one of the most important being the indebtedness of these laborers in general and to their prospective employers in particular. Informal attempts at fixing salaries and other terms of the contract can be effective over a subset of the employers and this subset could be caste based (7) When one tries to understand the factors that affect the status/bargaining power/freedom of attached laborers in the village economy, one can make certain interesting observations. While total debt affects remuneration, it does not affect either the ability of people to move away from an employer or the duration of their employment, both of which are affected by the debt to the employer. While belonging to the Scheduled Caste/Tribe affects the remuneration, it does not affect either the mobility or the duration of employment. Both total debt and debt to the employer can affect the ability of attached laborers to bargain collectively. So, the power/freedom enjoyed by the attached laborers in the agrarian economy is a complex function of the above factors, which do not work in the same direction. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 CHAPTER III A MODEL OF TIED-LABOR, EMPLOYER RELATIONS 1. Introduction One aspect of tied-labor that has been emphasized by several people who have studied this institution in many contexts in recent times is that tied-laborers move from one employer to another. A closely linked phenomenon that has also been pointed out in the above literature is that many laborers, both tied and non-tied, dislike the physical effort and dependence involved in tied-labor, and hence attempt to avoid tied-labor. As a consequence of both of the above, the duration that a tied- laborer works for a particular employer and the length of the tied-labor contract have come down in recent times. Consider the following description by Sarap (see Sarap (1991)) based upon his findings from the state of Orissa in Eastern India, Another change that has occurred (more so in wet areas) is that of circulation of farm servants among different employers and the shortening of the tenure of the contract. A similar description of mobility of tied-laborers among employers is reported by Jodhka (see Jodhka (1994)) in his study in the Indian state of Haryana. Shortening of tenure is reported by Jodhka in his above-mentioned study and Shrivastava (see Shrivastava (1997)) in his study of Chaukra village in the North Indian state of Uttar Pradesh. As pointed out in the previous chapter, in my field-work in the state of Andhra Pradesh in South India, I discovered that tied-laborers move from one Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 employer to another and also do not work for long periods continuously for one particular employer. The impetus for this movement and shortening of duration seems to come both from laborers and employers, depending upon the socio-economic context. On the employers’ side, the motivation for this is that they can get access to cheaper and/or more docile laborers. In the study by Sarap, the availability of migrant laborers provides the opportunity for the employer to do so. Breman (see Breman (1985) pp. 403) in his study of south Gujarat in India points out a similar phenomenon - The local employers, especially the big farmers get angry by the changed behavior of the local landless labor, who no longer conform to the stereotype of obedient and anxious to please farm servants of earlier time. In such a situation the employer may prefer the migrant workers who are docile, ready to work with lower wage, sometimes at irregular working hours and are easy to employ at short notice. Jodhka also points out that, to a certain extent the impetus comes from the employers. On the laborers’ side, this could be due to conflict with a particular employer, due to the offer of a higher remuneration, or because laborers see this movement as a reflection of their freedom. This is pointed out by Breman, Jodhka, and Sarap in their studies mentioned above. I also found from my fieldwork that tied- laborers v iew this m ovem ent as a sign o f freedom and they m ove from one em ployer to another in case they get a higher remuneration by doing so. Apart from movement from one employer to another, laborers also attempt to leave tied-labor for the reasons mentioned above - dislike for the effort and lack of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 freedom involved in tied-labor. In my fieldwork in Warangal, 20 % of tied-laborers interviewed moved away from tied-labor. 74 % of the laborers I studied in Warangal and Mahbubnagar districts, reported that if they did not have a need for credit (arising out of past debt or for other reasons), they would not have been tied- laborers. Some casual laborers I interviewed also cited excessive physical work and dependence on the employer as reasons for not opting for tied-labor. In his study in Papda village in Madhya Pradesh state in Central India, Datt (see Datt (1995) pp. 69) points out that, While the village had some attached laborers (locally referred to as harwahas), the laborers indicated a clear preference for casual labor contracts. We were told that no one becomes a harwaha by choice. All the attached laborers in the village became attached for want of a loan, usually to finance a substantial expense such as that for a marriage in the family.. .It was quite evident that being a harwaha indicated an inferior status and vulnerability. The laborers pointed out that working as a harwaha meant doing a ‘24-hour duty’ and that the harwahas were treated badly by their employers. Several other authors (e.g. see Brass (1999), Jodhka (1994) etc.) point out to similar situations. There is enough evidence to suggest that this ability of tied-laborers to move among employers or to leave tied-labor crucially depends upon debts to their employers and the availability of sources of borrowing. In the case of movement from one employer to another, usually the cash advance and any credit from the new employer are used to pay-off the debt of the current employer. This has been found in many studies, e.g. Jodhka finds from his above study that, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 Out of the 28 respondents who felt that they could easily stop working with the present employer, in 18 cases, they reported that they could borrow the amount from some other farmer and shift to him. Jodhka also points out that tied-laborers follow several strategies of reducing their borrowing from their employers. Similar findings are reported by Binswanger et. al.(see Binswanger et. al (1984)) in their study of Regular Farm Servants (RFS) in the ICRISAT village of Aurepalli in India, by Brass (see Brass (1999)) and other authors. In my fieldwork region, I found that the level of debt to the current employer is significant in explaining the probability that the laborer moves to a new employer and/or leaves tied-labor. This is because the laborer has to pay-off a significant portion of his debt to the current employer when he leaves, and he does so by using the cash advance and any credit that he can obtain from his new employer. I also found from my analysis that the higher the indebtedness to the present employer, the longer is the duration a tied-laborer works for a particular employer. In case the laborer moves away from tied-labor, he repays the debt to the current employer by borrowing from any other source at his disposal, e.g. relatives, local moneylenders etc. Apart from the availability of credit markets, the above phenomenon is also intimately related to strategic interaction among employers and strategic interaction between employers and laborers. Interaction among employers is manifested in the form of a refusal to accept laborers from other employers (which is a form of collusion) or competition among employers for laborers. Collusion could also take Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 the form of fixing wages, so that there is less incentive for the tied-laborer to move. There is evidence to suggest that both of these could happen, depending on the prevailing context. For example, in his study of rural South Africa in 1930 s, Lacey (see Lacey (1981) p. 163) points out that full-time servants, had limited options even within their own districts. Not only would a farmer be reluctant to poach his neighbor’s servants, but the farmers together tacitly agreed about keeping wages down at the same level. Successful collusion by employers is also pointed out in Binswanger et. al.(1984). As I have argued in Chapter 1, the interlinking of various markets, widely prevalent in agrarian economies of developing countries can facilitate collusion under certain conditions. This can explain to certain extent collusion among employers, who are both employers and creditors. While the above examples point out to collusive behavior, there is evidence for competition too. For example, Breman (Breman (1993) pp. 119) points out from his study in Gujarat that, The relationship between various patrons, who were divided into factions, was one of competition. By luring a servant from a fellow caste member they could cause the latter to lose prestige, for this proved that his servant was not loyal. Strategic interaction between employers and laborers is of course manifested in wage bargaining between employers and laborers, but it could take other forms, especially when we take credit into consideration. In this context, one interesting finding pointed out by Jodhka is the case where landlords who lend to laborers do so with the intention of eventually tying them. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 In light of the above discussion, it is clear that formalizing the movement of tied-laborers by incorporating credit and strategic behavior, will capture a phenomenon that has been observed in many contexts. At a more conceptual level this movement and the consequent shortening of duration can be interpreted in two ways which are related to each other. First, this represents an erosion of the element of patronage that used to characterize the relationship between tied-laborers and employers. This point has been made forcefully by Breman in the context of the disintegration of the hali system in South Gujarat, but it has also been made by others. Second, some people (e.g. Janakirajan (1997)) have seen this as characterizing “freedom” of laborers, as they are able to choose employers. However, as I pointed out in Chapter 2, this is a contentious issue on which there is a substantial methodological debate, especially because as pointed out by Brass (see Brass (1999) pp.208) some amount of such transfers of laborers occurred even under situations that most people would agree as unfree. As I pointed out in Chapter 2, it is also difficult to “measure” freedom, and notions of freedom only have utility in explaining historical change. Even if a formalization of this process will not be able to resolve methodological debates, to the extent that this process can explain institutional change, it is a worthwhile exercise. Moreover, it does give certain insights into the factors that can contribute to the erosion of patronage. Having noted this, when we observe the theoretical literature on tied-labor, we can see that it has shed light on several aspects of the institution - (i) the various Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 rationales for its existence, (ii) the impact of capitalist development (as reflected by mechanization, yield-improving innovations etc.), (iii) the impact of changes in financial markets, (iv) the impact of socio-economic factors like migration, etc. It has also blended very well with empirical findings from various parts of the world. However, the issue of mobility of tied-laborers and duration of their association has not been addressed. Hence, modeling this phenomenon would be a significant intervention into the theoretical literature on tied-labor. I, therefore develop a model of tied-labor which formalizes the movement of tied-laborers from current employers to new employers or to another source of livelihood. The model that I develop incorporates the role of credit and strategic behavior. The details of the model will be presented later, however I will briefly describe the model and the intuition for the main results. The model is based upon the framework of search and matching (for a discussion of the relevant literature, see Pissarides (2000 b)). In this model, there are two kinds of landlords - those who have tied-laborers and those who do not. There are two kinds of laborers - those who are employed as tied-laborers and a pool of casual laborers. The landlords are different in terms of the wages and credit that they can offer to the tied-laborers. The tied- laborers are different in terms of the debts they owe to their current employers. Landlords, who do not have tied-laborers and tied-laborers, search for matches. The search of the laborers is on the job i.e., they do not need to leave the current employer to search. The incentive for a tied-laborer to search is that he can find a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 better wage, and the incentive for the landlord to search is that he can find a tied- laborer. If a tied-laborer finds a match that can pay him a better wage, then he would like to move. However, before he can move, he has to repay the debt to the current employer. If the wage he finds is high enough, then he can use the wage and credit that he can obtain from the new employer and move. In case he continues to work for the current employer, he does not need to repay the debt immediately, but he can roll it over and pay it during his life-time. On the landlord’s side, if he finds a match, he would hire the tied-laborer if it is profitable to do so, and if he has the ability to lend enough so that the laborer can repay the debt to his current employer. Using this model, I will show that the higher the debt of a tied-laborer to the current employer, the less is the likelihood that the laborer will move from the current employer. As I mentioned in Chapter 2, this is a result that I obtained from my fieldwork. The intuition for this result is as follows. If there are no constraints on borrowing then once the laborer finds a landlord who can pay a higher wage, he can move because he will always get enough credit from the new employer to move. However, if there are limits on the amount he can borrow, there are two possibilities. First, he might not get enough credit so that he can pay off the current employer. Second, if the laborer continues with the current employer, he can roll over the debt, whereas if he moves to another employer, he needs to repay all the debt to the current employer. If there are constraints on the amount laborers can borrow, then this would mean that once the laborer moves after repaying the current employer, he Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 will be limited in the amount he can borrow to smooth consumption. The higher the debt, the more restrictive this would be. I introduce an alternate source of borrowing for laborers and show that this has interesting implications. Once there is an outside source of borrowing, we can introduce the possibility of the laborer leaving tied-labor itself. The model predicts that the higher the debt to the current employer, the lesser is the likelihood for the laborer to leave tied-labor. The intuition for this is quite similar to the case where the laborer moves to another employer. Introduction of an outside source of borrowing also helps us capture the impact of development of financial markets. The model predicts that an improvement in financial markets, as reflected by a decrease in the transaction costs associated with outside borrowing will increase the welfare of the laborers. The remaining part of the chapter is organized in the following manner. Section 2 presents a review of the theoretical literature on tied-labor and the gaps addressed by my contribution. Section 3 presents the basic model and some simple extensions to the model and Section 4 concludes. 2. Literature Survey While labor attachment has been studied for a long time by historians and economists, analytical models are more recent. In what follows, I briefly review the literature on labor-tying. Excellent reviews of the literature are present in Basu (1997) and Bardhan and Udry (1999). The purpose of presenting another review is as Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 much to present existing literature (including some models that are not presented in the above reviews) as it is to locate my intervention. Adopting the classification of Bardhan and Udry (1999), we will first consider models where the tied-laborers and casual laborers do the same tasks. The seminal models are Bardhan (1979), (1983). The argument in Bardhan (1979) hinges on the seasonality of agricultural production and costs of recruitment. The agricultural year is divided into a slack season and a peak season. The costs of recruitment, which are a measure of the tightness of the labor market, are higher in the peak season as compared to the slack season. The employer, who has monopsony power, has the incentive to offer a contract which pays the laborer a wage higher than the marginal product in the slack season, so that labor availability is assured in the peak season. Similarly, the wage in the peak season is lower than the marginal product in the peak season. The rationale for labor-tying in this model is the saving on recruitment costs in the peak season from labor-tying arrangements made in the slack season. Using this model, Bardhan derives the comparative static result that an increase in recruitment costs, which reflects an increase in the tightness of the market, increases the importance of tied-labor (in terms of the proportion of the tied- laborers to the total labor force). Bardhan (1983) explains labor-tying as a voluntary arrangement entered into by both tied-laborers and employers. The rationale for the exie for the exilabor hinges on the seasonality of agricultural production, the agricultural year being Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 divided into a slack and a peak period. For the laborers, the incentive for entering into the arrangement is to smooth consumption, and for the employers, it is the assurance of labor availability in the peak period. Using this model, Bardhan shows that tied-labor may be strengthened by capitalist agricultural development under certain conditions. In particular, (a) Yield-increasing improvements increase the importance of tied-labor as a proportion of total labor employed (b) Labor-saving technical progress, like some kinds of mechanization, reduces the importance of tied- labor as a proportion of total labor employed (c) The larger the total number of workers, the smaller is the importance of tied-labor as a proportion of total labor employed. The rationale for labor-tying in Guha (1989) also hinges upon seasonality, but it is based on the efficiency theory of wages. In the standard nutrition based efficiency wage model, productivity depends positively upon consumption, or nutrition. x In Guha’s model, the link between productivity and nutrition which is a characteristic of efficiency wage models operates with a lag of one period. There is no work in the lean season but laborers can generate income through a diminishing returns process (e.g. using the village commons). The casual laborers use the above process in the lean season, and in the peak season they work at a casual wage rate. The tied-laborers get paid by the employers both in the lean season and in the peak season. For the laborers, the choice between casual labor and tied-labor depends 1 A good discussion of the nutrition based efficiency wage model is in Bardhan and Udry (1999), Ch 4. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 upon the income streams from these two options. For the employers, there is a trade off in hiring tied-laborers because to get the productivity gain in the peak season, the employers have to pay the tied-laborers in the slack season (even though there is no work). In this model, there are two possible outcomes - (i) Only casual labor is employed (ii) Both casual and tied-laborers are employed. Guha also demonstrates all the comparative static results that Bardhan and Rudra observed - Decline in population and rise in productivity or demand, increase the importance of tied-labor (expressed in terms of the proportion of tied-laborers to the total labor). In all the above models, the tasks performed by the casual laborers and the tied-laborers are the same. Eswaran and Kotwal (see Eswaran and Kotwal (1985)) focus on the case where the casual laborers and tied-laborers perform different tasks. Seasonality still exists in agriculture however this is not what drives the result. The tasks that are required in the two periods are different. In the first period of the agricultural year, the tasks require more skill and are difficult to monitor (e.g. water resource management, application of fertilizers, maintenance of draught animals etc.) whereas the tasks in the second period are routine and less difficult to monitor (weeding, harvesting, threshing etc.). While casual laborers work only in the second period, tied-laborers work in both the periods. In essence, the model argues that the institution of tied-labor exists because it can elicit loyalty and trustworthiness from laborers - qualities that are important for the tasks that require skill and are difficult to monitor. Even though, the tasks in the first period are difficult to monitor, if the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 tied-laborer shirks in the first period, the outcome of this will be revealed at the end of the year (i.e. at the end of the second period). If the tied-laborer shirks in the first period, the contract can be terminated at the beginning of the second period. The threat of termination is real because the contract that is offered to the tied-laborers is over and above the opportunity income. The threat is credible because termination of the contract will not make the employer worse-off since the employer has a pool of workers at his/her disposal at the beginning of the agricultural year. Eswaran and Kotwal derive comparative static results similar to those in Bardhan (1983). Mukherjee and Ray (see Mukherjee and Ray (1995)) make an interesting intervention into the tied-labor literature. They introduce the possibility that the tied- laborer can default on the contract that is offered. The laborer has an incentive to do this because in the peak season, the wage that the laborer obtains (in the implicit contract) is lower than the wage for casual labor. The employer terminates the contract when the laborer breaches it, however the laborer can come back unscarred to get either a tied-labor or a casual labor contract in the subsequent periods. Hence, the employer has to give an incentive-compatible contract, which involves a premium to discourage the tied-laborer from defaulting. Mukherjee and Ray show that in this model casual labor and tied-labor can coexist. They also derive two interesting comparative static results - (i) any social or economic factor such as the breakdown of patron-client relations or increased migration of laborers that will lead to an increase in turnover rates, will lead to a fall in the proportion of tied-labor Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 contracts offered, (ii) a reduction in information regarding worker histories reduces the proportion of tied-laborers. Casselli (1997) introduces the possibility that laborers can obtain consumption credit. The introduction of borrowing opportunities has interesting implications for the evolution of tied-labor. As the financial sector becomes more efficient (as measured by a decrease in the transaction costs involved in borrowing) the proportion of tied-laborers decreases. He uses an extension of the model of Eswaran and Kotwal. The intuition behind the result is the following. Since casual laborers have irregular income, they gain from using the borrowing to smooth their consumption. There are two kinds of tied-laborers, those who are honest and those who shirk. Since honest tied-laborers have a smooth income path, they do not need to borrow, however tied-laborers who shirk can gain from borrowing. If we disallow tied-laborers from borrowing (since allowing tied-laborers to borrow would mean that the dishonest tied-laborers would reveal themselves), an increase in transaction costs would increase the well-being of the casual laborers, and this would make shirking more attractive. We have to note that for the above to hold, there should be full employment otherwise the unemployed casual laborers will bid the casual wage down and thereby offset the benefits from borrowing. Schaffner (1995) considers two aspects of labor-tying that have not been analyzed in the literature viz. “servility” and “limited horizons.” The first feature, “servility” refers to the docility (some other terms that can be used to describe this Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 are subjection, unfreedom, deference etc.) that is displayed by laborers involved in long-term contracts, a behavior that laborers find unpleasant, but which is desired by the employers. Some aspects of servile behavior are, being at the beck and call of the employer, showing loyalty and respect to the employer (which in some cases included supporting them in local politics) etc. The second feature, “limited horizons” refers to the phenomenon where the employers would like to limit the physical and mental horizons of the laborers by preventing them from getting into contact with non-servile workers or allowing them to stay with other agricultural workers, but preventing “outsiders” from coming into contact with them. The essence of “limited horizons” is minimum day-to-day contact with outsiders, rather than the absence of outside opportunities. Schaffner incorporates servility and limited horizons into a model of two-tiered agrarian labor markets, to derive two interesting comparative static results. First, a tightening of the labor market, due to capitalist development, leads not only to an increase in the proportion of tied- laborers (as predicted by Bardhan (1983) and Eswaran and Kotwal (1985)), but also to a change in the composition of tied-labor. As a result of the tightening of the labor market, the share of the tied-laborers of whom servility is required falls. In a sense, capitalist development leads to the withering away of servility. Second, employers may reduce employment and production relative to what they would be in case there was no market for servility. The welfare consequences of this are ambiguous, but if servility benefits employers to indulge in rent seeking activities, then the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110 consequences of the above are socially wasteful. Apart from the above two comparative static predictions, this model also has implications for agrarian revolts and the well-being of laborers, as compared to what it would have been if the labor market had been competitive. It is clear from the above survey that none of the above contributions has addressed the issue of movement of tied-laborers or strategic behavior. The model that I develop fills this gap. I will describe in greater detail, the setup of the model, but at the outset, I want to clarify that the model incorporates insights from both classes of models described above. Tied-Laborers and casual laborers perform different tasks as in Eswaran and Kotwal (1985); however, there is no element of patronage and loyalty in the model, which is the thrust of Eswaran and Kotwal’s argument. It is worthwhile to point out that Mukherjee and Ray’s analysis does deal with the issue of tied-laborers entering the casual labor market. But this is because of default and this situation is different from the situation where laborers choose to leave tied-labor at the outset. In one of the important extensions of the model, I consider the implications of availability of alternate sources of credit to the laborers. This comes close in spirit to Casselli’s contribution above. 3. A Simple Model In what follows, I will describe a basic model that formalizes the movement of tied- laborers. I will then present two simple extensions to this basic model. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I l l 3.1. Description of the Economy 3.1.1 Time and Seasonality There is one good in the economy. There is one period (a year), the beginning of which is denoted by date 0, and which is divided into a slack season and a peak season. There is no production of output during the slack season, while there is production of output in the peak season. There is work both in the slack and peak seasons. To make the model simple, we assume that there is no uncertainty, however we can incorporate uncertainty without changing the results that we obtain. 3.1.2 Laborers and Labor Contracts There are two kinds of labor contracts - casual labor and tied-labor. Casual labor contracts are spot market contracts available in the slack and peak seasons. A laborer and an employer enter into this contract for a certain wage. Tied-labor contracts are available only in the slack season. Tied-laborers enter into contracts with employers for equal payments in both the slack and peak seasons, for work to be performed both in the slack season and the peak season. This form of the tied-labor contract is standard (e.g. see Mukherjee and Ray (1995)) and captures the idea that the contract helps smooth consumption. Tied-laborers can also borrow from their employers during the slack season. A tied-labor contract specifies the wage and the interest rate on borrowing that the laborer needs to pay to the employer. In case a tied-laborer borrows from the employer, he/she needs to repay the principle and the interest during the peak season. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 Tied-labor involves different and physically more difficult tasks as compared to casual labor. However, unlike in Eswaran and Kotwal (1985) we abstract away from considerations of moral hazard etc. by assuming that the employer monitors these tasks. This is for ease of exposition, more importantly unlike in Eswaran and Kotwal (1985), we are dealing with a situation where there are no elements of patronage (display of loyalty, display of generosity etc.). Moreover, in reality in several contexts, the employer does monitor several tasks of the laborer. A good example of this situation, at least in one context is provided by Brass (Brass (1999), pp. 108) who points out that, Employers taunted their indebted permanent laborers with the constant refrain that ‘You have taken my money, Why aren’t you working’; from the workers viewpoint a major shortcoming of the extended work relationship was that many of his tasks were carried out in the house of the creditor- employer, and thus under the continuous supervision of the latter. Also, unlike in Mukherjee and Ray (1995) we assume that once a laborer enters a tied-labor contract he cannot default on the contract, by working as a casual laborer in the peak season. The utility function of each tied-laborer is assumed to be, U(cs,cp) = u(cs) + /3u(cp) - v u' > 0,m" < 0,0 < ft < 1 cs and cp are the consumptions in the slack season and the peak season respectively, ft is the seasonal discount rate, and v is the discounted marginal disutility of being a tied laborer. This is a measure of both the intense work done by tied-laborers and their dependence upon their employers. The form of the utility function captures the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 desire to smooth consumption on the part of the laborers and also the disutility from being engaged in tied-labor. Tied laborers are also indebted to their employers and are tied because of historical reasons that we abstract away from. As mentioned earlier, there is a pool of casual laborers, whose problem is being abstracted away from. The landlords and tied-laborers take the casual wage rate during both the slack season and the peak season as given. These are denoted by wc s and wc p respectively. 3.1.3 Landlords There are M landlords who have different landholdings. For ease of exposition, the landlords are represented in the following manner. Let tt,- be the profit of the landlord i and it depends upon whether the landlord has a tied-laborer or not. In case the landlord has a tied-laborer, then his profit from agriculture is represented as n\ (w, w] ,wc p,A, Lx , p) where the variables are as defined earlier and w, L, and A are the wage offered to the tied-laborer, land owned, and a technological parameter respectively. In case the landlord does not have a tied-laborer, then his profit is given by it* « ,wc p,A,Li,P') .Let the casual labor demand functions for landlord i in the cases where he has a tied-laborer and where he does not be CL\ (w, wc s,wc p, A, Li,P) and CL] (vt/, wp, A, Lx ,, /?) respectively. Note that these two labor demand functions are vectors because the landlord hires labor in both the slack and peak seasons. The assumption that the discount rate for the landlords is the same as the one for tied- laborers is made for ease of exposition and is not required for the results. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 In case a landlord has a tied-laborer, the tied-laborer can borrow from him, in which case, he can charge interest at an exogenous annual rate of R, which is also the opportunity cost of lending for each landlord. The assumption that R is the opportunity cost is not required for the results, but is made for ease of exposition. It simplifies the model considerably because we can abstract away from considerations of profit making from lending. In other words, we can focus on the labor-hiring role of the employer, while ignoring the money lending role. This is also a realistic assumption in light of the fact that in many studies it has been pointed out that landlords usually lend at a discount to their tied-laborers. We assume that /? R < 1 so that the laborer has the incentive to borrow. There are two crucial assumptions about landlords. The first assumption is that no landlord can hire more than one tied-laborer. We have to note that to hire a tied-laborer an employer needs to have the ability to give a wage in the slack season, and also if possible lend. This assumption says that none of the employers is “large” enough to hire more than one tied-laborer. As I show later, this assumption simplifies the model significantly. There are two groups of landlords - those with tied-laborers and those without. We represent these as Group 1 and Group 2 respectively. Given our assumption above, there are N landlords with tied-laborers and (M-N) without tied-laborers. The second assumption is that the amount a landlord can lend is limited. This is a reasonable assumption, given the fact that usually the sources of borrowing (e.g. institutional sources) and working capital at the disposal of the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 landlord are also limited. We assume that the amount that each landlord can lend is in the range [o, Z) ]. The assumption that the landlords are equal in terms of their lending ability is not required for the results, but is made for ease of exposition. Differences in lending can easily be introduced without changing the results of the model. 3.1.4 Timing and Sequence of Events Apart from the fact that each employer can hire only one tied-laborer, there are other market frictions. We assume that every landlord who wants to hire a tied-laborer cannot do so. Similarly, every laborer who wants to find a landlord cannot do so. We capture these frictions using a standard matching function, common in the literature on unemployment theory (see Pissarides (2000 a)). We assume that the number of matches, which is the number of laborers who find employers and also the number of employers who find laborers, is a function of the number of laborers and employers searching. This matching function is denoted by, m(X, Y) where X and Y are the number of laborers and landlords looking for matches respectively. This matching function has the standard properties - positive first derivatives in both the inputs, negative second derivatives in both inputs, and homogeneity of degree one. There are two other crucial assumptions regarding the landlords. We assume that all the laborers have the option of continuing working with their current employers at date 0 at a wagew . This wage is exogenously given and is not a choice variable on the part of the employers. What is important for the results below is that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 this is not chosen by the current employers by incorporating the fact that laborers will search with this offer in hand. The rationale for this assumption is that since tied-laborers at date 0 have worked for the employers in the past, the terms of their contracts are influenced by past variables. For example, if the laborers had received w in the past, then it is realistic to assume that the employers cannot offer wages below this and so they will offer this wage. This assumption also makes sense because we are primarily looking at a situation where the relationship structured in the past could breakup. As shown later, the results obtained from the model, can be obtained under a weaker condition and it is not necessary for the wages at which the tied-laborers can continue be same for all the landlords. As pointed out earlier, landlords are different in terms of their landholdings, and hence in terms of the wages they can offer to tied-laborers. Once a match is found between a laborer and landlord the wage that the tied-laborer gets is given by W j. w, is exogenously given (i.e. not determined within the model) and can be interpreted as a result of some surplus sharing rule between the landlords and laborers. In the literature on matching, wages are determined either using the framework of Nash Bargaining between firms and laborers or by “Wage Posting,” where firms post wage offers and workers either accept these offers or reject them (for a discussion on this issue, see Pissarides (2000 b)). In the simplest possible wage posting rule (see Diamond (1971)), firms pay the reservation wage, and hence this is equivalent to a Nash Bargaining situation where the workers get no share of the total Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 surplus. For simplicity, we abstract away from how this wage is determined, although this can be incorporated into the model without changing the results. The fact that W j varies from one landlord to another, is a reflection of the heterogeneity among landlords. The second assumption I make is that landlords in Group 2 are such that the values of w, are distributed in the range [wm in , wm a x ] with a distribution function G and a mean /u. This assumption implies that when a tied-laborer searches for a match, the wage he can expect to receive is from the above distribution. We have to note that the wage policies offered by landlords in Group 1 are different from those offered by landlords in Group 2. It is worthwhile to reiterate here that landlords can always hire casual laborers from a pool that is available to them. The landlords hire casual labor “residually,” i.e. after they go through the search process for a tied-laborer. After the landlords find out whether they have a tied-laborer or not, they decide the amount of casual labor they need and hire casual labor accordingly. Also, while landlords can hire only one tied-laborer, they can hire all the casual labor that they need. The wages at which casual labor is hired are exogenously given. As mentioned earlier, unlike in previous models of tied-labor, the choice between casual labor and tied- labor is not explicitly modeled here The sequence of events is as shown below, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 Slack Season (1) All laborers and landlords in Group 2 (who do not have tied-laborers) search for matches. Note that this implies that laborers can search “on the job.” This also implies that the landlords who have tied-laborers (i.e. those in Group 1) do not search. (2) Matches are found. A successful match occurs if a laborer gets enough credit from the new employer so that he can payoff the current employer, and also gets more utility from the new employer as compared to what he would get from continuing with the current employer. (3) Those laborers who find successful matches leave their current employers and those who do not, continue with their current employers. (4) Landlords hire casual labor for slack period operations. (5) Consumption occurs. Peak Season (6) Landlords hire casual labor for peak period operations. (7) Production and consumption occur. 3.2 Determination of the Equilibrium The definition and determination of the equilibrium is as follows. The equilibrium is defined as the sequence, ((pi, w , CL\ (W -, Lt, A), CL] « , w; , L,, A)), (p) , w ., CL) (wJ ;wc s,wc p,Lj, A), CL) « ,wc p,Lj, A)), (Pk^kA)) (i = I...N J = 1 ...(M -N ),k = 1...N) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 (1) The first term is for landlords in Group 1, who have tied-laborers at date 0. p) is the probability that landlord /' will retain his tied-laborer, the remaining terms are the wage paid to the tied-laborer, casual labor hired in case the employer retains his tied- laborer, casual labor hired in case the employer does not retain his tied-laborer. In case the tied-laborer does not leave, the employer pays the wage w to him and hires casual labor taking the wages for casual labor as given. (2) The second term in the sequence corresponds to landlords in Group 2. p * is the probability that the landlord j finds a tied-laborer and Wj is the wage offered. Casual labor hired is similar to the case above. (3) The third term in the sequence corresponds to the tied-laborers. p[ is the probability that the tied-laborer k moves, w * is the wage that the laborer gets. We have to note that wk - w in case the laborer stays with the current employer and in case he moves, W k depends upon the landlord that the laborer matches with, bk is the amount the laborer borrows in the slack season. 3.3 Results Let wR be the wage that the tied-laborer needs to draw from a prospective new employer so that he is indifferent between continuing with the current employer and moving to the new employer (i.e. at this wage, he would get the same utility from the current employer and the prospective new employer). This implies that after the tied- laborer searches for a new employer and finds a match he needs more than m P to induce him to move from the current employer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 Let us consider the problem of a typical tied-laborer who has an offer of w and is searching for a new employer. Let D be the debt of the tied-laborer to the current employer. The utility that the laborer gets if he continues with the current employer is the solution to the following problem, (l)m ax{u(cj + /?w(c/,)} {c„cp,b} s.t cs <(w + b) cp <(w-(b + D)R) - D < b < D We do not need to include the disutility term because it is a constant and hence does not affect the tied-laborer’s choice. We have to note that if the laborer continues with the current employer, he will get a wage w and he can borrow any amount in the range [o,d], b is the amount borrowed by the laborer, and the lower bound on b reflects the fact that the laborer cannot save with the employer. Lemma 1: If u'(w + D - D )- /? Ru'(w - DR) > 0 and D > 0 then wR > w Proof: To determine wR, we have to note that this is the wage that gives the laborer as much utility as what he would get from the current employer. The utility that the laborer can obtain from moving to the new employer is given by, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 (2)max{u(c,) + fiu(cp)} {cs,cp,b} s.t cs <(w + b - D ) cp < (w - bR) 0 <b<D w is the wage that is drawn. If we denote the net-borrowing (amount left after repaying the debt to the current employer) by b , we can rewrite the above problem as, (3)max{w(c0) + p w(c,)} {ct ,cp,b} s.t cs <(w + b ) cp < (w -(b+ D )R ) - D < b < (D -D ) (1) and (3) are similar except for the borrowing constraint in (3) which is more restrictive because the laborer has to repay the whole debt before moving. In fact, if there is no borrowing constraint, or if D=0, the two problems are identical. Comparing the problems (1) and (3), we can note that if the optimal amount of borrowing in (1) is greater than (D -D ), then the constraint on borrowing in (3) becomes binding forw = w . If this is the case, the wage w has to be greater than w for the laborer to move. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 The condition for the optimal amount of borrowing in (1) to be greater than (D - D) is given by, u’(w + D - D ) ~ 0 Ru'(w - DR) > 0 Hence, if the above condition is satisfied, thenwfi > w . Q.E.D Lemma 2: If u'(w + D - D ) - 0 Ru'(w - DR) < 0 or if D = 0 then wR = w . Proof: The proof is similar to the above. We have to note that if u'(w + D - D) - 0 Ru'(w - DR) < 0 the optimal amount of borrowing in (1) is less than or equal to (D -D ). If D = 0 then (1), (3) are identical Q.E.D For the analysis that follows, we will focus on the case where D> 0, since we have established that in the case where D - 0 , w r = w . Lemma 3: If u'(w + D - D) - 0 Ru'(w - DR) > 0 then (wR + D - D) <(w + b*) and (wR - DR) >(w-(b* + D)R) where b* is the optimal amount of borrowing if the laborer works with the current employer. Proof: u(wR + (D - D)) + 0 u(wR - DR) = v(D) = u(w + b*) + 0 u(w - (b* + D)R) Since w'(^ + D - D) - 0 Ru'(w - DR) > 0, b* > (D -D ) b* > (D -D ) (b* + D)R > DR =>(w-(b* + D)R) < (w - DR) From Lemma 1, we know that wR >w (since u'(w + D - D) - 0 Ru'(w - DR) > 0). Hence, (w - (b* + D)R) < (w - DR) < (wR - DR) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 Sinceu(wR + (D -D )) + J3u(wR -DR) = u(w + b*) + u(w-(b* +D)R) and (w* - DR) >(w-(b* + D)R) , (wR + D -D )< (w + b*). Otherwise the above equality cannot hold. Q.E.D The above lemma says that when the laborer moves to a new employer at wage w1 * , the consumption in the first period with the new employer would be lower than that with the current employer. Now, we are in the position to state the main result of the model. Before further analysis, we have to note that for the laborer to move, the wage that he gets should be greater thanw* .Hence, the proportion of landlords whom the laborer will move to, in case he matches with one of them, is clearly (1-G(wfl)). Lemma 4: If u'(w + D - D) - { 3 Ru'(w - DR) > 0 then the higher the debt, the more unlikely it is that the laborer will move to another employer. Proof: Given the above condition, the probability that the laborer will move is given by ^ N -’ -& ' N)\ \-G(wr )) where, ^ is the probability that the laborer will find a match given the fact that there are N laborers and (M-N) landlords who are searching. This is a constant. (1-G(wfi))is the probability that the laborer, having found a match, is matched with a landlord who can pay higher thanw*. u{wR +(D -D )) + /3 u(wR - DR) = v(D) = u(w + b*) + fi u(w - (b* + D)R) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 b* is the optimal amount of borrowing from the current employer. If b* is interior, then w'(w + b*)~ (5 Ru'(w -(b* + D)R) = 0. If the borrowing constraint is binding (b* = D) then, u'(w + D) - P Ru'(w ~(D + D)R) > 0 and v(D) = u(w + D) + {3 u(w ~(D + D)R) Differentiating the equation characterizing v f with respect to D, u'(wR + (D - D)){— — 1) + p u\wR - DR)— = - p Ru'(w - (b* + D)R) dD dD dwR = u'(wR + (D - D)) - p Ru\w - (b* + D)R) dD ~ (u\wR +(D-D )) + J3 u'(wR - DR)) dwR u'(wR+(D -D ))-u'(w + b*) If b is interior, = ---------- =>----------- — jy — J .— > 0 because dD (u'(w +(D-D)) + p u\wK- DR)) + D -D )< (w + b*)(from lemma 3 and from the fact that u is concave). Ifb* = D , u'(w + D) - p Ru'(w -(£) + D)R) > 0 u'(wR +(D-D )) > u'(w + D)> p Ru'(w ~(D + D)R) dwR = u'(wR + ( p - D ) ) - p Ru’(w -(b*+ D)R) dD ~ (u'(wR +(D-D)) + p u ’(wR -DR)) . u\ wr + ( D -D )) - u'(w + D± :Q (u\wR +(D-D )) + P u'(wR - DR)) » £ z G £ ! ! 2 = _g V ) — < o dD K J dD Hence, the higher the debt, the lower is the probability of moving to a new employer. Q.E.D Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 The intuition for the above result is that if debt is in the range where the borrowing constraint becomes binding when moving to the new employer, as debt increases the borrowing constraint becomes more and more restrictive. Hence the wage from the new employer has to be higher to induce the laborer to move. We can note that if there is no borrowing constraint, thenw* = w and the level of debt does not influence the probability that the laborer will move. We have to note that the above lemma says that if a laborer searches with a certain wage offer in hand then the higher his debt to the employer the more unlikely it is that he will move. However, it does not say anything about two different laborers in equilibrium because they could have two different wage offers. This is where the assumption about the offers made by the current employers becomes significant, as shown in the proposition below. The condition, u'(w + D - D ) ~ J3 Ru'(w - DR) > 0 can be rewritten in a more intuitive way as D> D* ,D* being defined by u'(w + D - D*)- ft Ru'(w - DR) = 0 .This is the threshold above which debt becomes important for movement. Proposition 1: In equilibrium, if Dt > D* thenDt > Dj => p\ > p l } Proof: From Lemma 4, we know that given the wage, the higher the debt of a laborer to his current employer, the lower is the likelihood that he will move (provided that this debt is higher than D*). From our assumption, all laborers have Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 the same wage offers, w from their current employers. Hence from Lemma 4, A >Dj =>p‘ > p ‘ We have to note that the size of Dj relative to A* does not matter. This is the main result of the model, that if there are limits to borrowing, then in equilibrium, laborers with lower debts are more likely to move as compared to those with higher debts, provided that these debts are large enough. This result captures the essential idea of how indebtedness can hinder mobility. It also captures the idea that as long as laborers can maintain small debts to their employers, they can move. As mentioned earlier, the above proposition holds under a weaker condition on the wage offers from continuing with the current employer. To see this, consider two laborers i and j. Now relax the assumption that the wages at which these laborers can continue with their current employers are the same viz, w .Let w, and w; be the wages paid to laborers i and j respectively in case they continue with their current employers, and let A and Dj be their debts to current employers. We can characterize threshold levels of debt, Z )*, and D*j and wages at which these laborers are indifferent between continuing with the current employers and moving, wR, and wR j. Note that these are different because the wages at which the laborers can continue with the employers, w, and w} are different i.e. w, ^ Wj. Now, consider the situation where A > Dj. If w, > Wj then it implies that the laborer with a higher debt gets a higher wage from continuing with the current employer. In this case, it is quite obvious that the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 proposition holds. To illustrate this, consider the case where debt does not hinder mobility (say because the limit on borrowing, D is very high) then the probabilities that / and j will move are ^ )) and m(N (M-N)) — — - :------ — (1 -G(wj)) respectively. Since w, > Wj and G is increasing, the N probability that i will move is lower. This result is only strengthened if the borrowing constraint becomes effective. The most interesting situation arises when w, < Wj, i.e. when the laborer with lower debt has a higher wage, if he continues with his employer. In this case the proposition holds if the difference between the wages Wi and w is small. To see why this is the case, consider the expression for w* when the borrowing constraint is binding, u(wR +(D -D )) + J3 u(wR - DR) = u(w + b*) + j3 u(w - (b* + D)R) (as earlier, b* is the optimal borrowing).It is obvious from the above that higherw s correspond to higher w* s. So, when we consider the probabilities that i and j will move, we have to realize that the impact of wages at which laborers can continue with their employers and the impact of debt go in opposite directions. In other words, while higher debt can hinder j, the fact that he receives a lower wage means that he has a higher probability of finding a new employer who can offer him a better wage. For the proposition to hold, the impact due to the debt has to dominate the impact due to the wage from continuing with the current employer, and this is the case if the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 difference between the wages from continuing with the current employers for different laborers is small. The limiting case is where the difference is zero, which is the case considered in the analysis (i.e. where all laborers can continue with their current employers at w). A sufficient condition that can ensure that the proposition will hold can be derived in the following manner. Consider the debts that the tied-laborers have to their current employers, let Dm m and Dm ax be the minimum and maximum debts. This implies that no tied-laborer has a debt lower than Dm in and a debt higher than Dm ax. Similarly, consider the wages at which tied-laborers can continue with their current employers, let wm in and w m a x be the minimum and maximum wages. This is the range of wages offered by landlords in Group 1, and note that this is not the same as the range of wages offered by landlords in Group 2. Letw* (w,D)be the wage at which a tied-laborer who has a debt D and who can continue with the current employer at a wagew, is indifferent between accepting the offer of the current employer and moving. It is obvious that the higher the value ofw* , the lower is the probability of moving. It is alt is also woile to reiterate here that wR is increasing in w and non increasing inD. The sufficient condition isw*( wm ax,D) < w*( Wmm,Z)') VD' > D . Consider two laborers. This condition implies that the range of debts and wages is such that even if the laborer with the lower debt gets the maximum possible wage, and the laborer with the higher debt gets the minimum possible wage, the likelihood that the laborer with lower debt will move is much higher than that of the laborer Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 with higher debt. It is obvious that if this condition is satisfied, we will be able to derive the proposition given any distribution of wages and debts over the laborers. This condition specifies a restriction on the wages, given the distribution of debts. Note that this condition will trivially hold ifwm a x = wm m = w . Now, we will consider an interesting comparative-static effect. An issue of interest in the context of the above model is the impact of technology. If technology improves, it will increase the profitability from the employment of both tied-laborers and casual laborers. If the impact of a technological improvement makes the employment of tied-labor more attractive, then this will have implications for mobility. If technological improvement makes tied-labor more attractive and it affects all landlords equally (i.e. if it does not affect the distribution), then this has the impact of pushing the range , wm a x ] to the right. This will in-tum mean that it is more likely with the improved technology to find a wage higher than for each laborer, which in-turn means that the likelihood that the laborer moves is higher. This is expressed in the proposition below. dw. Proposition 2: If > 0 V/ and the distribution ofw is unaffected by this technological change then — - > 0 \/k 6A Proof: The proof is straightforward and follows from the fact that — — ^ > 0 dA Q.E.D Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 For further analysis, the following notation is useful, v(D) - Utility of a tied-laborer with debt D, from continuing with the current employer. vn(w,D)~ Utility of a tied-laborer with debt D, from moving to a new employer, if the wage that he gets from moving is w. D* - Threshold level of debt, given by u'(w + D - D *) - f3 Ru'(w - DR) = 0 . wR - Wage at which the laborer is indifferent between continuing with the current employer and moving to the new employer. It is a solution to the following equation, u(wR +(D -D )) + J3u(wr -DR) = v(D).wR = w if D<D* P(D) - Probability that the laborer will stay with the current employer. 4. Extensions 4.1 Alternative Source of Credit So far we have been concerned only with movement from one employer to another. This is because there is no chance for the laborer to move out of tied-labor since the only source of credit is the employer. I will introduce an alternative source of borrowing in a very simple manner on the lines of Casselli (1997). Assume that apart t apart m plcb extent the laborer can borrow from sources like relatives, local moneylenders. However, these sources are usually either limited or the interest rate on these is very high. Also, alternative sources of borrowing are very uncertain. In some cases, this alternative source of credit is related to the occupation the tied-laborer moves to. For Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 example, in the case pointed out by Breman, some tied-laborers moved to work in brick kilns and the owners of these provided credit. In this context, we have to realize, as I pointed out in Chapter 2, that the cash advance paid by the employer is an interest-free credit and to a certain extent satisfies the credit requirements of the laborer. To capture the fact that alternative sources of credit are usually more expensive than the credit offered by the employer, we assume that the rate of interest on this outside source is OR where 0 >1. This also captures the transaction costs involved in outside borrowing. To capture the uncertainty involved in the outside option, we assume that the maximum amount that the laborer can borrow from this source, B follows the distribution H(B) defined in the range [o, 2?]. Once the laborer finds B, he can borrow any amount less than or equal to B. To visualize this, imagine the tied-laborer getting an offer like “I can lend you a maximum amount of 10,000.” To keep matters simple and to make things more realistic, it is assumed that this outside source is not available to the laborer to smooth consumption in case the laborer remains as a tied-laborer. There is an alternative occupation to tied-labor at the disposal of the laborer in case he can leave tied-labor. We will refer to this as an “alternative occupation” (AO). Examples of this could be working in a town, migrating to another village; casual labor, tenancy etc (see Sarap (1991), Breman (1993) for such alternatives in some contexts). In my fieldwork, tied-laborers suggested some alternatives like Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 traditional occupation, education, apart from casual labor. Let the income streams associated with the outside option be Is and Ip in the slack and peak seasons respectively. Now the laborer can consider three options- (i) Stay with the current employer (ii) Move to another employer (iii) Move to alternative occupation. In other words, with the current employer’s offer in hand, he can try to either leave tied- labor or to find a new employer, or both. This raises tricky issues in strategic behavior. So, to make the analysis simple, I focus on the case where AO is always better than working for the current employer, if it is possible for the laborer to move. This implies that Is and Ip are high enough compared tow and 0 is low enough. Also, the following restrictions are imposed on the laborers, (1) Each laborer finds out the maximum amount that he can borrow from the outside source of credit. In case the laborer gets enough credit to repay his current employer, and he finds AO preferable to working for the current employer, he chooses the amount to borrow, repays the debt to the current employer and leaves tied- labor. (2) Once the above process ends, there are some laborers left. These laborers and the landlords, continue the search process outlined in section 3.2, where there was no outside source of credit. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 Essentially, the laborer is prevented from searching for another employer with AO in hand or evaluating AO after searching for a new employer. Consider the problem of a laborer with a debt D to the current employer, and who is evaluating AO. The utility of the laborer from working with the current employer is represented by v(D) and the utility from AO is represented by v, (Is,7p,D,8). From 8v 8v our assumption v, (Is, I D, 8) > v(D) VZ) and —- < 0, —- < 0. The laborer can move 88 8D only if he can repay the debt to the current employer, for which he needs to be able to borrow an amount of at least D. The probability that the laborer can get this amount is given by (l-H(D)). For laborer /, we will represent this by q\. Now we are in a position to state the main results of the model. Proposition 3: In equilibrium the higher the debt to the current employer, the less likely it is that the laborer leaves tied-labor, i.e. for laborers /' and j, A > A => q\ < q ' t ■ Proof: The proof is straightforward and follows from the fact that q[ = (1 - H{Dk)), k = i, j and the fact that H(D) is increasing in D. The intuition is also straightforward. The higher the debt, the more unlikely it is to obtain the amount required from the outside source and hence the more unlikely it is to move. We can also consider the presence of outside source of credit on the welfare of the laborers and the employers. Let Case (1) represent the situation where there is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 no outside source of credit and Case (2) represent the case where there is the outside source. In Case (1) in equilibrium, the expected utility of laborer i is given by v(D,.) whereas in Case (2), it is given by EU, = H(Dt )v(A ) + (1 - H{Di ))v, (/,, I p, Dt, 0). Since v, (I s ,Ip,D,0)> v(D) VD , the expected utility of the laborer is always higher in Case (2) as compared to that in Case (1). So, the presence of an outside source of credit is welfare improving for the laborers. However, it is welfare decreasing for the landlords. In the case of the landlords who have tied-laborers at date 0, the reason for this is obvious - they lose laborers. In the case of the other landlords, the reason is not obvious, and it works through the matching function. After the initial stage, some laborers leave the tied-labor market, as a result of which there are fewer laborers searching for new employers, and hence fewer matches. The net result is that the probability of these employers finding a tied-laborer decreases, thereby reducing their welfare. Apart from the impact of the presence of an outside source of credit, we can also consider improvements in financial markets as represented by a decrease in 6 which represents a decrease in the transaction costs associated with borrowing. It is obvious that a decrease in the transaction costs increases the welfare of the tied- laborers. This can be clearly seen by differentiating the expected utility of the laborers with respect to 0, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.2 Credit Offer to Retain the Tied-laborer An interesting implication of the above model is that since the tied-laborer’s debt hinders his mobility, if the employer can control the laborer’s debt, he has the incentive to do so. We formalize this idea in the following manner by considering the problem of a typical laborer and a typical employer. Consider the model without outside credit, but with two periods. At date 0, the laborer is working for the employer and he has no other option, but to work for the employer. At date 1, the world is the same as in the previous model, where the laborer has the option of working for the employer or moving to a new employer. Since we do not have outside credit, the possibility of leaving tied-labor does not arise at date 1. At date 0, the laborer has no debt to the employer, but the employer can make a “take-it-or-leave-it” offer of credit so that the laborer has some debt outstanding at date 1, which in turn affects his probability of leaving. The sequence of moves is as shown below, Date 0 Slack Season (1) Employer makes a “take-it-or-leave-it” offer of credit. (2) Laborer either accepts the offer or rejects it. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 Peak Season (1) Laborer returns some portion of the borrowing including interest made in slack season. He could return the whole amount or return nothing Date 1 Slack Season (1) Laborer decides to either continue with the employer or move to another employer. In case he moves to another employer, he repays the whole debt outstanding. Otherwise he can roll over the debt. 4.2.1 The Laborer’s Problem At date 0, in case the laborer takes the offer, he will take into account the fact that he has the chance to move at date 1, and the amount of debt he has at date 1 hinders his ability to do so. In the section below, we will describe the laborer’s problem in greater detail. 4.2.2 The Employer’s Problem Since the employer lends at his opportunity cost, he makes no profit or loss from lending. So, he is indifferent about lending any amount. However, he chooses the amount to lend for “strategic” purposes, i.e. because it helps him to keep the laborer from moving. Suppose the employer makes an offer B. We will first consider the situation where the laborer accepts the offer. Let b* be the amount that the laborer returns in peak season at date 0, it could be 0, but it cannot be negative. The debt at Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 the beginning of date 1 is D = (BR - b*)R. In case the laborer accepts the offer, the laborer’s expected utility at date 0 is given by u(w + B) + pu(w-b*) + p [P(D)v(D) + (1 - P(D))E(vn (w, D))] The above expression needs explanation. The first term is the utility in the slack season of date 0, the second term is the utility in the peak season of date O.The third term is the expected utility during date l.With a probability P(D), the laborer stays with the current employer, in which case he gets a utility v(D), and with a probability (l-P(D)) the laborer moves. The expected utility from moving is given by E(vn (w,D)). E is the expectation operator and vn(w,D) is as defined earlier. In case the laborer rejects the offer, the laborer’s expected utility is given by u(w) + p u(w) + p [/>(0)v(0) + (1 - P(0))E(vn 0 ,0 ))]. This term is similar to the one where the laborer accepts the offer. If the laborer rejects the offer, he has zero debt at date 1 and hence the probability that he will stay with the current employer is given byP(Q) = F(w). When the employer makes the offer, he should ensure that the laborer accepts it i.e. the expected utility of the laborer from accepting the offer should be higher than the expected utility from rejecting it. Therefore the offer of the employer is such that, « 0 + B) + p u(w - b*) + P [P(D)v(D) + (1 - P(D))E(vn(w,D))] > u(w) + pu(w) + p [P(0)v(0) + (1 - P(0))E(vn 0,0))] Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 The above is the participation constraint for the laborer. One possible scenario for the participation constraint is as shown in the figure below. The straight line represents the expected utility of the laborer if he rejects the employer’s offer and the curve represents the expected utility if the laborer accepts the employer’s offer. Figure 4: Participation Constraint of the Laborer It is of course possible for the participation constraint to be always slack at a positive B. This occurs when the constraint on credit is binding, i.e. when the laborer would like to borrow the maximum that he can, which is D . The employer’s profit from making the offer is given by, p\p(D )x‘(w,V',wf ,L,A) + ( \- P m *'’ (w,. y>,, U A)] Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 We have to note that since the employer lends at his opportunity cost, his profit does not depend upon the amount that he lends. Note that7tl(w,ws,wp,L,A) > k c (ws,w p,L,A) otherwise the employer will not hire the tied-laborer at wage w 2 It is instructive to rewrite the above expression as shown below - 7c‘(w,ws,w p,L,A) + P [p(£>)<y (w, L, A ) - n c(ws,wp,L, A)) + n c(ws,wp,L,4)] Note that the above expression is increasing in P(D) implying that the higher the probability that the laborer stays with the employer, the higher is the expected profit. Hence, it is in the interest of the employer to ensure that P(D) is as high as possible. Since P(D) is non-decreasing in D (and increasing in D<=\p*,D ), it is in the interest of the employer to offer as much credit as possible while keeping the participation constraint satisfied. If we consider the case where the laborer has no option but to work for the employer, i.e. if he cannot move at date 1, then we can observe that in this case, the employer is indifferent to the amount lent. This is because the employer does not make any profit or loss from lending. He can even lend the amount that the laborer would have borrowed in case the laborer is allowed to borrow as much as he wants (subject of course to the credit constraint). We can now make an interesting 2 1 will ignore the trivial case where these profits are equal. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 proposition. Let B be the offer chosen by the employer and B be the amount the laborer would have chosen if he had the option of choosing the amount to borrow. Proposition 4: Given the assumption on R, that it is equal to the opportunity cost of the employer, it follows that B > B . Proof: The proof follows from the observation that at B , the participation constraint is slack. If5 < D , the employer can always increase the offer above B and still keep the participation constraint slack. The employer has an incentive to do so because he wants to lend the maximum possible amount. Hence, iff? < D thenB > B . If B = D , then the employer cannot increase the offer beyond f? and would therefore set B = D . Hence it follows that B > B Q.E.D The proposition above clearly demonstrates that when the employer has the bargaining power in the credit market, he can use it to affect the mobility of the tied- laborer. It also shows that in this situation, the employer has an incentive to “over lend,” i.e. lend more than what the laborer would have wanted to borrow, and also lend more than what he would have lent in case there are no other competitors for the tied-laborer. While the above result hinges on the fact that R is equal to the opportunity cost of the employer, it can be easily extended to other situations where R could be different. In case R is lower than the opportunity cost (so that the employer offers a discount) then the employer makes a loss by lending. In this case there is a trade-off between profit from keeping the tied-laborer and loss from lending, and we can Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 easily characterize the conditions under which the employer makes an offer that is higher than what the laborer would like to borrow. In case R is higher than the opportunity cost, the employer makes a profit from lending and the result is strengthened. This is because the employer has two incentives to lend more - profit from lending and increasing profitability from keeping the laborer. It is interesting to discuss the efficiency of the equilibrium when the current employer can make credit offers. Let us compare two scenarios -1 and 2 corresponding to situations where the current employers can make credit offers and where they cannot. From the above analysis, it is obvious the welfare of the current employers is higher in scenario 1 as compared to 2, and the reverse is true for the new employers and the tied-laborers. However, the total surplus could be higher either in scenario 1 or 2 depending upon the profit functions of landlords in Groups 1 and 2. If the average profitability from hiring tied-labor for landlords in Group 2 is sufficiently higher than that for landlords in Group 1, then by reducing the mobility of the tied-laborers, the current employers reduce the total surplus, i.e. the total surplus is higher in scenario 2 as compared to that in scenario 1 . 5 Conclusions In the above analysis, the movement of tied-laborers among employers, and the movement from tied-laborabor to(.Uive sources of livelihood have been formalized. Credit and strategic behavior, the main factors that have a bearing for this mobility have also been incorporated into this model. In this process, I have provided insights Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 into a phenomenon that has hitherto been unexplored in the literature on tied-labor, but which has been observed in many contexts. I have also demonstrated certain comparative-static effects like the impact of changes in technology and the introduction of alternative sources of borrowing for this movement. Far from being a purely theoretical exercise, the analysis above has important implications for the welfare and “freedom” of tied-laborers. The analysis shows clearly that the introduction of alternative sources of borrowing has important implications for the welfare of the laborers and their ability to leave tied-labor. The analysis also shows clearly that under certain conditions employers can successfully impede the mobility of the laborers by using their bargaining power in the credit market. While the above analysis has provided significant insights, the model clearly puts the impetus for the movement only on the tied-laborers. As pointed out in the introduction, some impetus for this movement does come from the employers. Also, this model does not explicitly consider the case of casual laborers who might also be searching for tied-labor contracts. Further research can attempt at formalizing these aspects. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 BIBLIOGRAPHY Abreu, Dilip. “On the Theory of Infinitely Repeated Games with Discounting.” Econometrica, March 1988, 56 (2), pp. 383-96. Amin, Shahid. Sugarcane and Sugar in Gorakhpur: An Inquiry into Peasant Production fo r Capitalist Enterprise in Colonial India, New Delhi, India: Oxford University Press, 1984. Bardhan, Pranab. “Wages and Employment in a Poor Agrarian Economy: A Theoretical and Empirical Analysis.” Journal o f Political Economy, June 1979, 87(3), pp. 479-500. _______________ . “Labor Tying in a Poor Agrarian Economy: A Theoretical and Empirical Analysis.” Quarterly Journal o f Economics, August 1983, 98(3), pp. 501- 14. _______________ . Land, Labor and Rural Poverty: Essays in Development Economics, New York, NY: Columbia University Press, New York, 1984. _______________ “A Note on Interlinked Rural Economic Arrangements,” in Pranab Bardhan ed. The Economic Theory o f Agrarian Institutions, Oxford, UK: Clarendon Press, 1989. _______________ “The Concept of Power in Economics.” Economics and Politics, November 1991, 3(3), pp. 265-77. ________________and Rudra, Ashok. “Interlinkage of Land, Labour and Credit Relations: An Analysis of Village Survey Data in India.” Economic and Political Weekly, February 1978, XIII (6, 7), pp. 367-84. _______________ a n d ______________. “Terms and Conditions of Sharecropping Contracts: An Analysis of Village Survey Data in India.” Journal o f Development Studies, April 1980, 16(3), pp. 287-302. _______________ a n d ______________. “Terms and Conditions of Labor Contracts in Agriculture: Results of a Survey in West Bengal 1979.” Oxford Bulletin of Economics and Statistics, February 1981, 43(1), pp. 89-111. ________________and Udry, Chris. Development Microeconomics, Oxford, UK: Oxford University Press, 1999. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 Basu, Kaushik. Agrarian Questions, New Delhi, India: Oxford University Press, 1994. ____________ . Analytical Development Economics, Cambridge, MA: MIT Press, 1997. ____________ and Bell, Clive. “Fragmented Duopoly: Theory and Application to Backward Agriculture.” Journal o f Development Economics, Jan 1991, 36 (1), pp. 145-65. Bernheim, B. Douglas and Whinston, Michael D. “Multimarket Contact and Collusive Behavior.” Rand Journal o f Economics, Spring 1990, 21 (1), pp.1-26. Bhadhuri, Amit. “A Study in Agricultural Backwardness Under Semi-Feudalism.” Economic Journal, March 1973, 83 (329), pp. 120-37. Bhalla, Sheila. “New Relations of Production in Haryana Agriculture.” Economic and Political Weekly, 1976, XI (13), pp. Binswanger, Hans; Doherty, Victor; Balaramiah, T.; Bhende, M.J.; Kshirsagar, K.G.; Rao, V.B and Raju, P.S.S. “Common Features and Contrasts in Labor Relations in the Semiarid Tropics of India.” in Hans Binswanger and Mark Rosenzweig eds. Contractual Arrangements, Employment, and Wages in Rural Labor Markets in Asia, New Haven, CT: Yale University Press, 1984. Brass, Tom. “Class Struggle and Deproletarianisation of Agricultural Labor in Haryana (India).” The Journal o f Peasant Studies, 1989, 18(1), pp. 36-87. ___________ . “Unfree Labour and Agrarian Change: A Different View.” Economic and Political Weekly, April 1995, XXX (13), pp. 697-99. ___________ . “Yet More on Agrarian Change and Unfree Labour.” Economic and Political Weekly, January 1996, XXXI (4), pp. 237-40. ___________ . “Misinterpreting Unfree Labour in Contemporary Haryana.” Economic and Political Weekly, A ugust 1996, X X X I (33), pp. 2262-64. ___________ . Towards a Comparative Political Economy o f Unfree Labor, London, UK: Frank Cass, 1999. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 Braverman, Avishay and Stiglitz, Joseph. “Sharecropping and Interlinking of Agrarian Markets.” American Economic Review, September 1982, 72(4), pp.695- 715. Breman, Jan. Patronage and Exploitation: Changing Agrarian Relations in South Gujarat, India, Berkeley, CA: University of California Press, 1974. ___________ . Beyond Patronage and Exploitation: Changing Agrarian Relations in Gujarat, New Delhi, India: Oxford University Press, 1993. Caselli, Francesco. “Rural Labor and Credit Markets.” Journal o f Development Economics, December 1997, 54(2), pp. 235-60. Dasgupta, Biplab. The Naxalite Movement, New Delhi, India: Allied Publishers, 1974. Datt, Gaurav. Bargaining Power, Labor and Unemployment: An Analysis of Agricultural Labor Markets in India, New Delhi, India: Sage Publishers, 1996. Datta, Samar; Nugent, Jeffery; Tishler, Asher and Jone-Lin Wang. “Seasonality, Differential Access and Interlinking of Labor and Credit.” Journal o f Development Studies, April 1988, 24(3), pp. 379-93. Desai, Meghnad; Rudolph, Lloyd; Rudolph, Susan and Rudra, Ashok. eds., Agrarian Power, and Agricultural Productivity in South Asia, Oxford: Oxford University Press, 1984. Diamond, Peter. “A Model of Price Adjustment.” Journal o f Economic Theory, June 1971,3(2), pp. 156-68. Eswaran, Mukesh and Kotwal, Ashok. “A Theory of Two-Tier Labor Markets in Agrarian Economies.” American Economic Review, March 1985, 75(1), pp. 162-77. Foucault, Michael. Power and Knowledge: Essays and Interviews Gangopadhyay, Subhashish and Sengupta, Kunal. “Small Farmers, Moneylenders and Trading Activity.” Oxford Economic Papers, June 1987, 39(2), pp. 333-42. Gour, Raj Bahadur; Chaudhary, Chandragupta; Hyder, Gulam and Paranjpe, B.S. Glorious Telengana Armed Struggle, New Delhi: Communist Party of India Publications. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 Guha, Ashok. “Consumption, Efficiency and Surplus Labour.” Journal o f Development Economics, July 1989, 31(1), pp. 1-12. Gupta, Dipankar. ed. Social Stratification, New Delhi, India: Oxford University Press, 1991. Hardiman, David. Feeding the Baniya: Peasants and Usurers in Western India, Oxford, UK: Oxford University Press, 1996. Harsanyi, John. “Measurement of Social Power, Opportunity Costs and the theory of Two-Person Bargaining Games,” in John Harsanyi ed. Essays on ethics, social behavior, and scientific explanation, Dordrecht, Holland: D. Reidel Publishers, 1976. Janakirajan, S. “Village Resurveys: Issues and Results.” in Jan Breman; Peter Kloos and Ashwani Saith eds., The Village in Asia Revisited, New Delhi, India: Oxford University Press, 1997. Jodhka, S.S. “Agrarian Changes and Attached Labour: Emerging Pattern in Haryana Agriculture.” Economic and Political Weekly, September 1994, XXIX (39), pp. A102-A107. __________ . “Agrarian Changes, Unfreedom and Attached Labour.” Economic and Political Weekly, August 1995, XXX (30, 31), pp. 2011-13. __________ . “Interpreting Attached Labour in Contemporary Haryana.” Economic and Political Weekly, May 1996, XXXI (21), pp. 1286-87. Krishniah, M. “Regular Farm Labour System in Semi-Arid Areas: Emerging Patterns in Andhra Pradesh.” Economic and Political Weekly, March 1998, XXXIII (13), pp. A20-A28. Lacey, M arian. Workingfor Boroko: The Origins o f a Coercive Labour System in South Africa, Johannesburg: Ravan Press. Lalitha, K.;Kannabiram, Vasantha; Melkote, Rama; Maheshwari, Uma; Tharu, Susie and Shatrugna Veena. We Were Making History: Women and the Telangana Uprising, New Delhi: Kali for Women, 1989. Lanjouw, Peter and Stern, Nicholas, eds. Economic Development ofPalanpur Over Five Decades, Oxford, UK: Clarendon Press, 1998. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 Moore, Barrington Jr. Social Origins o f Dictatorship and Democracy: Lord and Peasant in the making o f the modern world, Boston, MA: Beacon Press, 1966. Motiram, Sripad. “Globalization and the Shape of Agrarian Conflict in Telangana,” paper presented at the 28th Annual Conference on South Asia, University of Wisconsin at Madison, October 1999. Mukherjee, Anindita. “Casual Labor.” in Lanjouw, Peter and Stern, Nicholas, eds. Economic Development ofPalanpur Over Five Decades, Oxford, UK: Clarendon Press, 1998. Mukherjee, Anindita and Ray, Debraj, “Labor Tying.” Journal o f Development Economics, August 1995, 47(2), 207-39. Osmani, S.R. “Wage Determination in Rural Labour Markets: The Theory of Implicit Co-operation.” Journal o f Development Economics, November 1990, 34(1,2), pp. 3-23. Pissarides, Christopher. Equilibrium Unemployment Theory, Cambridge MA: MIT Press, 2000. ________________________. “The Economics of Search,” forthcoming, Encyclopedia o f Social and Behavioral Sciences, Amsterdam, Netherlands: Elsevier Publishers. Ransom, Richard and Sutch, Roger. One Kind o f Freedom: The Economic Consequences o f Emancipation, Berkeley, CA: University of California Press, 1977. Rao, Mohan. “Agrarian Power and Unfree Labour.” Journal o f Peasant Studies, 26(2,3), pp. 242-62. Rao, Smriti. “Production, Reproduction and the Role of Women in Agrarian Change,” paper presented at the 28th Annual Conference on South Asia, University of Wisconsin at Madison, October 1999. Ray, Debraj and Sengupta, Kunal. “Interlinking and the Pattern of Competition,” in Pranab Bardhan, ed., The Economic Theory o f Agrarian Institutions, Oxford: Clarendon Press, 1989. Sarap, Kailas. “Transactions in Rural Credit Markets in Orissa, India.” Journal o f Peasant Studies, October 1987, 15(1), pp.83-107. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 _____________ . Interlinked Agrarian Markets in Rural India, New Delhi, India: Sage Publishers, 1991. _____________ . “Changing Contractual Arrangement in Agriculture Labour Market: Evidence from Orissa.” Economic and Political Weekly, December 1991, XXVI (52), pp. A167-A176. Schaffner, Julie A. “Attached Farm Labor, Limited Horizons and Servility.” Journal o f Development Economics, August 1995, 47(2), pp. 241-70. Scott, James, Weapons o f the Weak: Everyday forms o f Peasant Resistance, New Haven: Yale University Press, 1985. Shrivastava, Ravi, “Change and Resilience in Producer Strategies in Uttar Pradesh Agriculture.” in Jan Breman; Peter Kloos and Ashwani Saith eds., The Village in Asia Revisited, New Delhi, India: Oxford University Press, 1997. Smith, Adam. An Inquiry into the Nature and Causes o f Wealth o f Nations, London, UK: Penguin Classics, 1986. Srinivas, M.N. Caste in Modem India and Other Essays, Bombay, India: Asia Publishing House, 1962. Stiglitz, Joseph. “Rational Peasants, Efficient Institutions, and a Theory of Rural Organization: Methodological Remarks for Development Economics,” in Pranab Bardhan, ed., The Economic Theory o f Agrarian Institutions, Oxford: Clarendon Press, 1989. Sundarayya, P. Telangana people's struggle and its lessons, Calcutta: Communist Party of India (Marxist), 1972. Taylor, Mlichael, Community, Anarchy and Liberty, Cambridge UK: Cambridge University Press, 1982. W harton, C. “Marketing, Merchandising, and Money-Lending: A Note on Middle- Man Monopsony in M a la y a Malayan Economic Review, 1962, 7(1), pp. 24-44. Vakulabharanam, Vamsicharan. “The Dark Side of White Gold: Globalization and Agrarian Change in Telangana,” paper presented at the 28th Annual Conference on South Asia, University of Wisconsin at Madison, October 1999. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ____________________________. “Globalization and Agrarian Change in Telangana,” Dissertation, forthcoming, University of Massachusetts at Amherst, 2003. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 APPENDIX I QUESTIONNAIRE ADDRESSED TO THE ATTACHED LABORERS Socioeconomic Background Name: Age: Marital Status: Household Size: Number of dependents in the household: Education: Employer: Caste: Assets (1) Amount of land owned: Wet1 : Dry: (2) Amount of land leased in Wet: Dry: (3) Amount of land leased out Wet: Dry: (4) Do you own any of the assets below? (a) Tractor (b)Buffaloes (c)Bullocks (d)Bullock Cart(e)Vehicles (please specify) (f) Others (please specify) Labor (1) Did you work as a tied-laborer last year? (2) How many years have you been working as a tied-laborer? How many years have you been continuously working for your current employer? (3) What is (are) the main reason(s) for entering into a tied-labor contract? 1 Wet land refers to land that is irrigated by a source of irrigation - well, tank etc. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 (4) If the answer to the above is that you had no choice (because of debt or otherwise), what would you have done if you had a choice given the resources you have? Would you rather be a casual laborer? (5) Do you think that you could have earned more if you had worked as a casual laborer? (6) Explain the terms of your contract including - (a) Salary (b) Non-monetary items (c) Number of Holidays (d) Mode in which the payment is made (i.e. installments, amounts etc.) (e) Others (please specify) (7) What is the total income of your household this year from the following sources? (a) Salary (b) Agriculture (i.e. Land) (c) Dairy (d) Income from Casual Labor (e) Others (please explain) (8) In case you worked as a tied-laborer last year, what was the income from each of the above sources last year? (a) Salary (b) Agriculture (i.e. Land) (c) Dairy (d) Income from Casual Labor (e) Others (please explain) (9) Have you also worked as a casual laborer this year? If so, for how many days? (10) Do you supervise casual labor? (11) For the tasks given below indicate whether you do the task. Also indicate whether casual labor is involved. You Casual Labor Ploughing Sowing Applying Fertilizers Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 Spraying Pesticides Harvesting Taking the crop to the market Selling the crop in the market Others (please specify) (12) Do you think that you do any unpaid labor? Please explain (13) Have you ever defaulted2 on a tied-labor contract in the past? If so, please give details. (14) Is the possibility of default an important factor in your opting for a tied-labor contract? (15) If you default on a contract, do you think that you can work for the same employer again as a tied-laborer or as a casual labor? (16) If you default on a contract, do you think that you can work for a different employer as a tied-laborer or as a casual laborer? What about an employer in another village? (17) If you have a conflict with your employer, please explain to me how it is resolved (especially how it is related to your caste group and your employer’s caste group) (18) How do you address your employer? How does he address you? (19) Do you work for others at the behest of your employer? Please explain. (20) Are you planning to work as a tied-laborer next year3 ? If so, can you explain the terms and who the employer is? 2 It was clarified to the laborer that default meant that the laborer left the employer completely and that this could be for various reasons - conflict, sickness etc. 3 This question was asked during the current year, 2000-2001. Since I stayed beyond 2001Ugadi in Warangal, I ensured that I got the a c t u a l choice of the attached laborers in the next year. Also, as mentioned earlier, due to the b a y a n a system in Warangal, most attached laborers had committed themselves to work for particular employers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 APPENDIX II QUESTIONNAIRE ADDRESSED TO THE EMPLOYERS Socioeconomic Background Name: Age: Marital Status: Household Size: Number of dependents in the household: Education: Caste: Assets (1) Amount of land owned: Wet: Dry: (2) Amount of land leased in Wet: Dry: (3) Amount of land leased out Wet: Dry: (4) Do you own any of the assets below? (a) Tractor (b) Buffaloes (c) Bullocks (d) Bullock Cart (f)Vehicles (please specify) (f) Others (please specify) Labor (1) How many tied-laborers do you have? (2) What is (are) the main reason(s) for employing them (answer for each of them separately)? (3) Do you hire casual labor also? If so how many days? (4) Did you have a case of a tied-laborer defaulting on his contract? Do you think that the possibility of defaulting is an important consideration for tied-laborers when they enter the contract? Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 (5) If your answer to the above is yes, how did you deal with that situation? (6) In case your answer to (4) is no, how would you hypothetically deal with such a situation? Credit (1) What is your total outstanding credit? (2) Give the following details about the loans that you made to your tied-laborers (for each of them separately)? Tied Laborer Debt Outstanding(to you) Duration Loan this year Collateral Rate of of this debt Interest (3) If you have made loans to other people in the village, give the following details Debtor Debt Outstanding(to you) Duration Loan this year Rate of Interest of this debt (4) If there are differences between the terms of credit that you offer to the tied-laborers and others, what is the reason for this? Collective Bargaining: and Collusion (1) Who do you think has more bargaining power in your relationship with your tied- laborer. Why? (a) You (b) The tied-laborer (c) Equal (2) Do you think that employers collude in setting tied-labor contracts? If not, why not? (3) Do you think tied-laborers collude (either formally or informally) among themselves? If not, why not? (4) (Question addressed to the employers in Warangal, where there was an unsuccessful effort to organize a union of tied-laborers) Can you give me some details of the attempt to form a union and bargain for higher salaries?. In your opinion, why was it unsuccessful? What role did the employers (including you) play in this attempt? Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Dual labor markets, public debt management, and exchange rate movements

PDF

Buscando la prosperidad: Migration as long -term investment in El Salvador

PDF

Dependence on foreign labor, quality of education and unemployment in the GCC countries: In search of solutions

PDF

Insurance mechanisms, forest clearance, and the effect of government policies in rural economies

PDF

Essays on auctions

PDF

Health interventions, health status, and labor market outcomes in Indonesia: Three essays

PDF

Essays on consumption behavior, economic growth and public policy

PDF

Essays on regulation of public utilities and the provision of public goods

PDF

Income inequality and economic growth: A theoretical and empirical analysis

PDF

Essays in supply chain management

PDF

Demoeconomic dynamics: Evidence from historic Europe

PDF

Essays in environmental policy

PDF

A statistical analysis of the formation and location factors of high -tech centers in the United States, 1950--1997: An evaluation using quasi -experimental control group methods

PDF

Essays on social security

PDF

Critical beauties: Aesthetics, gender, and realism in John Ruskin and George Eliot

PDF

An investigation of consumption, insurance and village institutions in India

PDF

An enquiry into the effectiveness of export assistance organizations serving southern California SMEs

PDF

A study concerning the relevance of using quality measures of education in economic growth models of sub-Saharan African nations

PDF

Facing the camera: Dickens, photography, and the anxiety of representation

PDF

Endogenous entry equilibrium in auctions and markets with standards

Asset Metadata

Creator Motiram, Sripad (author)

Core Title Essays on agrarian institutions

School Graduate School

Degree Doctor of Philosophy

Degree Program Economics

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Economics, Agricultural,economics, labor,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Nugent, Jeffrey B. (committee chair), Robinson, James (committee chair), [illegible] (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-542027

Unique identifier UC11334915

Identifier 3094361.pdf (filename),usctheses-c16-542027 (legacy record id)

Legacy Identifier 3094361.pdf

Dmrecord 542027

Document Type Dissertation

Rights Motiram, Sripad

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA