the same village cluster.18 Robert Wade studies collective action in 31 ... Nadu, Robert Chambers and John Harriss typologize villages based on irrigation and.
(published in India Review (2015), 14(4), 399–418)
What Do We Know About Village-Level Change and Stasis in India?
The study documents considerable spatial variation in change and stasis in development outcomes over the decade 2001-11 (proxied by women’s literacy and child sex ratio) even across villages within the same micro-region (taluk or sub-taluk) and with similar starting points. However, neither decentralization policy / practice nor other forms of public policy has identified village-level factors that mediate the impact of policy. Although extant literature has explored spatial variation, it has not explored such variation across different micro-regions of India, nor has it used methodologies that validate explanatory inference from spatial-longitudinal comparisons. The paper notes that the degree of spatial variation in change over such a short period of time is remarkably similar across different micro-regions of the country. It also proposes a tentative methodology for identifying village pairs to produce more rigorous comparative longitudinal analysis of the drivers of development change and stasis. Keywords: literacy; sex ratio; village studies; spatial variation; methodology
On November 7, 2014, leading his government’s initiative encouraging legislators to “adopt” villages for targeted development, Prime Minister Narendra Modi adopted Jayapur village in Varanasi taluk of his Lok Sabha constituency.1 Whatever the motivations of this policy from a political economy perspective, and whatever its merits and demerits from a policy perspective, such high-profile actions – as well as past attempts at promoting “model villages”2 – produce a focus, once again, on village-specific development. This micro-level focus complements recent moves, both in scholarly and policy-making circles, to disaggregate India’s regions and focus on pockets of backwardness. A recent study by Sanchita Bakshi, Arunish Chawla and Mihir Shah notes: ...an emerging characteristic of regional disparities in India is the presence of underdeveloped regions even within higher income states...what is even more remarkable, within relatively developed districts, we also find pockets of intense backwardness in some of their sub-districts. And conversely, some backward districts can have some of the most developed sub-districts. In fact we find many districts which include the most backward and most developed sub-districts of India.3 This study takes the argument to the next level of disaggregation: the village. Not only is there considerable variation in development outcomes across the districts of a state and across the subdistricts of a district,4 but there is also considerable variation across villages of a subdistrict – and indeed, even across villages of a sub-subdistrict. Further, one of the underappreciated aspects of village-level development – and the focus of this study – is the remarkable variation in village-level development even over a short period of time across villages in the same subdistrict. For instance, census data reveal that the women’s literacy rate in the Prime Minister’s adopted village Jayapur increased from 46% in 2001 to 54% in 2011 – close to a percentage point a year on average.5 But just ten kilometers away is the village of Ranibazar, which had a similar women’s literacy rate in 2001 (48%) but
nevertheless ended at 68% in 2011 – an increase that is double that of Jayapur. And only a little farther away is the village of Sikhari, which also started with a similar women’s literacy rate in 2001 (47%) but where it decreased over the next decade (to 45%). As these examples suggest, changes in development indicators exhibit a remarkable degree of spatial variation, even for villages that are relatively close to each other and have similar starting levels. Jayapur, Ranibazar, and Sikhari are all part of Varanasi taluk, and all had approximately the same women’s literacy rate in 2001, and yet within a decade they exhibited considerable diversity in a key development outcome, women’s literacy. The explanation of micro-level change over time is an important, vexing, and under-researched area of research. Particularly at the micro-level – villages in the same small geographical area, where many other conflating factors are likely to be similar – such large variations over relatively short periods of time afford an excellent opportunity to compare the specific local drivers of change (and stasis). As Donald Campbell and Laurence Ross noted many decades ago, “[w]hile the social scientist cannot as a rule experiment on a societal scale, societal “experimentation” or abrupt focused social change is continually going on, initiated by government, business, natural forces, etc”.6 This study systematically illustrates the extent of micro-level variation in the recent development trajectories of India’s villages. It shows that what is true for the JayapurRanibazar-Sikhari comparison in 2001-11 is also true for other village clusters: according to census data, even within a taluk and over only one decade, there are instances of villages with considerable increase in literacy (or sex ratio) coinciding with instances of villages with decrease in literacy (or sex ratio). The first objective of this study is to flag the important of this: such seemingly radically different short-term development trajectories even within village clusters produce a profound, under-researched puzzle. Second, the study shows that what is true for Varanasi appears to be also true for other parts of the country. To establish
this, besides Varanasi taluk of Varanasi district (Uttar Pradesh), the following taluks were chosen: Kherwara taluk of Udaipur district (Rajasthan), Jhalod taluk of Dahod district (Gujarat), and Shorapur taluk of Yadgir district (Karnataka).7 The broad pattern of intra-taluk village-level variation in change in literacy and sex ratio in 2001-11 is consistent across these very different taluks from different geographical and historical regions of India. This preliminary finding is hardly intuitive, and suggests the need for greater research on whether larger forces shape micro-level variations in development changes. Third, the study argues that the study of contrasting development trajectories of comparable villages is critical for calibrating policies for micro-level development, which is an unstated requirement for the success of present strategies such as that of village adoption. The remainder of this study is organized as follows. The following section provides a brief overview of the literature on Indian villages from a comparative and longitudinal perspective. The next section uses census data for 2001 and 2011 to document the considerable variation in growth / stagnation of women’s literacy across villages in different micro-regions of the country, with a further focus on Shorapur taluk. This is followed by exploration of a different development outcome, child sex ratio, where again similar patterns of variation are observed. Based on these descriptive findings, the study then engages with recent work on subdistrictlevel variation and the importance of parsing out factors at the village-level and the microregion-level. The concluding section discusses broad implications for scholarship on the drivers of development change as well as for contextualized public policy.
Related Literature Departing from the “book view” of an earlier generation of studies, post-independence explorations of Indian villages have taken the “field view”.8 An early, systematic effort occurred through the Village Level Studies program in university Agro-Economic Research Centers in the 1950s and ‘60s.9 The Village Studies Programme at the Institute of Development Studies (IDS), Sussex in the 1970s systematically studied village-level development across world regions, including India.10 Focusing on one state (Tamil Nadu) alone, K. Nagaraj and R. Rukmani identify 678 village studies, most of them PhD and MPhil theses.11 However, as Barbara Harriss-White and John Harris note, this empirical social anthropological tradition became “unfashionable” in later decades,12 and the emphasis shifted to discourse / constructivism as well as theorization that glosses over contextual and spacetime specificities.13 Many empirical village studies note the existence of considerable micro-level diversity in development trajectories.14 Writes Robert Wade: “The fact is that rural societies of the nonwestern world are marked by greatly varying features and tendencies, both in their internal ecology and culture, and in their connections with markets, state structures and other external influences before and during western penetration... Only a few miles may separate a village with a great deal of public organization from one with very little.”15 Studying two Bihar villages over decades, Amrita Datta and co-authors note: “each village has its own pattern of connections with the process of development, because of its location, resources, economic structure, social composition and relations, and history. The overall pattern of change in rural Bihar is then a composite of many different village development paths, in which the average may hide a great deal of variation in social and economic patterns and trends.”16 A large literature explores spatial differences in development trajectories of macro-regions in India, most commonly inter-state comparisons but also comparisons of sub-state regions.17
The literature comparing micro-regions in close proximity, including villages in the same cluster – which, arguably, can yield more precise accounts of the factors driving change given the in-built control over a range of factors in such research designs – is smaller. The challenge is to construct research designs that take both time and micro-level space seriously. The remainder of this section describes some of the work in the comparative village study tradition, the longitudinal village study tradition, and the still smaller literature on village studies that are both comparative and longitudinal, although the attempt here is not to provide a comprehensive survey of such studies. Based on this, the section ends with a methodological critique of extant literature, which the methodology outlined in this paper seeks to address. Comparative Village Studies Several studies in the Indian literature focus on spatial variation across villages, many within the same village cluster.18 Robert Wade studies collective action in 31 irrigated and 10 dry villages of Kurnool district of Andhra Pradesh using a research design that explicitly fleshes out micro-level spatial variation.19 He identifies scarcity and risk as key determinants of collective action in these villages, and explores how these are also shaped by social structure, demography, extra-village economic links, and the role of the state. David Mosse uses a broadly similar approach for Ramnad in Tamil Nadu.20 He finds that not only does collective action depend upon livelihoods determined by ecology, but cultural and political factors also mediate this relationship. Studying twelve villages of the old North Arcot district of Tamil Nadu, Robert Chambers and John Harriss typologize villages based on irrigation and associated labor needs. They provide a tentative explanation of village variation in economic activity based on location, resources and population, and the nature of the production process.21 Grace Carswell studies two villages near Tiruppur in Tamil Nadu and notes that Mathari (dalit) households in one village have less access to garment jobs in Tiruppur than
Matharis in the other, and this can be traced to specific path dependencies.22 Studying 16 villages in Bhilwara, Rajsamand, and Udaipur districts of Rajasthan, Anirudh Krishna establishes considerable variation, for instance in land disputes (ranging from 2.79 land dispute cases per hundred villagers in one village to only 0.25 in another).23 He argues that differences in development are due to the levels of social capital along with the presence of capable and effective village agents. Studying 105 villages in Udaipur, Kota, Ajmer, and Jodhpur districts of Rajasthan, Gabrielle Kruks-Wisner notes variation in the incidence and practice of “claim-making” (petitioning and lobbying) on the state, and attributes this to the degree to which an individual has contacts and exposure across space and social groups.24 Micro-level variations in development occur not only across villages within the same region, but also, for instance, across slums within the same city. Adam Auerbach argues that variations in public services and infrastructure across slums in a city are due to differences in party network density, since such density determines the extent of clientelist linkages to public services and infrastructure.25 Longitudinal Village Studies There has also been a handful of studies of village-level change in India.26 Studying selected Thanjavur villages over four decades, Kathleen Gough explores the changing face of the Indian state and development policies from the ground up, and how social stratification inflects state policies.27 Unsurprisingly, longitudinal village studies suggest that the overwhelming dominance of traditional elites has declined in many instances. For instance, for his study village of Sripuram in Thanjavur, Andre Beteille shows that while a small Brahmin group once dominated along the dimensions of caste, class, and political power, over the decades this dominance declined in different ways across these different dimensions.28 Other studies complement this narrative – for instance, work by Anirudh Krishna and Oliver Mendelsohn.29 However, countering this evidence of weakening local
caste dominance over the decades, other village studies suggest that its sources in cultural differentiation enable it to continue to produce class differentiation.30 Importantly, there are substantial spatial variations in the extent and nature of decline in local caste dominance. John Harriss notes: “There are huge difficulties with generalizations about the decline of dominance based on land control. There is enormous variation from region to region and from locality to locality even within one region.”31 The literature on village restudies has been an important avenue to construct longitudinal narratives of stability and change at the micro-level.32 The series of North Arcot studies is an important example.33 In one instance, John Harriss, J. Jeyaranjan, and K. Nagaraj further restudy the villages studied a century ago by Gilbert Slater and later restudied by S. Guhan.34 By layering contemporary fieldwork over previous findings, they uncover a narrative of continuity rather than change in landlord power. Nevertheless, there was some limited empowerment of dalits through political mobilization: “We observed the pride of dalit women, especially in their girl children’s school uniforms and their school books. This, according to some women with whom we spoke, was a big change of the last 25 years” (p. 50). Focusing on Palanpur village in Uttar Pradesh, Peter Lanjouw and Nicholas Stern note the small uptick in women’s literacy relative to men’s literacy in 1957-1993, although the increase in both was poor in absolute terms.35 They locate the explanation in poor parental motivation, poor schooling system, and low priority of education in local politics. The ICRISAT longitudinal village studies from the semi-arid south-central region also fall into this category – for instance, work by Thomas Walker and James Ryan, Reena Badiani and co-authors, and Uttam Deb and co-authors.36 The utilization of village-level census data across decades to explore longitudinal change – the route taken in this study – has also been suggested previously, for instance by Christophe Guilmoto for irrigated lands and K. Nagaraj for agrarian structure.37
Comparative and Longitudinal Village Studies The literature on micro-level studies that are both seriously comparative and longitudinal is slim. In the case of the North Arcot Studies, John Harriss explains inter-village longitudinal change (over the decade of the ‘70s) based on degree of irrigation/cultivation, labor and nonfarm employment opportunities, and village class structures.38 Jan Breman traces the dynamics of rural transformation in four villages in South Gujarat, with the underclass on the one side and the landed upper castes on the other.39 Scarlett Epstein, A. P. Suryanarayana, and T. Thimmegowda explore the trajectories of a wet and dry village in Mandya district over four decades, finding that the former had greater “introversion” and the latter “extroversion” in social relations.40 In their comparative study of two Bihar villages, Amrita Datta and coauthors find greater improvement in girls’ education and in gender relations in one village (Mahisham) than in another (Chandkura) – resulting in a closing of the gender gap in literacy in Mahisham but not Chankura in 1981-2009 – and they trace these outcome differences to differences in male out-migration.41 The small literature on comparative longitudinal village studies suggests the presence of considerable local-level variation in village development trajectories. However, it is also possible that studies in this tradition chose villages which do indeed show such variation. This possibility, along with the fact that this literature is small, suggests the need for more systematic exploration of village development trajectories in different micro-regions of the country. Later sections of this study attempt to do so. Methodological Limitations of Extant Literature The literature on village studies, selectively surveyed here, exhibits two important methodological characteristics. The first is that the methodology of village choice typically does not hew to the principles of comparative research designs that yield explanatory
inference. The second is that longitudinal process-tracing is typically absent. As I elaborate below, both characteristics compromise the potential for empirical inference and explanation. The two points are made using two examples (the work of Robert Chambers and John Harriss and the work of Anirudh Krishna) showcasing the older and newer traditions of village studies.42 The comparative approach to explanation can be traced to John Stuart Mill and his “method of difference.”43 In this approach, units chosen for comparison are similar except for a specific outcome and variable of interest, so that variation in outcome can be potentially causally traced to variation in the other variable.44 Consider the work of Robert Chambers and John Harriss in this light.45 They compare the villages of the first round of systematic North Arcot village studies from 1973. The intention of the North Arcot studies was to compare rice regions with similar agro-climatic conditions in India and Sri Lanka. To this end, 11 villages were selected through (stratified) random sampling.46 This implies that together the villages were expected to be representative of the region, but they were not expected to be similar to each other because of the random selection. This effectively precludes application of the comparative approach which seeks to control for a large swathe of common factors across villages. Village selection through random sampling – and more generally, when villages are not purposively selected for starting similarity – makes comparisons of those villages fundamentally different from the comparative framework outlined in this paper. The study by Chambers and Harriss is nevertheless useful, of course, in the sense of case juxtaposition and what Theda Skocpol and Margaret Somers call “contrast of contexts”, but the approach to sample selection considerably compromises on explanatory power.47 A similar issue holds for the work of Anirudh Krishna, who has a chapter entitled “Understanding Economic Development: Why Do Some Villages Develop Faster than
Others?”48 To address this question, Krishna categorizes villages into those with contemporaneous “high” and “low” performance on a development index. Factors producing these different outcomes can be isolated using the above-described comparative approach only if the villages were sufficiently similar on a large set of variables at some point in time prior to that of the fieldwork. Krishna explicitly acknowledges this: “I had to select villages that had started off at a roughly similar level…” (p. 118). However, in fact villages were chosen through stratified random sampling,49 compromising explanatory power through sampling/research design.50 In later sections, this paper tries to directly address the challenge of explanatory inference from spatial-longitudinal comparison by suggesting a methodology of purposively identifying villages with similar starting points but strikingly varying trajectories leading to different end points within a relatively short period of time. The second methodological challenge is in taking the longitudinal aspect seriously. A rigorous explanatory comparative-longitudinal study would need to trace the processes through which villages that were similar at one point in time end up on different time trajectories with substantially different end points. The time element is explicit in Krishna’s chapter subtitle. However, the empirical analysis uses comparative statics rather than longitudinal process-tracing. Establishing that there is contemporaneous conditional correlation between variables (development level, social capital, and local agency capacity) is insufficient to explain “why some villages develop faster than others.” What is required is some form of process-tracing – what John Gerring describes as “contextual evidence and deductive logic to reconstruct causality … [through] long causal chains” (p. 172-3).51 In the case of the North Arcot village comparisons by Chambers and Harriss, and even more in the second set of comparisons by John Harriss after the second round,52 there is indeed a limited degree of process-tracing, although it can be critiqued for being insufficiently detailed.53
Variations in Village Development Trajectories: The Case of Women’s Literacy In order to establish that village-level development trajectories vary widely even within the same micro-region, I focus on two development outcomes: women’s literacy and child sex ratio. Four micro-regions from different parts of the country are considered: Varanasi taluk (Varanasi district, Uttar Pradesh), Kherwara taluk (Udaipur district, Rajasthan), Jhalod taluk (Dahod district, Gujarat), and Shorapur taluk (Yadgir district, Karnataka). The use of census data ensures that the treatment of each selected micro-region is comprehensive. Literacy Trajectories in Four Different Micro-Regions Consider first the case of women’s literacy. Village-level women’s literacy rates in 2001 and 2011 are presented in Figure 1. The sub-figures plot, for each village with over 250 population in 2001 and 2011, the literacy rate in 2011 against the literacy rate in 2001.54 The solid line is the line of equality: villages above it experienced an increase in literacy rate in that decade and villages below it experienced a decrease. As Figure 1 suggests, taking a cross-section of villages in any taluk (Varanasi, Kherwara, Dahod, or Shorapur) in any year (2001 or 2011), there is considerable variation in women’s literacy rates. This is in itself interesting and worthy of careful investigation, although it is hardly a new finding. But what is surprising is that villages also vary considerably in the extent of change over the decade 2001-11. In the sub-figures, each observation’s vertical distance from the line of equality gives the extent of change in women’s literacy. The following points stand out. First, hearteningly, in all taluks the vast majority of villages exhibit positive vertical distances, that is, increase in literacy. Second, in each taluk there are nevertheless villages where literacy reduced over the decade (given by points below the line of equality). Third, among the majority of villages exhibiting positive change, the variation in extent of change is
considerable. In each taluk, there are several villages where the literacy rate increasedbyover25 percentage points, but also several villages where the literacy rate increased by less than ten percentage points (including Jayapur village of Varanasi).55 Focus on a Micro-Region: Shorapur Villages To take a more in-depth look at spatial variations in patterns of development change and stasis, I explore the villages of Shorapur taluk specifically (Figure1, last subfigure). Variations across Sub-Taluk Regions Are the patterns of village-level variation across taluks – represented by Figure1 – also replicated within different regions of the same taluk? To examine this for Shorapur, I divide the taluk into four subregions. As the map in Figure 3 shows, the taluk has relatively clear western and eastern portions; further, since the large majority of villages are in the central portion, I make an additional north-south distinction for the central portion.56 Figure 2 presents village-level variation in change of women’s literacy for each of the four sub-taluk regions (West, East, North-Central, and South-Central) of Shorapur taluk. It turns out that the overall intra-taluk pattern (Figure2) is remarkably similar to the overall inter-taluk pattern (Figure 1): again, villages within each of the newest micro-regions exhibit considerable variation in women’s literacy rates for both 2001 and 2001, and importantly, there are sharp variations in the degree of change across villages – within each small micro-region there are villages experiencing near-stasis and villages experiencing considerable increase in literacy. That is, even as subnational regions are divided and further subdivided, the pattern of villagelevel variation appears unchanging. In this sense, the subregion appears to be a microcosm of the larger region, however the two are defined. This finding is intriguing and needs greater exploration beyond the scope of this study, although in the concluding section I discuss some of its implications.
Identifying Villages for Trajectory Comparisons Of the 181 villages with over 250 population in both years in Shorapur taluk, there were eight villages where the women’s literacy rate increased by over 25 percentage points in these ten years, a very substantial increase. At the same time, there were (coincidentally) eight villages where the women’s literacy rate decreased a little. Interestingly, increases and decreases have occurred for villages with both relatively high and relatively low starting literacy levels in 2001. For instance, the villages Halgera and Mavinmatti both had approximately 6% women’s literacy in 2001, but while Halgera increased rapidly to 37% by 2011, Mavinmatti was stagnant at the extremely low level of 6%. Similarly, while Bekanhalli and Sadab had relatively better starting rates (26% and 28% respectively) in 2001, Bekanhalli increased rapidly to 52% by 2011 while Sadab stayed stagnant. In fact, Halgera, starting with a much lower literacy rate than Sadab in 2001, more than compensated the difference by 2011. Not only are the longitudinal stories of Halgera and suchlike villages remarkable, but so also are the surprising stories of stasis in villages such as Mavinmatti and Sadab – remarkable in their seeming imperviousness to the forces that shaped progress in literacy in Halgera. Variations across Neighboring Villages Besides the considerable variation in extent of increase and decrease in literacy over the decade, it is also the case that villages of these different types are often neighbors. This is shown by Figure 3 (left map) for Shorapur taluk.57 There are several instances of villages with large increase in literacy that are spatially close to villages with small increase or even decrease in literacy.58 The Puzzle of Micro-Level Variation in Literacy These micro-level variations in change over a relatively short period of time, especially among villages that are so close to each other, constitute not only a profound puzzle but also
an opportunity to understand the spatially diverse processes of change (and in many cases, stasis).We know little in a rigorous, systematic way about what drives change in the Halgeras and Bekanhallis of Shorapur, and what impedes change in the Mavinmattis and Sadabs of Shorapur. They appear to be subject to similar taluk-level, district-level, state-level, and national-level forces, and yet their significantly different trajectories of change (and nonchange) call for greater scrutiny. What are the village-level and community-level dynamics that account for these different trajectories?59 Can these differences in dynamics be attributed to systematic factors, or are they idiosyncratic? Unfortunately, extant research offers only a limited set of answers to these micro-level questions, which are nevertheless very pressing from a policy perspective. The research techniques that need to be deployed to form this kind of micro/granular understanding of the dynamics of change and non-change lie within the realm of qualitative research, and typically have to be re-invented in application for each particular context. Those familiar with these villages, and those who work in and with them, may have a sound sense of what drives these remarkable changes and these remarkable cases of stasis. However, there is a need to identify and systematize such knowledge using sound methodology, and subject it to empirical tests beyond the realm of intuition and experience, in order to explore replicability, as challenging as that may be. As Michael Lipton notes: “why do some villages grow and develop, while other villages, with similar initial resources, languish? … Such questions require comparative VS [village studies approach] – not necessarily in a large sample of villages.”60
Variations in Village Development Trajectories: The Case of Child Sex Ratio The sex ratio is, in some contexts, a rough-and-ready measure of the gender gap in development indicators, particular for health but more broadly as well. I focus on the child sex ratio, defined as the number of girls for every 100 boys in the under-6 population.61 This indicator is also a proxy for severe neglect of the girl-child. There has been concern in policy circles that in several districts of India, particularly in the North and West of the country – including in some experiencing rapid economic growth – child sex ratios have declined in 2001-2011.62 Sex Ratio Trajectories in Four Micro-Regions Figure 4 shows child sex ratios in the villages of the four selected taluks in 2001 and 2011. Dashed lines separate each sub-figure into four quadrants. Villages in the northeast quadrant have more girls than boys in both census years, villages in the southeast quadrant have more girls than boys in 2001 but not 2011, villages in the southwest quadrant have fewer girls than boys in both years, and villages in the northwest quadrant have fewer girls than boys in 2001 but not in 2011. As would be expected for these regions, a large number of villages had fewer girls than boys in both years, as shown by the density of the southwest quadrant compared to other quadrants in each sub-figure. However – and encouragingly, given the prevailing malebias – there is nevertheless a set of villages in each taluk where there were more girls than boys in both years (northeast quadrant). While the imbalanced sex ratio and its variation across villages is cause for concern and merits investigation, here I focus on changes in the sex ratio in this decade. In all four taluks, the number of villages where it increased and decreased was approximately the same (as shown by villages above and below the solid line of equality). Even among the majority of
villages with adverse sex ratio for girls in 2001, in Varanasi and Kherwara taluks only a little more than half the villages improved their sex ratio by 2011, and this is clearly a matter of concern.63 This goes to support the point made by Scott Fulford “The differences between states or districts are masking the much larger variation at the local level...The variation in the fraction of girls from village to village and decade to decade with in each village is so large...”64 In Jayapur village itself, the child sex ratio favored boys in 2001 (92), and declined even further by 2011 (82). Interestingly, the nearby village of Ranibazar not only improved women’s literacy faster over the decade despite starting at a similar level as Jayapur (as described earlier), but also reduced the imbalance in the child sex ratio (91 to 95) despite starting at a similar level as Jayapur.65 This may point to factors driving change simultaneously in the literacy and sex ratio indicators, and to the extent that the change in one may or may not influence change in the other.66 Clearly, greater field-level qualitative research is required to illuminate these possibilities. The Puzzle of Micro-Level Variation in Child Sex Ratio In all the taluks, while there are several instances of sex ratio going from below 100 to above 100 in 2001-2011, there are even more instances of the opposite trend (comparing the northwest and southeast quadrants). All of this suggests wide variation in the gender gap in health and gender-specific neglect of children even in these geographically clustered sets of villages. While the literature has built up an impressive knowledge base of the generic factors affecting child sex ratio and sex ratio in general,67 these micro-level variations – over a short period in villages with seemingly more similarities than differences to start with – afford an opportunity to isolate and understand local, contextual factors that drive change. To the extent that kinship systems and female autonomy matter, and there is discrimination against
the girl child (at both the family and the community level) as regards food allocation and healthcare, and to the extent that sex-selection technology is accessible to parents, we would expect it to be the case for all parents and communities living across villages in spatially proximate clusters. Therefore, while such important generic factors can explain the level of the child sex ratio in 2001 or 2011, they are unlikely to substantially explain why neighboring villages may display stark differences in both the level and – importantly, for the purposes of this study – change in the child sex ratio over the decade. Spatial variation within village clusters must depend on less generic, more context-specific factors, and these are crucial for effective micro-level policy. Further, it is likely that the factors driving differential change across villages in such clusters are more amenable to analysis as a complex of factors rather than taken separately. The identification of such spatial variations in change invites greater scrutiny to understand what is going “right” with one set of villages (especially those high above the solid equality line in the sub-figures) and what is not going right in another set (especially those below the line). What are the endogenous social, political, and public health processes at work? What are the roles of outsider interventions, particularly the different manifestations of the state? These are matters of pressing interest that cannot be answered but for careful on-the-ground research of the complex forces creating stasis and change.
The Search for Explanations: Village-Level or Region-Level? This study has taken a descriptive route and explicitly avoided the substantive question of what explains village-level variations in development change/stasis even within microregions. Extant literature on comparative longitudinal studies is relatively small and unsystematic, and we do not know the extent to which nuances of each micro-region affect the development trajectories – both change and stasis – of its villages.68 In the related context of explaining variation of development outcomes across subdistricts within a district, Sanchita Bakshi and co-authors make a persuasive case for one specific factor at the level of the micro-region, namely presence of Scheduled Tribe (ST) populations;69 this may itself proxy other geographical, political, and social factors. They note that within most districts, subdistricts with greater ST populations exhibit lower values on their development index, raising the possibility of micro-level core-periphery processes.70 One of their several examples is the (undivided) Thane district of Maharashtra,71 where northern subdistricts with low levels of development – Palghar, Dahanu, Vikramgadh, Talasari, Mokhada and Wada – coexist with southern subdistricts that are among the most developed in the country. Consider the opposite cases of the subdistricts Mokhada (backwardness rank of 539 out of 5955) and Kalyan (backwardness rank of 5907 out of 5955). Figure 5 shows that Kalyan villages indeed have decisively greater women’s literacy, and also that they have far smaller presence of ST population, suggesting that this microregion-level factor (or its correlates) is likely to influence women’s literacy. However, the evidence presented in this study suggests that what Bakshi and co-authors argue is true of subdistricts within a district – namely, considerable variation in development outcomes across subdistricts within a district – is likely also true across villages within those very subdistricts. For instance, Figure 5 (right graph) for Kalyan shows that there is a strong
negative correlation between ST population and women’s literacy. Were these authors to disaggregate below the subdistrict level, they might have concluded that ST population may be a village-level factor after all. Further, in the case of Mokhada (Figure 5, left graph), there is considerable variation in literacy rates across villages despite largely similar ST populations, suggesting that other village-level factors maybe influential. Whether these are idiosyncratic or systematic, or in between, is a matter for detailed fieldwork to unearth. Finally, consider Figure 6, which shows women’s literacy for 2001 and 2011 for the villages in the two taluks. The scatter plots are spread out in a manner similar to Figure 1, suggesting yet again that there is considerable intra-taluk variation in the extent of progress in literacy over the decade even among villages with similar starting levels. This suggests that variations in village-level development outcomes may indeed be driven by village-specific factors, not just factors common to the micro-region. Which kinds of factors – village-level or microregion-level – matter more? These are the kinds of interesting questions that we cannot currently answer, but which must be asked, especially from a policy perspective.
Conclusion This study has documented remarkable variations in decadal changes in two key development outcomes (women’s literacy and child sex ratio) across villages within small micro-regions of India. Further, such remarkable variations in development change and stasis appear to be true in disparate micro-regions across the country, as evidenced by the case of the four taluks considered here. The objective was not to explain such variations, but merely to document and highlight them using a systematic approach based on census data. It appears that, but for small strands within an older social anthropology literature, the extent of local spatial
variations in inter-temporal development outcomes have not been sufficiently appreciated in the development literature. There are two consequences to this. First, from a policy perspective, it is important to understand the well-springs of development in each village. Statewide or even districtwide policies appear to have greater development impact in some villages over others, presumably because mediating factors differ across villages. This is not an explicit focus of extant policyformulation at the state level, nor is it explicitly explored by those taking / implementing policy decisions at the district or block / taluk levels. It is a point that was made several decades back by Robert Chambers and John Harriss in the context of their North Arcot village comparison: “reformulating approaches to rural development planning on the basis of improved understanding of economic and social processes, in their interaction at village level.”72 That policy should be tailor-made for each village is, of course, one of the foundational principles of decentralization. However, neither decentralization policy nor practice emphasizes the need for comparing villages experiencing sharp upticks in development outcomes with nearby villages experiencing stasis, in order to identify policy-relevant drivers of development in that micro-region. In the context of the new village adoption scheme, for instance, a policy to improve women’s literacy in the Prime Minister’s adopted village of Jayapur should be explicitly informed by knowledge of what seems to have worked in nearby Ranibazar, and what seems to have failed in nearby Sikhari. Michael Lipton provides an example of how village studies can produce contextually sensitive policy: “… we have many VS [Village Studies] illustrating debt structures and marketing problems in Indian rice villages. Only because of these can we now use hypothesis-testing, policy-oriented VS to discover what sorts of storage improvements are likely to benefit small farmers and landless labourers in different types of villages.”73
Second, scholarship on the drivers of development can benefit from micro-level comparative longitudinal research designs. For instance, a comparison of Jayapur-Ranibazar-Sikhari in Varanasi taluk, or Halgera-Mavinmatti or Bekanhalli-Sadab (both pairs in Shorapur taluk) – all comparative cases where women’s literacy was similar in 2001 but ended up differently in 2011 – may pinpoint the specific, local drivers of change and stasis. This study has also suggested a simple methodology for identifying villages for comparative longitudinal research using census data: for the outcome variable of interest,74 use scatter plots such as Figure 1 or 4 to identify villages with similar starting levels in 2001 but very different ending levels in 2011, just as with Halgera-Mavinmatti or Bekanhalli-Sadab. Other information can then be overlaid on such promising village pairs, for instance information on population or presence of Scheduled Castes or Tribes, or location within the village cluster (as in Figure 3) – all of which can improve the choice of villages for comparative longitudinal exploration.
Figure 1: Women’s Literacy, 2011 vs 2001
Varanasi 60 40 20
2011
40
Jayapur
0
0
20
2011
60
80
80
Kherwara
0
20
40 2001
60
80
0
20
Jhalod
60
80
0
0
20
40
40
2011
60
60
80
80
Shorapur
20
2011
40 2001
0
20
40 2001
60
80
0
20
40 2001
60
80
Note: Census data for villages with population over 250 in 2001 and 2011. Solid line is line of equality.
Figure 2: Women’s Literacy in Subregions of Shorapur Taluk, 2011 vs 2001
West
60 20 0
0
20
40
60
0
20
40
60
2001
2001
North-Central
South-Central
40 0
0
20
20
40
2011
60
60
0
2011
40
2011
40 20
2011
60
East
0
20
40 2001
60
0
20
40
60
2001
Note: Census data for 2001 and 2011. Solid line is line of equality. Villages in the East and West regions are defined as those with longitude lower than 76.4116 and greater than 76.66101, respectively. For the remaining villages, defined as Central region, North-Central and South-Central are distinguished by latitude 16.54974. Note that all villages of the taluk are included here, unlike Figure1which includes only villages with population over 250.
Figure 3: Spatial Variation of Change, Shorapur Taluk
Change in Women's Literacy
Change in Child Sex Ratio
(30,40] (20,30] (10,20] (0,10] (-10,0] [-20,-10]
(75,100] (50,75] (0,50] (-25,0] [-75,-25] No data
Note: The map uses census data layered on village geographic boundaries digitized by Sharad Lele from District Census Handbooks. Color shades reflect percentage point increase / decrease (in intervals) of women’s literacy and child sex ratio in 2001-11.
Figure 4: Child Sex Ratio, 2011 vs 2001
Varanasi
50
50
75
100
2011
100
Jayapur
75
2011
125
125
Kherwara
50
75
100
125 2001
50
75
Jhalod
125
50
50
75
100
2011
100
125
125
Shorapur
75
2011
100 2001
50
75
100 2001
125
50
75
100 2001
125
Note: Census data for villages with population over 750 in 2001 and 2011. Child sex ratio is defined as number of girls for 100 boys for children under six years. Solid line is line of equality. Northeastern quadrant shows villages with more women than men in both years; southwestern quadrant shows villages with more men than women in both years; southeastern quadrant shows villages with more women than men in 2001but not 2011; and northwestern quadrant shows villages with more women than men in 2011 but not 2001.
Figure 5: Women’s Literacy and Scheduled Tribe Population in Mokhada and Kalyan Taluks, 2011
Mokhada
60 40 0
0
20
20
40
60
women's literacy rate in 2011
80
80
Kalyan
0
25 50 75 ST population (%) in 2011
100
0
25 50 75 ST population (%) in 2011
Note: Census data for villages with population over 250 in 2001 and 2011.
100
Figure 6: Women’s Literacy in Mokhada and Kalyan Taluks, 2011 vs 2001
Mokhada
60 0
0
20
40
2011
40 20
2011
60
80
80
Kalyan
0
20
40 2001
60
80
0
20
40 2001
60
80
Note: Census data for villages with population over 250 in 2001 and 2011. Solid line is line of equality.
1
The scheme is called Sansad Adarsh GramYojana and it has a dedicated website at http://www.saanjhi.gov.in/. Interestingly, Home Minister Rajnath Singh also adopted a village in Lucknow District (village Benti), but falling under a different constituency. See Subir Sinha, “Lineages of the Developmentalist State: Transnationality and Village India, 1900-1965,” Comparative Studies in Society and History Vol. 50, No. 1 (2008), p. 48. 2
Sanchita Bakshi, Arunish Chawla and Mihir Shah, “Regional Disparities in India: A Moving Frontier,” Economic and Political Weekly Vol. L, No. 1 (2015), pp. 44–52. 3
4
On the fomer, see Bibek Debroy and Laveesh Bhandari, District-Level Deprivation in the New Millenium (New Delhi: Konark Publishers, 2003); on the latter, see Bakshi, Chawla and Shah, “Regional Disparities in India.” 5
All figures for literacy rate are from the decennial census and refer to individuals over six years of age. Donald T. Campbell and H. Laurence Ross, “The Connecticut Crackdown on Speeding: Time-Series Data on Quasi-Experimental Analysis,” Law & Society Review Vol. 3, No. 1 (1968), p. 52. 6
7
Varanasi was chosen because Jayapur village in it was adopted by the Prime Minister. Kherwara was chosen because I have worked there previously, and Jhalod and Shorapur because I am currently researching these regions. Given their diversity in geographical, cultural and historical terms, they constitute an unsystematic “random” sample. The analysis presented in this study has also been performed on other regions – taluks of Mandya district (Karnataka), taluks of Banswara district (Rajasthan), taluks of Dhamtari district (Chhattisgarh) and remaining taluks of Udaipur district (Rajasthan) – with the same qualitative results as those emerging from Varanasi, Kherwara, Dahod, and Shorapur taluks. See Surinder S. Jodhka, “From “Book View” to “Field View”: Social Anthropological Constructions of the Indian Village,” Oxford Development Studies Vol. 26, No. 3 (1998), pp. 311–331. 8
See Walker and Ryan, Village and Household Economies. Michael Lipton describes “the Indian VS [Village Studies] cycle: large, ambitious but underfunded programmes of villagelevel fact-finding, which are abandoned at the very moment when they should, instead, be redirected towards smaller, hypothesis-testing programmes, in which each village study is staffed and funded adequately to draw conclusions relevant to policy.” (Michael Lipton, “Village Studies and Alternative Methods of Rural Research,” in Biplab Dasgupta, ed., Village Studies in the Third World (Delhi: Hindustan Publishing Corporation, 1978), p. 24. 9
10
The program was directed by Michael Lipton and focused on issues of livelihoods, migration, and nutrition. Its first published book volume was on India (Claire M. Lambert, ed., Village Studies: India, 1950-75: Data Analysis (London: Mansell Publishing, for the Institute of Development Studies, 1976). K. Nagaraj and R. Rukmani, “Bibliography of Village and Town Studies of Tamil Nadu,” Discussion Paper No. 76, Kerala Research Programme on Local Level Development (Thiruvananthapuram: Centre for Development Studies), 2004. Besides unpublished theses, their comprehensive listing includes studies from the IDS bibliography, the Madras Institute of Development Studies library, and Census monographs. The studies on Tamil Nadu are not representative of studies in other states because Tamil Nadu has received disproportionately more attention (see John Harriss and J. Jeyaranjan, “Rural Tamil Nadu in the Liberalisation Era: What Do We Learn from Village Studies?,” Working Paper No. 183, Institute of South Asian Studies, National University of Singapore, 2014). 12 Barbara Harriss-White and John Harriss, “Green Revolution and After: The ‘North Arcot’ Papers and Long Terms Studies Of the Political Economy of Rural Development in South India,” Working Paper No. 146, QEH Working Paper Series, 2007; see also Christopher Fuller and Jonathan Spencer, “South Asian Anthropology in the 1980s,” South Asia Research Vol. 10, No. 2 (1990), pp. 85–105. The break in the older village studies tradition may be 11
also partly because the coherence of the village unit has diminished considerably (see Dipankar Gupta, “Whither the Indian Village: Culture and Agriculture in “Rural” India,” Economic and Political Weekly Vol. 40 (2005). Nevertheless, there continues to be new work framed around villages; for instance see Diane P. Mines and Nicolas Yazgi, eds., Village Matters: Relocating Villages in the Contemporary Anthropology of India (Delhi: Oxford University Press, 2010). 13 Interestingly, both the discourse / constructivist school and the deductive-nomological / statistical school – at opposite ends of the epistemic spectrum – nevertheless imply decline in the older village studies tradition. 14 See Amrita Datta, Gerry Rodgers, Janine Rodgers and B. K. N. Singh, “Contrasts in Development in Bihar: A Tale of Two Villages,” The Journal of Development Studies Vol. 50, No. 9 (2014), pp. 1197–1208; K. Nagaraj, “A Note on Methods of Village Study,” paper presented at “Studying Village Economies in India: A Colloquium on Methodology” (2008); K. Narayanan Nair and Vineetha Menon, eds. Social Change in Kerala: Insights from MicroLevel Studies (Delhi: Daanish Books: 2008). 15
Robert Wade, Village Republics: Economic Conditions for Collective Action in South India. (San Francisco: Institute for Contemporary Studies, 1994), p. 3 & 6. 16
“Contrasts in Development in Bihar”, p. 1198.
For a review of this literature, see Suraj Jacob, “Towards a Comparative Subnational Perspective on India,” Studies in Indian Politics (forthcoming). 17
18
The focus here is on studies of natural variation. In addition, a recent wave of randomized control trials (RCTs) has created variations through experimental interventions, but I do not survey them here. 19
Robert Wade, Village Republics.
20
David Mosse, The Rule of Water: Statecraft, Ecology and Collective Action in South India (New Delhi: Oxford University Press, 2003). Robert Chambers and John Harriss, “Comparing Twelve South Indian Villages: In Search of Practical Theory,” in B. H. Farmer, ed., Green Revolution? Technology and Change in Rice–Growing Areas of Tamil Nadu and Sri Lanka (London: MacMillan, 1977). Some instances of cross-sectional differences that they note: Randam had almost three times the percentage of landless households as Vinayagapuram (48 vs 18%), and the daily wage rate for ploughing in paddy fields was almost five times in Dusi as in Veerasambanur (Rs. 3.5 vs 0.75). The only longitudinal difference they incorporate is demographic: for instance, Vengodu had no increase in population, while Kalpattu had a 30% increase in 1961-71. 22 Grace Carswell, “Dalits and Local Labour Markets in Rural India: Experiences from the Tiruppur Textile Region in Tamil Nadu,” Transactions of the Institute of British Geographers Vol. 38, No. 2 (2013), pp. 325–338. 21
23
Anirudh Krishna, Active Social Capital: Tracing the Roots of Development and Democracy (New York: Columbia University Press, 2002). Krishna studies a total of 69 villages (including villages in Ajmer and Dungarpur districts), but 16 of these were studied intensively.
Gabrielle K. Kruks-Wisner (2012), “Claiming the State: Citizen-State Relations and Service Delivery in Rural India,” PhD thesis, MIT. However, the study explores variation in claim-making not only across villages but also within villages. 24
Adam Auerbach, “Clients and Communities: The Political Economy of Party Network Organization and Development in India's Urban Slums,” World Politics (forthcoming). 26 For a review, see Rajshri Jayaraman and Peter Lanjouw, “The Evolution of Poverty and Inequality in Indian Villages,” The World Bank Research Observer Vol. 14 No. 1 (1999), pp. 1–30. 25
27
Kathleen Gough, Rural change in Southeast India: 1950s to 1980s (Delhi: Oxford University Press, 1989). 28
Andre Beteille, Caste, Class, and Power: Changing Patterns of Stratification in a Tanjore Village (Berkeley: University of California Press, 1965). Anirudha Krishna, “Politics in the Middle: Mediating Relationships Between Citizens and the State in Rural North India,” in Patrons, Clients, and Policies: Patterns of Democratic Accountability and Political Competition, ed., Herbert Kitschelt and Steven Ian Wilkinson. (Cambridge: Cambridge University Press, 2007), pp. 298–383; Oliver Mendelsohn, “The Transformation of Authority in Rural India,” Modern Asian Studies Vol. 27 No. 4 (1993), pp. 805–42. 29
30
For instance, Jan Breman, The Poverty Regime in Village India (New Delhi: Oxford University Press, 2007). John Harriss, “Reflections on Caste and Class, Hierarchy and Dominance,” Seminar Vol. 633 (2012). 31
32
For instance, see Peter Kloos, Jan Breman, and Ashwani Saith, eds., The Village in Asia Revisited (New York: Oxford University Press, 1998). 33
The study was initiated by the Madras University-Cambridge University Project on Agrarian Change. (One of these, Dusi, was studied by Gilbert Slater and his students in 1916.) This initial set of studies was published in Farmer, Green Revolution?. In 1983, the 11 villages were resurveyed by a joint team from the Tamil Nadu Agricultural University and the International Food Policy Research Institute. This second set of studies was published in Peter B. R. Hazell and C. Ramasamy, eds., The Revolution Reconsidered: The Impact of High-Yielding Rice Varieties in South India (Baltimore: John Hopkins University Press, 1991). In the mid-1990s these 11 villages were again resurveyed by Barbara Harriss-White and a team from the Madras Institute of Development Studies. This third set of studies was published in Barbara Harriss-White and S. Janakarajan, eds., Rural India Facing the 21st Century: Essays on Long Term Village Change and Recent Development Policy (London: Anthem Press, 2004), pp. 1-539. 34 John Harriss, J. Jeyaranjan and K. Nagaraj, “Land, Labour and Caste Politics in Rural Tamil Nadu in the 20th Century: Iruvelpattu (1916-2008),” Economic and Political Weekly Vol. LV, No. 31 (2010), pp. 47–61. 35
Peter Lanjouw and Nicholas Stern, eds. Economic Development in Palanpur over Five Decades (Oxford: Clarendon Press, 1998).
36
Thomas S. Walker and James G. Ryan, Village and Household Economies in India's SemiArid Tropics (Baltimore: Johns Hopkins University Press, 1990); Reena Badiani, Stefan Dercon, Pramila Krishnan and K. P. C. Rao, “Changes in Living Standards in Villages in India 1975-2004: Revisiting the ICRISAT Village Level Studies,” Working Paper 85, Chronic Poverty Research Centre (2007); Uttam Kumar Deb, G. D. Nageswara Rao, Y. Mohan Rao and Rachel Slater, “Diversification and Livelihood Options: A Study of Two Villages in Andhra Pradesh, India 1975-2001,” Working Paper 178, Overseas Development Institute, 2002. Although they trace trajectories of multiple villages over time, the focus is not on comparing trajectories. Christophe Z. Guilmoto, “Irrigation and the Great Indian Rural Database: Vignettes from South India,” Economic and Political Weekly Vol. 37, No. 13 (2002), pp. 1223–1228; K. Nagaraj, “A Note on Methods of Village Study,” paper presented at “Studying Village Economies in India: A Colloquium on Methodology,” 2008. Further, some studies have followed this research design by matching villages across censuses, for instance Scott L. Fulford, “The Changing Geography of Gender in India,” Working Paper, Department of Economics, Boston College (2013). The graphs presented in this study adopt a similar strategy. 37
John Harriss, “Population, Employment, and Wages: A Comparative Study of North Arcot Villages, 1973-1983,” Chapter 6 in Hazell and Ramasamy, Green Revolution Reconsidered. 39 Breman, The Poverty Regime in Village India. In these villages, landless tribal Halpatis and Dublas form the underclass, and Anavil Brahmins and Kanbi Patidars form the landed upper class. In Chikhligam and Gandevigam villages, he concludes that the lot of the landless Halpatis changed in form but not substance, with bonded laborers becoming farm servants and then casual labor, but predominantly still in agriculture. Even irrigation-induced agricultural development (in Bardoligam) only benefited the landed rich but not the landless poor. The lot of the underclass even in the “golden industrial corridor” was little different in substance, as the fourth village (Atulgam) reveals. 38
40
T. Scarlett Epstein, A. P. Suryanarayana and T. Thimmegowda, Village Voices: Forty Years of Rural Transformation in South India (New Delhi: Sage, 1998). “Contrasts in Development in Bihar.” Mahisham had greater migration, which increased the participation of women in economic and other activities. 41
42
As a general matter, village studies in India often do not carefully articulate or legitimate the methodologies they employ. As Harriss-White and Harriss (“Green Revolution and After”) note, they “are hampered both by intervillage variation and by individualistic methodologies” (p. 15). 43 Mill, John Stuart, The System of Logic, Ratiocinative and Inductive: Being a Connected View of the Principles of Evidence, and the Methods of Scientific Investigation (London: Longmans, Green, 1843/1872). For a methodological discussion in the Indian context, see Jacob, “Comparative Subnational Perspective.” Mill also proposed a “method of similarity” where chosen units are different except for outcome and variable of interest, allowing the analysis to establish robustness but not explanation per se; see John Gerring, Case Study Research: Principles and Practices (New York: Cambridge University Press, 2007). Note that both these “Mill’s methods” come from a positivist epistemological tradition, as do the works surveyed here, including the studies by Chambers and Harriss as well as Krishna. For an elaboration of the abductive approach to 44
comparison, see Dvora Yanow, “Interpretive Analysis and Comparative Research,” in Isabelle Engeli and Christine Rothmayr, eds., Comparative Policy studies: Conceptual and Methodological Challenges (London: Palgrave Macmillan, 2014), pp. 131–159. Chambers and Harriss, “Comparing Twelve South Indian Villages.” Farmer, Green Revolution, p. 7; Chambers and Harriss, “Comparing Twelve South Indian Villages,” p. 301 & 303; and Harriss-White and Harriss, “Green Revolution and After.” 47 Skocpol, Theda and Margaret Somers, “The Uses of Comparative History in Macrosocial Inquiry,” Comparative Studies in Society and History Vol. 22 No. 2 (1980), pp 174–197. 45 46
48
Anirudh Krishna, Active Social Capital. Choice of village was based on government data regarding villages’ participation in a Watershed Development Program in which Krishna was involved some years prior to the start of fieldwork (Appendix A). Specifically, among participating villages, some had “high” achievements in the watershed program, others were in the “middle” category, and still others in the “low” category. The fieldwork villages were a random sample from each of these categories. 50 Krishna has another chapter, titled “Examining Community Harmony: Why Are Some Villages Peaceful and Others Not?”, which exhibits the same methodological problem. Note that although villages were selected randomly rather than purposively, Krishna bases it on Mill’s method (p. 97) because there is variation in the dependent variable (level of development or community harmony) across cases. However, this is insufficient to establish a valid causal design. Further, Krishna’s strategy of controlling for conflating factors through statistical rather than design procedures is insufficient to establish comparability for inferential purposes; for an elaboration of this point, see Jacob, “Comparative Subnational Perspective.” 49
In several other cases as well, village selection occurred through random sampling. In the ICRISAT studies, representative villages were chosen from representative taluks of representative districts of India’s semi-arid tropics (Walker and Ryan, Village and Household Economies). In Krus-Wisner’s study, village selection was through random sampling after block-level stratification by literacy rates. In the case of Robert Wade’s study, the sample was not chosen randomly – rather, it was chosen purposively to showcase villages with all, some, or none of four corporate institutions (council, fund, field guards, common irrigators). However, “starting” or other kinds of similarity of these villages is insufficiently explored, compromising explanatory inference. Gerring, Case Study Research; see also Jacob, “Comparative Subnational Perspective.” Harriss, “Population, Employment, and Wages.” 53 Similarly, in Robert Wade’s study (Village Republics), there is insufficient tracing of longitudinal trajectories although some longitudinal aspects are mentioned in passing. For example (p. 155): “It is said that until 10 to 15 years ago Polur had a village council, fund, field guards and common irrigators much like Kottapalle, but that it all finished because of constant conflicts within the council.” 51 52
54
The sub-figures present villages that could be matched between the two censuses; only a very small minority of villages were left unmatched in each taluk. The choice of threshold village size (250), while arbitrary, was designed to ignore overly small villages which might present different dynamics. However, it turns out that inclusion of these villages does not change the overall qualitative results reported in this study.
55
This is not to deny differences across taluks; for instance, Varanasi had a greater proportion of villages with 25 percentage point increase compared to Shorapur taluk. 56
Villages in the West and East regions are defined as those with longitude lower than 76.4116 and greater than 76.66101, respectively. For the remaining villages, defined as Central region, North-Central and South-Central are divided by latitude 16.54974. Note that this is, obviously, a rough-and-ready division unaligned with ecological, topographical, or demographic distinctions –but it is sufficient to serve the present purpose of showing that village-level variation replicates itself even when subregions are further subdivided. 57
The map uses census data layered on village geographic boundaries digitized by Sharad Lele. I am grateful to him for making these files publicly available. Nevertheless, as a technical matter, estimation of the Moran’s I statistic suggests the presence of spatial autocorrelation – that is, the extent of change in a village’s literacy rate is likely to be correlated with the extent of change for its neighbors. 58
There is, of course, a large literature on women’s literacy, which has identified several Potential drivers of change in literacy. For instance, Aparna Sundaram and Reeve Vanneman identify one context where greater participation of women in the formal labor force leads to girls being withdrawn from school, thereby stunting their formal education as well as widening the gender gap in education (Aparna Sundaram and Reeve Vanneman, “Gender Differentials in Literacy in India: The Intriguing Relationship with Women’s Labor Force Participation,” World Development Vol. 36, No. 1 (2008), pp. 128–143). The point, however, is that despite a large literature on generic factors, we need micro-level understanding of actual factors driving change and stasis in each micro-region. 59
Michael Lipton, “Village Studies and Alternative Methods of Rural Research”, in Biplab Dasgupta (ed) Village Studies in the Third World. Hindustan Publishing Corporation, Delhi, 1978; p. 25. 60
61
Although it is more common to define sex ratio in terms of 1000 boys, the definition used here has the advantage of interpretation as a percentage figure. 62
Jean Dreze and Amartya Sen, An Uncertain Glory: Indian and its Contradictions (Delhi: Penguin, 2013). 63
The relative proportion was slightly better in Jhalod and Shorapur taluks.
64
“The Changing Geography of Gender in India.”
65
Further, the other village mentioned earlier, Sikhari, saw a sharp fall in the child sex ratio even though it started in 2001 favoring girls (105 to 85). 66
I am grateful to Sreeparna Ghosh for suggesting this.
Perianayagam Arokiasamy, Isabelle Attane and Christophe Z. Guilmoto, “Sex Ratio at Birth and Excess Female Child Mortality in India: Trends, Differentials and Regional Patterns,” in Watering the Neighbor’s Garden: The Growing Demographic Female Deficit in Asia, ed. Isabelle Attane and Christophe Z. Guilmoto (Paris: Committee for International Cooperation in National Research in Demography, 2007), pp. 49–72; Tim Dyson and Mick Moore, “On kinship structure, female autonomy, and demographic behavior in India,” 67
Population and Development Review Vol. 9, No. 1 (1983), pp. 35–60; Monica Das Gupta, “Selective Discrimination against Female Children in Rural Punjab, India,” Population and Development Review Vol. 13, No. 1 (1987), pp. 77–100. 68
Those inclined towards conventional statistical techniques may prefer to set up a regression model instead, in order to provide a statistical explanation via estimation of average conditional relationships between literacy or sex ratio and an array of independent variables. However, this cannot identify either the complex of factors that drive development in microregions, or the underlying mechanisms at work – and most importantly, it may end up conflating explanation with correlation or statistical explanation (Jacob 2015). 69
“Regional Disparities in India.”
70
Sanchita Bakshi and co-authors also note that other factors at the level of the micro-region may be at work, for example to explain sharp variations among micro-regions of districts such as Bilaspur or Durg which do not have substantial ST concentrations (“Regional Disparities in India”). 71
Palghar district was partitioned from Thane in August 2014, and it is constituted by the northern poorer-performing subdistricts identified by these authors, although they refer to the undivided district. Chambers and Harriss, “Comparing Twelve South Indian Villages,” p. 301. They also write: “Why are villages so different? And what of it so far as policy is concerned? The possible absurdity of the first question is reduced by the practical significance of the second.” 72
Lipton, “Village Studies.” Besides gender-specific literacy rates and sex ratio, the census has village-level data on village land-use (including irrigated area) and a variety of physical and social infrastructure variables. 73 74
