Class, power, and agrarian change land and labour in rural west java

Class Power and Agrarian Change Land and Labour in Rural West Java Jonathan Pincus Management Information Expert Food and Agriculture Organization &... MARTIN'S PRESS, INC., Schola

STUDIES ON THE ECONOMIES OF EAST AND SOUTH-EAST ASIA General Editors: Peter Nolan, Lecturer in the Faculty of Economics and Politics, University of Cambridge, and Fellow and Director of Studies in Economics, Jesus College, Cambridge, England; and Malcolm Falkus, Professor of Economic History, University of New England, Armidale, New South Wales, Australia

In the last decades of the twentieth century the small and medium-sized nations of East and South-East Asia have begun a process of potentially enormous political and economic transformation Explosive growth has occurred already in many parts of the region, and the more slowly grow- ing countries are attempting to emulate this vanguard group The impact

of the region upon the world economy has increased rapidly and is likely

to continue to do so in the future

In order to understand better economic developments within this vast and diverse region, this series aims to publish books on both contemporary and historical issues It includes works both by Western scholars and by economists from countries within the region

Published titles include:

Melanie Beresford

NATIONAL UNIFICATION AND ECONOMIC DEVELOPMENT IN VIETNAM

John Butcher and Howard Dick (editors)

THE RISE AND FALL OF REVENUE FARMING

Mark Cleary and Shuang Yann Wong

OIL, DEVELOPMENT AND DIVERSIFICATION IN

BRUNEI DARUSSALAM

Yujiro Hayami and Toshihiko Kawagoe

THE AGRARIAN ORIGINS OF COMMERCE AND INDUSTRY Jomo K S

GROWTH AND STRUCTURAL CHANGE IN THE MALAYSIAN ECONOMY

Medhi Krongkaew (editor)

THAILAND'S INDUSTRIALIZATION AND ITS CONSEQUENCES Lee Sheng-Yi

MONEY AND FINANCE IN THE ECONOMIC DEVELOPMENT OF TAIWAN

Rajah Rasiah

FOREIGN CAPITAL AND INDUSTRIALIZATION IN MALAYSIA

Class Power and

Agrarian Change

Land and Labour in Rural West Java

Jonathan Pincus

Management Information Expert

Food and Agriculture Organization

&

First published in Great Britain 1996 by

MACMILLAN PRESS LTD

Houndmills, Basingstoke, Hampshire RG21 6XS

and London

Companies and representatives

throughout the world

A catalogue record for this book is available

from the British Library

ISBN 0-333-64578-2

First published in the United States of America 1996 by

ST MARTIN'S PRESS, INC.,

Scholarly and Reference Division,

Class power and agrarian change : land and labour in rural West

Java / Jonathan Pincus

p cm — (Studies in the economies of East and South-East

Asia)

Includes bibliographical references and index

ISBN 0-312-15827-0 (cloth)

I Land reform—Indonesia—Java 2 Social

classes—Indonesia Java 3 Power (Social sciences)—Indonesia—Java I Title

All rights reserved No reproduction, copy or transmission of

this publication may be made without written permission

No paragraph of this publication may be reproduced, copied or

transmitted save with written permission or in accordance with

the provisions of the Copyright, Designs and Patents Act 1988,

or under the terms of any licence permitting limited copying

issued by the Copyright Licensing Agency, 90 Tottenham Court

Road, London W1P9HE

Any person who does any unauthorised act in relation to this

publication may be liable to criminal prosecution and civil

claims for damages

10 9 8 7 6 5 4 3 2 1

05 04 03 02 01 00 99 98 97 96

Printed in Great Britain by

The Ipswich Book Company Ltd, Ipswich Suffolk

as

m

Contents

List of Tables vii List of Figures xi Preface xii

1 Introduction 1

2 Methodological Issues 19

3 Measuring Class Differentiation 37

4 Wage Labour Relations in Agriculture 93

5 The Process of Accumulation 147

6 Conclusion 188

Appendices

B Real Wage Trends in Agriculture 198

List of Tables

1.1 Producer subsidies in rice in selected Asian cou ntries 6

1.2 Distribution of operated farm sizes, Java 1963-1983 10

1.3 Average annual growth of wage labour force and 11

population of self employed farmers, Java 1980-1990

1.4 Sectoral distribution of increase in wage employment, 12

Java 1980-1990

1.5 Real daily wages for male hoers in some rice- 13

cultivating regions of Asia, 1987

2.1 Population density and agrarian density in North and 23

South Subang subdistricts, 1971 -1990

2.2 Components of the possessions score 28

2.3 Possessions score summary statistics 30

3.1 Female-headed households 43 3.2 Spearman correlation coefficients for possessions 49

scores and control over irrigated rice fields

3.3 Mean possession scores by area controlled 50

3.4 Correlations of labour hiring, family labour, posses- 53

sions scores and access to land

3.5 Correlations of labour hiring, family labour, posses- 54

sions scores and access to land, small and middle

farmers

3.6 Spearman correlation coefficients for agricultural 56

labour days hired out with possessions scores and

access to land, farming households

3.7 LOGIT analysis of participation in agricultural wage 57

labour among farming households

3.8 Mean values of selected indicators for landless 61

agricultural labour households

3.9 Income from nonagricultural wage labour by 69

land-controlling category

3.10 Migrants to domestic work in the Middle East 70

3.11 Mean earnings from trade and industry by 72

land-controlling groups

3.12 Distribution of households by class 79 3.13 Post-hoc comparisons of group means for small and 80

middle farmers, South and East Subang

4.1 Location of pre-harvest wage employment, agricul- 102

tural workers from North Subang, dry season 1990

4.2 Labour days worked locally by North Subang farm 103

labourers by wage system, dry season 1990

vn

viii List of Tables

4.3 Location of harvest employment, North Subang 105

4.4 Labour days worked locally by East Subang farm 109

labourers by wage system, dry season 1990

4.5 Harvest shares under bawon and ceblokan systems 111

by class, East Subang, dry season 1990

4.6 Crosstabulation of ceblokan arrangements by class 113

category of farmer and worker, East Subang

4.7 Correlation matrix of ceblokan earnings from large 114

farmers, East Subang, dry season 1990

4.8 Location of harvest employment, East Subang 115

4.9 Location of pre-harvest wage employment, East 116

Subang

4.10 Location of pre-harvest and harvest labour days, 118

South Subang

4.11 Labour days worked locally by South Subang farm 120

labourers by wage system, dry season 1990

4.12 Earnings per hour for one, two and three task 122

ceblokan arrangements, South Subang (combined

harvest and pre-harvest wages)

4A3 Correlation matrix of ceblokan earnings from large 123

farmers, South Subang, dry season 1990

4.14 Real wages for daily wage and contract labour tasks, 129

dry seasons 1971, 1978/79 and 1990, (Rupiah per

hour, constant 1971 prices)

4.15 Labour and Capital's share of rice production per 132

hectare for large farmers, constant 1971 prices

5.1 Mean output per hectare and capital costs in rice 150

production by class category, dry season 1990

5.2 Distribution of bank borrowing by class category (Rp 152

'000)

5.3 Mean output, returns and interest payments for 155

sharecroppers by class category, dry season 1990

5.4 Mean income per hectare from own-cultivation and 159

leasing by class category, dry season 1990 (Rp '000

List of Tables ix

Subang District, 1970-1989

5.10 School enrolment rates for children 7 to 18 years of 171

age, West Java, Java, Indonesia and study villages

1989

5.11 Participation in salaried employment by class 172

category

5.12 Status of Village Co-operative Units and Farmer 174

Credit (Kredit Usaha TanU Subang District

1989-1990

A 1 Weights used for Subang 9 commodity price index 195

A.2 Comparison of the Subang rural price index with 196

Subang rice prices and the Java-wide rural price

index

B.l Percentage Increase in Real Hoeing Wages From 199

Village Studies, 1970 to 1987

B.2 Selected indicators of changes in labour demand and 203

supply, Java 1980-1990 (average growth rates per

annum)

C.l Size distribution of paddy field ownership, area 204

operated and area controlled, dry season 1990, North

Subang

C.2 Size distribution of paddy field ownership, area 205

operated and area controlled, dry season 1990, South

Subang

C.3 Size distribution of paddy field ownership, area 206

operated and controlled, dry season 1990, East

Subang

C.4 Ownership of dry land fields and fishponds, South 207

Subang

C.5 Agrarian density, gini ratios for sawah ownership 208

and per cent landless in 17 Javanese villages

C.6 Post-hoc comparisons (Scheffe test) of possessions 209

score means for land-controlling groups

C.7 Categorical groups for logit analysis 210 C.8 Participation in nonagricultural wage labour 211

C.9 Mean wages and labour days worked, nonagricul- 212

tural and agricultural wage labour, dry season 19901

C 10 Trade and industry activities 213

C 11 Work participation rates by class 214 C.l2 Pre-harvest labour use by task, North Subang 1979 215

and 1990 dry seasons (person-hours per hectare)

C.l3 Pre-harvest labour use by task, East Subang 1990 216

x List of Tables

dry season (hours per hectare)

C14 Pre-harvest labour use by task, South Subang 1978 217 and 1990 dry seasons (hours per hectare)

C.l5 Crosstabulation of ceblokan arrangements by class 218 category of farmer and worker, South Subang

C.16 Educational attainment of rural population 10 years 219

of age and older, West Java, Java, Indonesia and

study villages

C.17 Educational attainment levels for residents ten years 220

of age and older

List of Figures

3.1 Agrarian Density and Land Ownership 46 3.2 Agrarian Density and Landlessness 47 3.3 Income Sources For Landless Agricultural Labourers 63

5.1 Flow of consumption loans within and between 178

classes following partial harvest failure, East

Subang

B.l Real Hoeing Wages in Java, 1976-1990 200 B.2 Real Tranplanting Wages in Java, 1976-1990 201

xi

Preface

This book is a revised version of my dissertation I would like to thank the editors of the series, Peter Nolan and Malcolm Falkus, for the chance to present my results and views to a wider audience

The research on which this book is based was funded by the Board of Graduate Studies of Cambridge University, the Cambridge Political Economy Society and Wolfson College Additional financial support for fieldwork in Indonesia was provided by the US Agency for International Development and the Ellen McArthur Fund I am grateful to Michael Hammig, then with USAID, Jakarta, for his constant encouragement and support Thanks are also due to the Pusat Penelitian Sosial Ekonomi Pertanian (PSE) for serving as my official host in Indonesia, and to Mohamad Saat of the Central Bureau of Statistics in Jakarta for his patience with my many queries This book could not have been written without the practical assistance and insightful comments of John Sender I have also learned much from comments given by Terry Byres, Peter Kenmore, Mushtaq Khan, David McKendrick, Peter Nolan, Gabriel Palma and Rizal Ramli at various stages of the book's preparation I alone, however, am responsible for all remairing errors of fact and interpretation

xn

The proposition that local patterns of agrarian change are closely bound up with class structure and class power runs counter to two common assumptions about Asian villages in general, and Javanese villages in particular The first relates to the relevance of social class to the analysis of agricultural development Even in the face of profound economic and social change, policy makers, scholars and development workers continue to cling to an outmoded view of the countryside as populated by homogeneous, small-scale producers farming on the basis of family labour and supported by 'traditional' village institutions.1 The historian Richard Hofstadter, noting a similar bias in American political discourse, used the term 'agrarian myth' to describe the resilience of these populist images despite their increasing irrelevance to rural economic structure Sadly, his observation that 'the agrarian myth came to be believed more widely and tenaciously as it became more fictional' applies equally well to Java (and to many other regions of Asia) as it does

to the US (1956, 30)

Failure to consider the issues of class and class power has also contributed to mainstream economists' preoccupation with demographic factors as the sole or main instigator of economic change at the village level An important example of this ten-

1

2 Class Power and Agrarian Change

dency is the neoclassical theory of 'induced innovation' This approach attributes changes in production relations and income distribution to movements of relative prices resulting from population growth (Hayami and Ruttan 1985; Ruttan 1978) In

their influential book Asian Village Economy at the Crossroads

(1981), Hayami and Kikuchi found support for induced innovation in their study of two villages in the Subang district of West Java Both of these villages have been resurveyed for this book The evidence collected ten years after Hayami and Kikuchi completed their survey indicates that the induced innovation hypothesis represents a vast oversimplification of agrarian dynamics in the study villages More specifically, the authors' focus on population pressure on land resources can only be sus-tained if it is assumed that the Subang villages are essentially closed systems, in which both labour and capital are immobile These two assumptions about agrarian change have proven particularly durable in the case of Java This is partly explained

by the island's extremely high population density and relatively

small average farm sizes.2 Social scientists have also played a part in perpetuating the agrarian myth in Java In his celebrated theory of 'agricultural involution', Clifford Geertz dismissed the relevance of class formation in the Javanese context in favour of his notion of 'shared poverty', in which villagers adopt increas-ingly elaborate tenancy and work-spreading arrangements in response to mounting population pressure.3 In answering his numerous critics (see White 1983; Alexander and Alexander 1982; Kano 1980), Geertz later claimed that the concept of shared poverty has been broadly misinterpreted, and that it was meant solely 'as a sensitising, heuristic concept designed to elucidate situations to which it could be applied to the degree that it could, and, contrastively, those to which is could not' (1984, 527) Whatever his original pedagogic intent, the notion of shared poverty has gained wide currency among Indonesian academics and policy makers, and continues to represent a substantial obstacle to the acceptance of a more realistic interpretation of agrarian change

These views still find support in the work of contemporary economists of a more technocratic bent Tabor, for example, concludes:

Although access to land is far from completely equal,

land-holdings remain far more evenly distributed in Indonesia

than in many other developing countries According to the

1983 Agricultural Census, only 4.4 per cent of Javanese and

Introduction 3

1.2 per cent of non-Javanese agricultural households are landless This, coupled with the relatively low rural to urban migration rates [sic] and the rise in agricultural employment levels, provides empirical support for the argument that green revolution advances have increased labour demand and improved returns in agriculture without marginalising the small farmers (1992, 167)

Yet, as Breman notes, 'The way in which such statistical material

is interpreted is strongly dependent not only on the reliability of methods for data collection but also on its style of presentation' (1983,127) Nowhere in the above quotation, nor in the following text, is there mention of the fact that the term 'agricultural households' as used in the census refers only to households in which at least one individual is directly engaged in crop production, fisheries or raising livestock This definition excludes

an estimated 40 per cent of rural Javanese households which neither own nor operate land (see White 1991, 57).4

The political ideology of the Suharto regime has also worked

to sustain the agrarian myth in Indonesia A central pillar of New Order agrarian policy has been the rigid control of political activity in rural areas In the years following the violent destruction of the Indonesian Communist Party in 1965-66, the government took immediate steps to remove any further threat from class-based movements in the countryside Under the so-called 'floating mass' policy all political activity is banned below the district level except for the brief (and closely monitored) campaigns prior to general elections Independent organisations

of farmers and farm workers are forbidden, and the regime has employed an effective strategy of bureaucratisation and militarisation of local government to restrain all forms of political initiative and resistance in the countryside (Husken and White 1989, 249-251) Into the ideological vacuum created by these policies, the government has injected a strong rhetorical

emphasis on populist themes such as 'mutual self-help' (gotong

royong) and the role of state-sponsored village co-operatives In this environment it is hardly surprising that discussion of class formation and rural class conflict is met with disbelief and in some cases hostility among Indonesians both within and outside

of government circles

Much of what is contained in the following pages represents

a challenge to the set of beliefs which constitute the Javanese agrarian myth Some readers, including some who consider themselves to be reasonably familiar with the patterns of Java-

4 Class Power and Agrarian Change

nese village life, will not recognise the image of the island presented here.5 However, it should be stressed from the outset that this book is not an attempt to introduce an alternative

uniform view of rural Java It is obvious that substantial variation exists between villages in terms of rural social structure, class relationships and dynamics of economic change Instead, the aim here is to demonstrate the centrality of class relations for a fuller understanding of these differences, and the locally-specific factors which give rise to them

Organisation of the Book

The book is organised as follows The remainder of this chapter provides a brief overview of the macro level evidence concerning trends in Javanese agrarian change Chapter 2 de-scribes the methodology used in the field studies, and addresses some common problems encountered by researchers studying agrarian change at the village level Two methodological issues of immediate relevance are discussed in detail These are the use of the 'village' as a unit of analysis, and problems associated with the use of resurvey evidence in studying processes of change The chapter also introduces the 'possessions score', a measure

of household well-being based on the ownership of durable goods

Chapter 3 presents an analysis of class structures in the three study villages It is argued that landholding is at best a partial measure of class differentiation, since the group of small farm households—which is often presented in the literature as

an economically homogeneous group-4n fact encompasses both extremely poor and well-off villagers This group is more accurately differentiated in terms of the use of hired labour as opposed to reliance on family labour for cultivation, and the participation of household members in the labour market as sellers of wage labour

Wage relations in agriculture are examined in Chapter 4 The proposition that production relations adjust to restore equil-ibrium to the labour market is rejected in favour of a less deterministic framework focusing on class structure and the relative bargaining power of classes at the village level

Chapter 5 examines strategies of capital accumulation ployed in the three villages These encompass own cultivation, land leasing, land acquisition, non-agricultural activities, salaried employment and 'political accumulation' pursued through

em-Introduction 5

positions in village government and co-operatives The ship between agrarian crisis and accumulation is analysed through an account of a partial harvest failure in one of the study villages in the 1989-1990 wet season

relatiai-THE MACRO SETTING: EVIDENCE OF AGRARIAN CHANGE IN

JAVA

Studies conducted in a range of Asian regions have sised several common themes in the development of post-colonial agriculture.6 These include the heightened role of the state in the agrarian economy, the commercialisation of peasant agriculture, accelerated processes of capital accumulation and proletarianisation and changes in relations of production, most notably the casualisation of the wage labour force and the rationalisation of labour use All of these trends are discernible

empha-in the Javanese context, although, as noted above, the quality and coverage of the macro-level evidence is uneven and hence often subject to misinterpretation

in the late 1960s to increase rice production Most prominent among these was the rehabilitation and expansion of irrigation systems in Java Between 1969 and 1980, extension or rehabili-tation of irrigation works covered more than half of all paddy field area in Java, and about 20 per cent of area outside Java (Booth 1988, 144) Related investments included the develop-ment of the domestic fertiliser industry, seed production facilities and a national network for transporting and storing rice and production inputs (Fox 1991, 61-63)

These infrastructural improvements facilitated the rapid adoption of new technologies based on modern rice varieties and more intensive use of inorganic fertilisers Another policy instrumental to the increase in fertiliser use was government intervention in markets for both fertilisers and rice From 1970

to 1989, domestic prices for urea and phosphatic fertilisers

6 Class Power and Agrarian Change

averaged approximately 40 to 50 per cent of world prices, a level

of subsidisation which effectively shifted part of the burden of risk associated with technological learning from farmers to the government (Tabor 1992, 180; Timmer 1985, 70) In the rice market, the State Logistics Board (Bulog) was assigned the task of defending official 'floor' prices for rice purchased from farmers, prices which during the 1980s were consistently above world market levels (World Bank 1992, 90).7 The combination of fertiliser subsidies and state intervention in the rice market resulted in a modest level of protection for farmers, amounting

to a strategy of import substitution in rice with the aim of generating employment in rural areas and saving scarce foreign exchange (Table l.l).8

Table 1.1: Producer subsidies in rice in selected Asian

Source: Gulati and Sharma 1992, Table 4b

However, not all components of Indonesia's rice tion programme have been equally successful The fixed input packages provided as part of the BIMAS ('Mass Guidance') rice intensification programme were often inappropriate to local agro-ecological conditions, and in some areas farmers were coerced by local officials into taking loans and inputs they did not want In response to mounting debt arrears, and the with-drawal of smaller farmers from the programme, BIMAS was discontinued in 1984 (Robinson and Snodgrass 1987) Until they

intensifica-Introduction 7

were abolished in 1988, subsidies on pesticides encouraged inchsaiminate use of broad spectrum insecticides which induced the highly disruptive brown planthopper outbreaks of the 1970s and mid-1980s (Kenmore et al., 1994; Fox 1991, 75) Finally, state-sponsored village co-operatives have not lived up to expectations in terms of the provision of farm inputs or as a marketing channel for rice output, and their administration has been prone to manipulation by village elites (Booth 1988, 256) Although rice intensification has received the most attention

in the literature, other aspects of state penetration into the agrarian economy have also played a large role in shaping patterns of change Not least among these has been the state's tight control over village pohtical life Selection of village heads is screened by district officials, and previously elected village councils have been replaced by 'village community security bodies' (LKMD) appointed by the village head The centralisation

of local power in the hands of village, sub-district and district heads, in tandem with the suppression of autonomous pohtical activity, has deprived the rural labouring classes of protection from illegal levies and other abuses of power on the part of village and supra-village elites The absence of social control on the behaviour of village elites has naturally widened the scope for various forms of pohtical accumulation The expansion of government employment, particularly during the 1970s and early 1980s, also fit neatly into the diversified accumulation strategies pursued by village elites Meanwhile, failure to implement the 1960 land reform laws removed the last legal threat to the power of large landowners, who have benefited hugely from the economic and pohtical patronage of the New Order state (Gunawan Wiradi 1984; Rajagukguk 1988, 64; Hart 1986a, 40ff)

Accumulation and Proletarianisation

Owing to the accelerated pace of change associated with the rice intensification effort, it is perhaps understandable that observers of the agrarian scene in the 1970s tended to interpret the increasingly commercial orientation of Javanese rice farmers, particularly the rationalisation of labour relations, as a direct effect of the adoption of new production technologies (see, for example, Collier et al 1978) Subsequent work, however, has placed greater emphasis on the historical roots of contemporary developments (Booth 1988; Hiisken and White 1989, 237).9

Although this renewed sensitivity to historical factors is to be

8 Class Power and Agrarian Change

welcomed, there is still little agreement as to the content and direction of change, particularly with reference to the structure

of agrarian society

Debate in the literature has centred around conflicting pretations of empirical evidence pertaining to the structure of landholdings, the degree of landlessness and growth of the wage labour force Booth, for example, argues that:

inter-A process of levelling down has been taking place in Java until the 1970s, whereby a growing population and agricul- tural labour force have been absorbed on limited land through a gradual reduction in the average holding size, with the dispersion of holdings remaining roughly the same (1988, 55)

This conclusion, however, is based on a comparison of data from

the colonial administration's 1903 welfare survey on land

ownership with data from the 1973 agricultural census on area

operated, and is therefore not justified (see Hasselman 1914, app.R) Moreover, this view is inconsistent with evidence pre- sented in large-scale colonial surveys as well as historical studies

of changes in specific regions of Java Although conditions varied markedly from region to region, these sources point to several salient features of Javanese agrarian structure during the late colonial period According to the aforementioned colonial welfare survey, at the turn of the century one-third of the agrarian population owned no land, while at the other extreme the nine per cent of landowners holding more than 1.4 hectares accounted for more than one third of all land (Hiisken and White

1989, 240)

This already pronounced degree of inequality increased over the ensuing years, particularly during the boom years following the First World War (Ibid., 242ff.; Elson 1984, 218ff.; Breman

1983, 74ff;) Rapid commercialisation and differentiation during the late 19 th and early 20 rh centuries was also accompanied by the decline of communal land tenure and its gradual replacement with individual ownership rights (Elson 1984, 219; Burger 1984, 90-91; Kroef 1984,155)

These trends were stalled with the onset of the Great pression of the 1930s, and did not regain momentum until after the Japanese occupation and the ensuing struggle for independ- ence (White 1989b, 69; Elson 1984, 233ff) Thus, rather than a gradual process of 'involution', the first half of this century appears to have been characterised by a period of nascent capitahst development, followed by stagnation and reversal

De-Introduction 9 during the difficult years of 1930s and 1940s Despite the disruptions of depression and war, however, pervasive inequali-ties in access to land as well as high levels of landlesaiess persisted, forming the social basis of the agrarian conflicts of the early 1960s (Lyon 1970, 20ff.; Ten Dam 1966, 349; Mortimer 1972; 34ff)

Attempts to trace changes in landholding patterns since the 1960s have focused on the three decennial agricultural censuses, the first of which was carried out in 1963 The censuses suggest

a pattern of holdings dominated by small farms, a structure which has remained remarkably stable over the 20 year period covered by the surveys (Table 1.2) Yet as Hiisken and White note, 'these data in fact serve mainly to highlight the limitations

of farm-size statistics in telling us anything about agrarian change, especially in contexts of population and productivity growth' (1989, 255) To begin with, as noted above, the agricul-tural censuses provide no information on land ownership or landlessness, and thus are of limited value in addressing questions of agrarian structure.10 Second, the reliability of the census data is open to question, as total land area recorded in these surveys differs widely from other sources, such as land tax

data (Booth and Sundrum 1976, 102) Moreover, land leased in greatly exceeds land leased out as recorded in the 1983 census,

an indication that landlords systematically underreport farm sizes Village resurveys, although still few in number, also point

to a process of land concentration which remains undetected at the macro-level For example, surveys carried out in nine lowland rice-producing villages in 1971 and 1981 reveal a dramatic increase in the share of rice fields controlled by farmers cultivating more than one hectare, accompanied by a sharp rise

in landlessness (White and Gunawan Wiradi 1989, Table 13.4).n

Most importantly, farm size statistics tell us nothing about

the degree to which and processes by which small farmers and

landless households have become increasingly dependent on the sale of wage labour for their subsistence As Karl Kautsky remarked nearly a century ago,

We now see that both tendencies toward centralisation and towards fragmentation can work alongside each other The

number of small farms whose owners appear on the market

for commodities as proletarians and sellers of labour-power is

increasing But their landholdings are only relevant outside the sphere of commodity-production, in the sphere of pro-duction for the household Once this stage is reached, any in-

10 Class Power and Agrarian Change

crease in small rural enterprises simply becomes one

particu-lar form in which the number of proletarian households

in-creases — a process which runs hand in hand with the

mul-tiplication of large-scale capitalist enterprises (1988, 179,

em-phasis in original).12

This aspect of proletarianisation is particularly relevant in Java where farms of less than one-half hectare predominate numeri-cally (although not in terms of land area) and where, as we have seen, landless households have historically accounted for a large share of the total rural population From this perspective, the process of proletarianisation in Java is not restricted to the separation of small farmers from their land, but also encom-passes the processes through which petty producers within and outside of agriculture are transformed into sellers of wage labour

Table 1.2: Distribution of operated farm sizes, Java

1 The 1963 census did not collect information on farms of less than 0.10 ha In

the 1973 census, the minimum sizes were 0.05 ha for wet rice fields and 0.10 ha

for dryland fields

2 Rice farms only

Sources: 1963 Agricultural Census, Final Report; CBS 1975; 1985

Table 1.3 compares the average annual growth in the wage labour force with growth of the total population and self-employed farmers for the period 1980 to 1990 Unfortunately, because of a change in the definition of 'employee' beginning with the 1980 population census, labour force data collected before this date are not comparable with post-1980 data.13 Data from the 1971 population census are also incompatible as a more stringent definition of employment was apphed in that

Introduction 11 year.14 As shown in the table, the census figures indicate that average annual growth of the wage labour force in Java (6.25 per cent) far exceeded the rate of growth of the working age population (2.43 per cent) during the 1980s Meanwhile, the number of self-employed farmers was nearly constant, growing

at only 0.65 per cent per annum Even these figures do not tell the full story, since they only cover the employment status of the

individual's primary occupation during the reference period15 as reported by the respondent They therefore prcvide no indication

of the extent to which small-scale producers, particularly small farmers, are dependent on the sale of wage labour for their subsistence This latter issue is considered in greater depth in the context of the three Subang villages in Chapter 3

Table 1.3: Average annual growth of total wage labour force and population of self-employed farmers, Java 1980-1990

0.94% -0.49% 0.65%

2.52% 2.35% 2.43%

As shown in Table 1.4, an important feature of the growth of wage employment during the 1980s was the large role played by the manufacturing sector, which accounted for 40 per cent of the total increment in wage employment, and nearly half of the increment among women As a share of total employment in Java, manufacturing increased from 11 to 15 per cent during the

12 Class Power and Agrarian Change

same period Construction has also been a particularly dynamic sector, accounting for 18 per cent of the increment in male wage employment At the same time, however, the capacity of agriculture to absorb new entrants into the labour force has declined, as opportunities for new investments in irrigation diminish and as mechanisation of rice agriculture begins to gather pace (World Bank 1992, 32; Naylor 1992) Rapid employ-ment growth in sectors outside of agriculture is closely related to the growing importance of migration, particularly circular migration, as noted in village studies (Hugo 1985; Manning 1988b, 61).16

Table 1.4: Sectoral distribution of increase in wage

1 Percentage of total increment in wage employment accounted for by each of the sectors listed

"' Includes utilities, financial and business services, real estate government

employment and other services

Source: CBS 1983, CBS 1992a

Despite the rapid growth of wage employment, wages in agriculture remain exceeding low by international standards This is demonstrated in Table 1.5, which presents figures for real wages (as measured in rice equivalents), rice yields per hectare and rural population density for a range of Asian rice-producing countries and regions for the year 1987 As shown in the table, daily wages for male hoers in Java ranked among the lowest in the sample despite the fact that Java recorded the highest aver-age paddy yields.17

1987

(A/B)

4.3 2.4 1.6 3.6 3.3

5.3 4.2 7.8 2.9 5.3 6.3 4.5 5.3 3.0

(C)

4.55 4.83 4.26 4.94 4.74

2.86 1.69 2.60 1.67 4.01 2.36 2.62 3.13 1.40

(D)

8.90 10.93 9.10 9.27 9.56

3.33 5.88 8.33 2.94 5.26 5.56 3.51 4.49 8.51 (A) Daily money wage, adjusted to reflect an eight-hour work day (B) Retail rice price per kilogram

(A/B) Rice equivalent of daily money wage

(C) Average paddy yield per hectare

(D) Population density 4

1 Agricultural wages in Java published by the Central Bureau of Statistics include meals Money wages were obtained by reducing the total wage by a factor of 35 per cent, an estimate based on the food component of hoeing wages recorded by the National Farmer Sample, PATANAS (calculated by the author using 1987 PATANAS wage data)

2 Wages refer to agricultural tasks for workers in rice and maize

3 Wages refer to all agricultural tasks

4 Population density for Javanese provinces and Indian states refers to rural population per hectare of arable land; for other countries this figure refers to members of agricultural households per hectare arable land

Sources: Money wages CBS 1991a; Krishnan 1991, A-83 (India); ESCAP

(Philippines); Central Bank of Sri Lanka 1988; Bangladesh Bureau of Statistics

1993 Rice prices CBS 1992b; Tyagi 1990,232 (India); Philippines, Sri Lanka and Bangladesh as above Paddy yields CBS Statistical Yearbook; Chandhok et al

1990 (India); FAO Production Yearbook (Philippines and Sri Lanka); Bangladesh as above Population density CBS 1991a Qava); Krishnaji et al 1991, A-67 (India); FAO Production Yearbook (Philippines, Bangladesh and Sri Lanka)

14 Class Power and Agrarian Change

While Java's high level of population density has no doubt contributed to this situation, the figures in Table 1.5 reveal that

no close correlation exists between population density and wage rates, indicating that other factors, such as the large share of landless households in the rural population and substantial inequality in landholdings, have also played a key role Of equal importance, although less frequently mentioned in the literature, has been state pohtical control over rural areas As Hart notes, the strengthening and militarisation of the local bureaucracy and the suppression of rural pohtical activity 'almost by definition operated so as to reheve the rural elite of the constraints to which they were subject during the late Sukarno period and simultaneously to circumscribe the bargaining power of the rural poor' (1986,190).18

While economists have engaged in a lively debate over

movements in Javanese real wages, the low absolute level of

agri-cultural wages has received less attention.19 This in part reflects the influence of a number of unabashedly sanguine poverty studies published in recent years which purport to show a massive decline in the incidence of rural poverty since the mid-

1970s Most notable among these is the World Bank's Poverty

Assessment and Strategy Report, which claimed that the incidence of poverty in rural Java declined by one third over a

period of only three years (World Bank 1990).20 The intent of this report was to demonstrate that macroeconomic adjustment measures taken following the decline in world oil prices in 1983 were consistent with improvements in living standards among the poorest Indonesians.21 While there is little doubt that the incidence of rural poverty has declined since the mid-1970s — primarily as a result of the increased availabihty of wage employment (Manning 1992, 33) — such overly optimistic estimates serve to distract attention from the consequences of restrictions placed on the bargaining power of rural wage workers

Recognition of the central role played by the relative gaining power of rural classes figures prominently in the analysis presented in the following chapters However, it is argued that class power at the village level is not simply a reflection of national or regional trends A central aim of this book is to explore the ways in which locally specific historical, pohtical and economic factors influence the bargaining power of the poor and

bar-in turn help to shape the organisation of production and patterns of change

Introduction 15

NOTES

1 As one prominent survey of the recent literature concludes, 'Asian agriculture is dominated by peasants who own small landed prop- erties and operate them on a family basis This peasant mode is augmented by the use of tenancy contracts that facilitate land transfers from relatively land-abundant households to households with little land so as to make the ratio of operational farmland to family labour homogeneous across farms' (Otsuka et al 1992, 1970)

2 According to the 1990 population census, 107.5 million people lived on Java, an island — at 132 thousand square kilometres — which is roughly the size of Greece, which has a population of 10 million (CBS 1992a)

3 Geertz writes, 'Under the pressure of increasing numbers and limited resources Javanese village society did not bifurcate, as did that of so many other "underdeveloped" nations, into a group of large landlords and a group of oppressed near-serfs Rather, it maintained a comparatively high degree of social and economic homogeneity by dividing the economic pie into a steadily increas- ing number of minute pieces, a process to which I have referred elsewhere as "shared poverty"' (Geertz 1963, 97)

4 The same error is made the World Bank in its recent study of poverty in Indonesia: 'Given that the typical poor household in rural areas already derives income from a variety of sources, the determinants of income (and hence the causes of poverty) are both varied and to some extent interrelated The most obvious single determinant of farm incomes is the amount of land the family cultivates In determining income, it is access to land, not neces- sarily land ownership, that is important (the 1983 agricultural census found that only 4 per cent of agricultural households culti- vated no land at all compared to around 20 per cent who owned

no land)' (World Bank 1990, 49)

5 An example is the widespread tendency of Indonesian ment organisations to focus exclusively on small farmers while more often than not failing even to detect the presence of large populations of landless workers in the regions in which they oper- ate The invisibility of the landless is both testimony to the power

non-govern-of the agrarian myth and the difficulties associated with designing development programs which address their needs Ironically, this focus on small farmers reproduces long-standing biases in gov- ernment rural policy originating with the Dutch colonial admini- stration (see Breman 1983,109)

6 See, for example, Schendel 1991, Breman and Mundle, eds 1991, Ramachandran 1990 and Byres 1981

16 Class Power and Agrarian Change

7. Bulog was established in 1967 and has sole authority to import and export rice The organisation is also responsible for distribut- ing rice to the armed forces and civil servants, and maintaining a national buffer stock to stabilise prices

8 Import substitution in rice was in part necessitated by the thinness

of the world rice market, in which trade accounts for only 3 to 4 per cent of world production (Gulati and Sharma 1992, A110) Given the size of Indonesia's annual rice requirements (26.1 mil- lion tons of milled rice in 1988), the country's imports are a major determinant of world rice prices (Timmer 1985, 55; World Bank

1992, 28)

9 As Manning concludes, 'the Green Revolution hardly marked a transition from subsistence to commercial farming for a peasantry long exposed to the vagaries of international and domestic mar- kets and with a long history of involvement in wage labour rela- tionships' (1988a, 78)

10 In an attempt to obtain an indirect measure of the increase in landlessness over this period, Hiisken and White compared popu- lation census figures from 1961, 1971 and 1980 with the 1963,

1973 and 1983 agricultural censuses They found that while in

1963 about 73 per cent of households operated a farm of more than 0.10 hectares, this figure had fallen to 57 percent by 1983 (Hiisken and White, 255) It should be noted, however, that this figures includes owner-lessors who do not cultivate land

11 As White comments in reference to the results of the 1973 tural census 'This finding [of stability in farm sizes] once again surprises those who have done field-research in Java and have re- turned with the strong conviction (even if they were unable to demonstrate it by accurate measurement) that a fairly rapid con- centration of holdings has been occurring, including rapid in- creases in absentee ownership by urban elites' (White 1979,102)

agricul-12 Compare this to Booth's caricature of the Marxist position: 'The Marxist argument postulates that the average holding size will increase over time regardless of the mounting population pressure, and that farm numbers will decline and a growing population will have to find employment as hired agricultural workers' (Booth

employ-Introduction 17

may include some persons whom later surveys classified as employed' (Korns 1987, 15) This definitional change undermines Booth's claim, based on a comparison of the 1976 and 1982 labour force surveys, that 'The recent evidence that the number of food- crop holdings in Java has been growing at the same time as the agricultural labour force has been virtually constant, or even de- clining, suggests that the number of households completely with- out access to land but dependent on agriculture for their livelihood

self-in Java may also have been declself-inself-ing' (1988, 50; emphasis added) Problems associated with seasonally in agriculture also limit the usefulness of comparisons between the 1976 and 1982 labour force surveys, as the former was carried out in the four month period from September to December of 1976, while the later was conducted in February, May, August and November of 1982

14 Employment in the 1971 population census was defined as work over a minimum period of two days over the reference week, as compared to one hour in subsequent surveys

15 The reference period for the two censuses was one week prior to the survey Both censuses were carried out in October

16 Macro-surveys have been less successful in enumerating migration patterns The population censuses, for example, exclude intra- provincial migrants, the total for which, according to Hugo, ex- ceeds other forms of migration by a factor of six (1985, 68)

17 Manning reached the same conclusion using 1980 wage data from Java, South Korea, Thailand, the Philippines, Sri Lanka and Bangla- desh (Manning 1988b, Table 3)

18 Indonesian industrial workers are also among the lowest paid in Asia: for example, a study comparing real wages in the textile and garment industries across 50 producing countries found that Indonesian daily and hourly wages were the lowest in the sample (Ahmad Istiqom Garinsia et al 1990, 26) According to the gov- ernment's own estimates, the statutory minimum wage amounted

to only 68 per cent of the 'minimum physical needs' of a single individual in 1991 (Manning 1992, 35) Even this extremely low standard is not actively enforced (Ahmad op cit., 24)

19 See Appendix B for a discussion of real wage trends

20 This result is based on data from the 1984 and 1987 National Socio-economic Survey (SUSENAS) using the official poverty line calculated by the Central Bureau of Statistics The methodology apphed in the study can be criticised on several grounds First, the credibility of the data is open to question As the World Bank itself reports, a reliability study carried out by the Central Bureau of Statistics found that expenditures on food among low income households was significantiy overestimated in the 1987 SUSENAS (Ibid., Annex I) In addition, itinerant individuals, a large proportion

18 Class Power and Agrarian Change

of which are poor, are underrepresented in the SUSENAS sample Second, the rural poverty lines adopted in the study were Rp 7,746 (US$7.55) and Rp 10,294 (US$6.26) per person per month for 1984 and 1987, respectively, levels which are sufficient to cover only basic caloric requirements and an extremely small increment for non-food items (these figures are country-wide averages of re- gional poverty lines) Aside from the problems involved in using standardised caloric cut-off points as a measure of poverty (see Sukhatme 1982), these cut-off points are substantially lower than those in other studies covering the same period (see Booth 1993)

21 The blatantly pohtical intent of the World Bank report is illustrated

by the fact that the incidence of rural poverty in 1987 reported in

an earlier draft of the text obtained by the author was more than

60 per cent higher than the figure included in the published sion It is also worth mentioning that this huge fall in the reported incidence of poverty coincided with an increase in the infant mor- tality rate in the second half of the 1980s (World Bank 1993b, 17)

ver-CHAPTER Two

Methodological Issues

INTRODUCTION THE FOLLOWING chapters present results from a survey of three villages in the Subang district of West Java This district was selected owing to the availabihty of baseline data from earlier surveys carried out in the 1960s and 1970s The intention was to resurvey the two villages for which baseline data exist in order to track changes in patterns of differentiation, accumula-tion and production relations A third village in the same district was surveyed for purposes of comparison A limited amount of background information was available for this additional site from a separate survey from the early 1980s

The decision to survey three locations within a small graphic area reflects two of the book's central hypotheses The first is that locally specific factors, such as the history of village formation, agro-ecological conditions and the balance of class power, play a major role in shaping patterns of agrarian change

geo-at the village level The second hypothesis holds thgeo-at these local factors influence agrarian outcomes primarily through their effect on the formation and development of inter-household relations at the village level

However obvious these premises may seem to readers iar with the day-to-day patterns of life in Asian villages, they are not easily reconciled with mainstream economic theories of agrarian change Economists have been inexplicably slow to relinquish their vision of rural households as independent decision-making units differing only in terms of 'initial endov-ments' and 'preferences' (Epstein 1975, 36; Appadurai 1989, 254) Microeconomic analysis at the village level has frequently overemphasised the 'choices' made by individuals while neglect-ing the power relationships which underlie and circumscribe these choices

fami-This chapter describes the methodology of the present study and discusses several more general problems confronting

19

20 Class Power and Agrarian Change

researchers studying the impact of structural and contextual factors on the path of agrarian change It consists of three parts The first describes the process of village selection and provides some basic information on the Subang district and study villages The mechanics of household sampling are taken up in the second section The chapter concludes with a discussion of two specific issues which arose during the course of the Subang survey: first, problems associated with the use of the village as a unit of analysis; and second, questions specific to the use of resurveys to analyse patterns of agrarian change

THE STUDY VILLAGES

Village Selection

The selection of study locations was governed primarily by the availabihty of suitable baseline data from previous village studies It was hoped that the resurvey strategy would provide important information concerning changes in economic struc-ture and production relations As discussed below, however, these expectations were only partially fulfilled

Three villages were selected from the Subang district of West Java Two of these villages, North and South Subang, were included in the series of studies carried out from 1968 to 1973

by the Agro-Economic Survey.1 The same villages were also studied in the late 1970s as part of the Rural Dynamics Study carried out by the same organisation, and by Yujiro Hayami and

Masao Kikuchi in preparation for their book Asian Village

Economy at the Crossroads (1981) A third village (East Subang) was also added Although no information for this location was available from the 1970s, this village was surveyed by a team of Japanese and Indonesian researchers in 1983-1984 as part of a comparative study of rice productivity in Thailand and Indonesia (Fujimoto andMatsuda 1985).2

The Subang District

Subang district is located on the northern coast of West Java about 150 kilometres east of Jakarta For most of the colonial period the area which is now Subang formed part of a private

estate (Pamanoekan en Tjiasemlanden) producing coffee, tea,

rubber and quinine Much of the district was only settled in the early years of this century with the arrival of migrants from the

Methodological Issues 21 eastern regions of West Java and from Central Java Administra-tion of the district was reclaimed by the colonial government in

1925

Agriculture remains the dominant sector, employing nearly half of the district's labour force in 1990.3 The northern half of the district consists of a flat, lowland plain lying between zero and 25 metres above sea level Irrigated rice monoculture has been the main agro-ecosystem of the northern plain since the completion of the Jatiluhur irrigation system in the late 1960s.4

Elevation rises steadily in the southern part of Subang, reaching a peak of 1,500 metres in the extreme south of the district Agricultural production is more varied here, and includes irrigated rice and other foodcrops as well as smalholder rubber, coconut, tea, coffee, cloves and fisheries Three state-owned plantations producing tea, rubber and sugarcane also operate in Subang

Two of the three study villages (North and East Subang) are located in the northern lowlands The third (South Subang) is situated near the centre of the district approximately 80 metres above sea level Irrigated wet rice is the predominant agro-ecosystem in all three locations

North Subang

The village of North Subang is situated approximately 15 kilometres south of the Java Sea in heart of the Subang's

northern lowland plain The kampung surveyed as part of this

study—one of the three hamlets which make up North Subang village—lies at the end of a dirt road of some three kilometres in

length linking the kampung to the district's main north-south

trunk road Access to the main road can be difficult, particularly

during the wet season, although becak (pedicabs) and ojeg

(motorcycle rides for hire) regularly ply this route.5 The hamlet is located about ten kilometres from a two-lane highway linking Jakarta in the west and Cirebon in the east

The kampung itself lies on a patch of high ground

sur-rounded on three sides by rice fields and bordered on the north

by a secondary irrigation canal A primary school is located at the entrance of the hamlet, and a newly renovated mosque stands some 20 metres from the canal Houses constructed of cement, stone and tile stand on relatively large plots in the centre of the hamlet These houses are occupied by the descendants of the hamlet's original settlers, many of whom are also the largest landholders These sturdy, and in some cases ornate, houses

2 2 Class Power and Agrarian Change

present a stark contrast to the bamboo huts—mostly occupied

by migrant labourers—packed tightly together along the

flood-prone fringes of the kampung The lack of available land to

expand the hamlet has also reduced the size of the wealthier residents' house plots as houses are built for successive genera- tions Unlike the other two hamlets of North Subang village, this

kampung is not yet electrified

North Subang is of fairly recent origin Older residents report that the first houses appeared in the 1920s as the surrounding land was opened up for cultivation The first irrigation system was constructed by the colonial government in 1935, although water was sufficient only for the wet season rice crop (Colter

1981, 3.17) By the late 1940s some fifty households hved in the

kampung, some relocating from nearby villages and others migrating from Cirebon and Indramayu The presence of migrant farm labourers, including workers from the north coast of Central Java, was already common at this time Some of these migrants settled permanently in North Subang and the surround- ing villages

This process of rapid population growth was brought to an abrupt halt in the early 1960s, when repeated crop failures forced many local residents into migrant farm work, particularly

in the Banten region Such was the scarcity of food at this time that a parcel of rice fields could be purchased for as little as 20 kilograms of unhusked rice Households holding surplus grain bought up land owned by departing migrants, resulting in the highly concentrated pattern of ownership which still character- ises North Subang By the mid-1960s only twenty or so house-

holds still hved in the kampung throughout the year, opening the

way for another influx of migrant labourers as conditions improved—particularly after the rehabihtation of the irrigation system in 1966

Double cropping of rice became possible for the first time with the completion of the Jatiluhur irrigation system in 1968 Total area double-cropped expanded during the early 1970s as tertiary canals were extended Increased demand for labour associated with double cropping encouraged further in- migration As shown in Table 2.1, the period 1971-1980 was characterised by rapid population growth, although the concomi- tant expansion of area under cultivation meant that population per hectare of rice fields actually declined slightly 6 Despite slower population growth during the 1980s, agrarian density

Methodological Issues 2 3

increased as sawah area remained virtually stable over the

decade

Table 2.1: Population density and agrarian density in North

and South Subang Subdistricts, 19711990

North/East Subang South Subang Subdistrict Subdistrict

Popula- Sawah Density Popula- Sawah Density

(A) (B) (A/B) (A) (B) (A/B)

1 Irrigated rice fields

Sources: 1971,1980 and 1990 Population censuses (Subang Statistical Office);

Subang District agricultural office

East Subang

The village of East Subang borders North Subang on the eastern side of the north-south trunk road A hardened stone and dirt road of less than two kilometres links the study

kampung with the main road, providing easy access to the market town and Jakarta-Cirebon highway

The study kampung—one of two hamlets in East Subang —is

surrounded by rice fields on three sides, and separated from its sister hamlet by a small stream and stretch of waste land Unlike

North Subang, the layout of the kampung does not follow an

obvious historical pattern Labourers' bamboo huts are partly interspersed with large, sturdy structures occupied by the hamlet's better-off households Another noticeable difference between East and North Subang is the larger land area available for dwellings Plots inhabited by poorer households are therefore

not as cramped as those in North Subang Although the

kam-pung is not yet electrified, cables strung to electrified houses in the neighbouring hamlet were already supplying power to several wealthier households in 1990

24 Class Power and Agrarian Change

The history of East Subang roughly parallels that of North Subang Settled in the second decade of this century, rice cultivation in East Subang was limited to one crop per year until the completion of the Jatiluhur system in 1969 As in North Subang, land was concentrated in the hands of a few wealthy

households during the 'maize era' (zaman jagung) of the early

1960s.7 This period also witnessed an outflow of farmers who travelled west in search of wage labour and land Unlike North Subang, however, population growth in East Subang following the onset of double-cropping has been slower, and primarily the result of natural increase and marriage (typically in-migration of husbands) rather than the settlement of migrant labourers In

1990 only six of the 135 households in the survey kampung did not contain at least one person born in East Subang village The web of social relations generated by this pattern of development helps explain the high incidence of sharecropping in East Subang

in comparison with North Subang

South Subang

South Subang village is situated some 20 kilometres from North and East Subang, and approximately five kilometres from the district capital The village centre is connected to Subang City

by a paved road, although the study kampung itself is located at

the end of a short dirt track Both the village centre and local

markets are within easy reach of the kampung throughout the

year Minibuses run regularly along the Subang City-South Subang road

South Subang was settled earlier than the other study lages, probably during the last decade of the 19h century A large majority of current residents can trace their ancestry back to these original settlers Although hiring of migrant labourers was common during the 1960s and 1970s, few of these workers settled permanently in the area The hamlet has slowly expanded northwards from the original settlement, with the newest houses

vil-built along the hills at the extreme north of the kampung A

striking feature of the hamlet, however, is the absence of sharp differences in the quality of houses, with the notable exception

of a few modern structures built by local ehtes in recent years At the time of the survey South Subang was not yet electrified.8

Double-cropping came to the village with the construction of the Leuwinangka system in 1928.9 The early arrival of double-cropping partly explains the large number of micro-holdings in

the kampung, as the greater income-generating capacity of the

Methodological Issues 25

land has encouraged repeated partitions of family holdings The Leuwinangka system has undergone six rounds of rehabihtation since 1977, although the area covered has not expanded significantly As shown in Table 2.1, population density in this subdistrict far exceeds that of the North/East Subang subdistrict, and has increased at a more rapid rate over the past two decades It is important to note, however, that these figures actually understate the difference in agrarian density in the two subdistricts, since rice yields in North and East Subang are consistently higher than those in South Subang

THE HOUSEHOLD SURVEY

Sampling Procedure

The village (desa) in Indonesia is essentially an administrative unit, usually containing several hamlets (kampung or dusun) and

up to several thousand households In each of the villages

selected, one kampung was chosen for study based on the

location of the baseline surveys

The first stage of the fieldwork consisted of a complete

cen-sus of households in each kampung in order to determine the

size of the population and to gain a rough idea of the economic structure of each hamlet The results of the census indicated that populations in two of the three locations (South and East Subang) were small enough to permit complete coverage of all house-holds within the geographical boundaries of the hamlet In North Subang, however, the study hamlet contained more than 200 households It was decided that it would not be feasible to

survey every household in the North Subang kampung, and a

random sample of 100 households was taken The sample was not stratified in order to include a proportionate number of farmers and labourers in the survey The use of a random sample of North Subang households did not affect comparisons with earlier surveys of the same location, as these studies had also found it impossible to conduct a full enumeration of the

kampung

The use of this sampling method in North Subang served to illustrate some of the problems involved in working with random samples when analysing processes of rural change After beginning the household survey in North Subang it was discov-

ered that the largest farmer in the kampung—dn individual

owning 60 hectares of irrigated rice fields—was not included in

26 Class Power and Agrarian Change

the sample This farmer had (predictably) underreported his

control over land during the kampung census, and the true

extent of the household's holdings only became apparent as interviews with other local households revealed the scope of this farmer's activities in the village.10 The household was therefore added to the sample during the course of the survey, as it was felt that the gain in terms of coverage was more important than the loss of 'randomness' in the sample This incident points to the main weakness of random sampling in the context of village studies: since this method assumes an approximately normal distribution, random samples in economically polarised localities are likely to result in partial coverage and hence misleading survey results (Casely and Lury 1987, 67)

Another problem with random sampling which became parent during the North Subang survey is that it tends to frustrate attempts to unravel the complex web of social and economic relations between households Such relationships are frequently multidimensional in nature, encompassing employ-ment, land leasing, credit, pohtical patronage, kinship and so forth The practice of interviewing only one of the parties involved in these dyadic linkages carries with it the danger that the information collected will be incomplete or erroneous Landlords, for example, may be willing to report participation in sharecropping arrangements, but may be less forthcoming about the interest charged on loans to tenants For this reason, it is generally necessary to check responses given by both parties, a process which often requires repeated visits Thus, random sampling in North Subang did not in the end save as much time

ap-as wap-as originally anticipated, since interviews were carried out with a large number of households not selected as part of the sample as a means of verifying data

The Household Survey

Following the completion of the kampung census, a

ques-tionnaire was drawn up, tested in pilot interviews and corrected During the first round of the survey two research assistants helped carry out interviews Upon collating the results of these interviews, however, it soon became apparent that certain questions asked during the interviews were eliciting formulaic responses In addition, cross-checks of answers involving inter-household relationships showed that many responses were inconsistent: employers and employees failed to agree on

Methodological Issues 27

duration of work and wages paid; lessors and lessees gave differing responses concerning contract terms, and so forth

Three corrective measures were taken to improve the quality

of the data First, it was decided that all further interviews would

be carried out by the author Second, questionnaires were not used after the completion of the first round of interviews, since

it was apparent that the presence of the questionnaire itself was

a major obstacle to frank and informal discussion Instead, interviews were more loosely structured around a hst of ques-tions compiled for each household Because of the sheer quantity

of information involved, it was unavoidable that notes be taken during the interviews: however, there was in fact a noticeable difference in the tone of discussions after the questionnaires were no longer used Third, a hst of conflicting information pertaining to inter-household relationships was drawn up, and the households involved were revisited separately until these conflicts were resolved

The Possessions Score

Many researchers have remarked upon the difficulties volved in collecting information on household income and wealth Unlike responses pertaining to inter-household relaticn-ships, this information cannot be verified through cross-checks, and therefore the researcher must rely on the memory and frankness of the respondents themselves However, since written records are rarely kept, it is highly impractical to expect respoi-dents to recall income data with any amount of precision over a period of more than a week or two Systematic underreporting is also common, particularly among better-off households Moreover, income surveys in developing rural areas raise a number of thorny issues relating to the definition of household 'income' (Casely and Lury, 167) Is agricultural production consumed within the household, for example, to be considered

in-as part of income, and if so how is it to be valued? Should earnings from petty trade and handicraft production be counted

as net or gross? Is it possible to separate the cost of inputs (particularly credit) from the broader household budget?

Another problem with income surveys is that earnings by rural households are often intermittent and therefore vary substantially from month to month and from season to season 'Snapshots' of one reference period may therefore reflect transitory earnings, thus resulting in a distorted estimate of the