Understanding, Measuring and Utilizing Social Capital

AGRICULTURAL SYSTEMS Agricultural Systems 82 (2004) 291–305 www.elsevier.com/locate/agsy

Understanding, measuring and utilizing social capital: clarifying concepts and presenting a field application from India Anirudh Krishna

*

Sanford Institute of Public Policy, Duke University, P.O. Box 90245, Durham, NC 27708, USA

Abstract Social capital has been defined as a resource, a propensity for mutually beneficial collective action, that communities possess to different extents. Communities with high levels of social capital are able to act together collectively for achieving diverse common objectives. While the concept of social capital is valid universally, the measure of social capital will vary by context. It must be related in each case to aspects of social relations that assist mutually beneficial collective action within that particular cultural context. A locally-relevant scale of social capital was developed to assess whether and how social capital mattered for development performance in 69 north Indian villages. Variables corresponding to other bodies of explanation, including extent of commercialisation, relative stratification, and relative need were also examined, but a combination of high social capital and capable agency was found to associate most closely with high development performance. Agency is important particularly in situations where institutions are not available that enable citizens to connect with the state and with markets. The productivity of social capital is considerably reduced on account of this institutional gap. Development performance can be improved in these situations by adding to the stock of social capital and also through enhancing agency capacity. Ó 2004 Elsevier Ltd. All rights reserved.

Keywords: India; South Asia; Social capital; Agency; Development

*

Tel.: +1 919 613 7337; fax: +1 919 681 8288.

E-mail address: krishna@pps.duke.edu.

0308-521X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.agsy.2004.07.003

292

A. Krishna / Agricultural Systems 82 (2004) 291–305

1. Introduction Social capital has become prominent within the vocabulary of development practitioners, but there is little consensus about what it is, how it is observed and measured, which outcomes it supports, and more importantly, which outcomes it does not support. In popular imagination, and also in some scholarly writings, social capital is often equated with all that is good about human and social relations. Such a view lacks concrete empirical referents and makes it extremely difficult to study this variable independently of its effects. Consequently, people – particularly busy practical people – tend to be somewhat cynical when references are made to the power of social capital.

There is considerable hope, however, in the idea of social capital if we can clarify the concept and associate it with images of reality that we see around us in our everyday work. By helping focus attention on talents and energies that exist at the grassroots – within indigenous communities and local groups – concern with social capital helps rectify an imbalance in development practice, as discussed below. It frees up space for communities to plan their own futures, and it forces national planners to recognize that communities are not all alike.

Development practitioners have long been aware that programme results vary considerably from one location to another, but so far it has been hard to account for these differences. Why should one community derive higher benefits from some programme than another, especially when both communities received the same level of overall assistance? A number of different reasons – quality of leadership, effectiveness of programme staffs, etc. – can be suggested to explain these observed differences. Social capital is one other possible explanation that must be considered.

Social capital refers to the quality of human relations within some well-defined social group that enables members of this group to act in cooperation with one another for achieving mutual benefits. More formally, it is defined as ‘‘features of social organization such as networks, norms and social trust that facilitate cooperation and coordination for mutual benefit’’ (Putnam, 1995, p. 67).

It is important to note that different definitions of social capital abound within the burgeoning literature on this concept (Baron and Hannan, 1994; Robison et al., 2002). ‘‘The point is approaching,’’ asserts Portes, 1998, p. 2), ‘‘at which social capital comes to be applied to so many events and in so many different contexts as to lose any distinct meaning’’. It is important, therefore, to clarify exactly what one implies while referring to social capital before indicating any effects to which this concept might give rise. In the definition considered for this article social capital is regarded as the common property of a group that facilitates and promotes collective action for the mutual benefit of group members. Alternatively, following Lin (2001), social capital could have been considered as resources embedded within social networks that are used by individuals to facilitate particular actions, i.e., social capital could have been seen as an individual-level and not a group-level asset. Inkeles (2000, p. 247)) stresses, however, that ‘‘what makes the study of social capital compelling is the assumption of added value. . . it permits communities to do what they could not do. . . without the addition of social capital. By contrast, studies focused on the indi-

A. Krishna / Agricultural Systems 82 (2004) 291–305

293

vidual generally say nothing about added value, and concentrate rather on competitive advantage in the gaining of shares from a fixed pie. They almost invariably deal with a win–lose situation. It is a much more attractive proposition for humankind if we can establish that adding social capital at the communal level is a win–win situation for both the individual and the community at large’’.

We are all aware of communities that are more cooperative, where members care for one another and where they are prepared to act collectively for common purposes, whereas other groups have more infighting, discord, jealousy, and selfishness.

The definition of social capital followed here formalizes the common observation of differences in cooperation among communities or groups, and asserts that intergroup differences have important consequences for the results that these groups can achieve in practice.

More cooperative groups will achieve better development and governance outcomes, it is claimed, while groups with lower levels of social capital will have lower achievements in the social, political and economic realms. 1 What happens within the group influences its achievements in the world outside.

This is a claim worth taking seriously. It adds an important new element to development practice, and it asserts that this element can make the crucial difference between success and failure. Plans and programs are central for development, it was believed earlier, and better plans and programs were sought in order to achieve better results. However, performance continued to be mixed even as programs of assistance were refined and improved. Some communities performed very well and other communities performed very poorly, and the reasons for success and failure were never completely well understood.

It is important to understand these reasons better. Knowledge of the reasons for success and failure can help ratchet up the pace of development and make significant improvements in peopleÕs quality of life. It is thus worthwhile to examine whether development outcomes can be improved significantly through paying attention to social capital.

Section 2 of this paper presents the methodology of the study, which provided indepth qualitative as well as quantitative data for assessing the effects of social capital.

Section 3 discusses issues related to the measurements of social capital.

Section 4 investigates development outcomes in 69 north Indian village communities. The findings indicate that social capital matters significantly for development outcomes. Villages that have high levels of social capital also have significantly higher development performance. However, the utility of social capital is enhanced considerably when this resource is utilized strategically. The capability of a particular set of village agents is of special importance in this respect. Villagers who are knowledgeable and well versed in dealing with government agencies and market operations help target villagersÕ social capital toward incentives available in the external environment. This suggests that in addition to high social capital, agency capacity 1 It is possible that what is regarded as a benefit by group members might constitute a loss for society at large (Rose, 1999). Members of a mafia group, for example, can have high social capital, which enables them to achieve and share large benefits from criminal activities that damage the overall good of society.

294

A. Krishna / Agricultural Systems 82 (2004) 291–305

matters. 2 Villages that do not have capable agents achieve much lower development success even when they have high levels of social capital.

Finally, Section 5 completes the paper, discussing an agenda for action to help capture the potential of social capital while avoiding some pitfalls. Policy recommendations underline the importance of strengthening intermediary institutions so that social capital can flow without hindrance to improve development results.

2. Study methodology The fieldwork started in the summer of 1998 and lasted for over two years. It involved living among village communities in the Indian states of Madhya Pradesh and Rajasthan, where I had previously lived and worked for 15 years as a government official. I did not present myself as a government official, however, and selected parts of the state where I had not worked before, using public transportation and living in public buildings made available by villagers.

The first six months, I spent time observing patterns of collective action in these villages to identify which activities villagers undertook collectively, and which individually. The activities that most villagers thought appropriate to execute with others provided culturally appropriate indicators of social capital (as discussed in Section 3).

The case study investigation covered 16 villages, with long-term stay in the villages and speaking with a wide cross-section of villagers. Later, the study was extended to 69 villages. This extended survey was carried out with the help of a group of 16 field investigators, equally men and women, who are themselves villagers from this region.

Together, the investigators and I interviewed over 2000 villagers, selected by random sampling from all adults in the 69 villages, as described in Krishna (2002).

The average population of these villages is about 1100 persons, divided into about 200 households. The average area of a village is a little under 2000 acres, almost half of which consists of wasteland and scrub forest. The main occupation of 90% of village residents is agricultural production. However, average landholding is small, about half an acre per capita, and land quality is generally quite poor. The area is mostly semi-arid and drought is frequent, with rainfall well below average in two years out of every five. Nearly 45% of village residents live under the official poverty line.

3. Measuring social capital Putnam et al. (1993) measures social capital in Italian regions with reference to the density of citizensÕ membership in formal organizations operating in these regions. It needs to be noted that this is a proxy measure of social capital: it is not directly con2 Agency is understood here in terms of the definition provided by Sen (1999, pp. 18–19) of agent as ‘‘someone who acts and brings about change, and whose achievements can be judged in terms of her own values and objectives, whether or not we assess them in terms of some external criteria as well.’’

A. Krishna / Agricultural Systems 82 (2004) 291–305

295

cerned with norms or with trust but it looks, instead, at a manifestation that accompanies social capital in a specific setting. It is not obvious that similar manifestations of social capital will be observed in other cultures. Diverse forms of human activity are likely to develop in different ecological and cultural settings. Networks, roles, rules, procedures, precedents, norms, values, attitudes and beliefs will differ, and measures of social capital that are relevant for one set of cultures can be quite irrelevant for others.

Density of formal organizations is a particularly inappropriate indicator for Rajasthan villages, because it does not reflect self-organization, since nearly all formal organizations have been initiated by some government agency. Villagers join these organizations mostly in order to gain some immediate economic benefits and usually no collective action is associated with such government-sponsored organizations. Formal organizations in this context do not, therefore, provide any reliable indication of voluntarism and cooperation among villagers. However, several informal networks exist and are very active.

I developed a locally-relevant scale for measuring social capital in Rajasthan which relies upon assessing participation in informal networks. Not all activities are valid for investigating cooperation and coordination within any particular context. Since social capital exists ‘‘in the relations among persons’’ (Coleman, 1988, pp. S100–101), only those activities that a particular culture considers best undertaken collectively rather than individually should be considered for measuring social capital.

In our study area six such activities were identified as possibly constituting a measure of social capital: (i) Membership In Labour-Sharing Groups: The survey asked if the person was a member of a labour group in the village, i.e., whether he or she worked very often with the same group, sharing the work either on her/his fields, or in some public work, or for a private employer. Responses were coded 0 for ‘‘no’’ and 1 for ‘‘yes’’. These responses were aggregated by village, thereby measuring the proportion of villagers participating in such networks. More than 80% of respondents answered ‘‘yes’’, though this proportion varied from 98.5 to 73% among different villages. (ii) Dealing with crop disease: With regard to crop disease the survey asked ‘‘if a crop disease were to affect the entire standing crop of their village, who would come forward to deal with this situation’’. Responses ranged from ‘‘Every one would deal with the problem individually’’, scored 1, to ‘‘The entire village would act together’’, scored 5. IndividualsÕ responses were averaged for each surveyed village. (iii) Dealing with natural disasters: The survey asked ‘‘if there were some calamity in the village requiring immediate help from government, e.g., a flood or fire, who in this village would approach government for help?’’ Responses varied from ‘‘No one’’, scored 1, to ‘‘The entire village collectively’’, scored 5. (iv) Trust: The survey asked ‘‘if a friend in the village faced the following alternatives, which one would he or she prefer’’:

296

A. Krishna / Agricultural Systems 82 (2004) 291–305

To own and farm 10 U of land entirely by themselves (scored 1) To own and farm 25 U of land jointly with one other person in this village (scored 2) The second alternative would give each person access to more land (12.5 U, instead of 10 U represented by the ďŹ rst option), but they would have to work and share produce interdependently. The more people trusted other people in their village, the more they preferred the second option. The question was framed so that the respondent was not making an assessment of his or her own level of trust, but rather of how trusting other people in the village were in general. (v) Solidarity: The survey asked if it were possible to conceive of a village leader who would puts aside his welfare and that of his family to concern himself mainly with the welfare of village society. Responses ranged from ‘‘Such a thing is not possible’’, scored 1, to ‘‘Such a thing happens quite frequently in this village’’, scored 3. (vi) Reciprocity: The survey asked ‘‘if some children of the village tended to stray from the correct path, for example, if they were disrespectful to elders, or they disobeyed their parents, or were mischievous, who in this village would feel it right to correct other peopleĂ•s children.’’ Four alternatives were posed: ‘‘No one,’’ scored 1; ‘‘Only close relatives’’ scored 2; ‘‘Relatives and neighbours,’’ scored 3; and ‘‘Anyone from the village,’’ scored 4. Average responses by village ranged from 3.45 to 1.70.

Responses to these six questions were highly correlated with one another and they all loaded highly on a single common factor, indicating that villages that have high scores on any one manifestation of social capital also tend to have high scores on the other ďŹ ve manifestations. Village scores on the six separate items were aggregated to form the social capital index (SCI). This index is highly correlated with each of its six constituent factors, and with other indicators of local norms and informal networks.

However, there is hardly any correlation between the SCI and any of the other structural and agency variables considered below.

It is important to note that social capital is not directly observable; people carry it inside their heads. What one can observe and measure are some manifestations or behavioural consequences of social capital, including both structural and cognitive elements (Upho, 2000). Dierent cultures uphold dierent expressions of social capital; hence its observable aspects will vary contextually. Dierent measures of social capital are appropriate for dierent social and cultural contexts. The measures that were developed and used for Rajasthan may be adapted to other developing country contexts. 3 But this claim needs to be veriďŹ ed through careful ďŹ eld inquiries. Developing a locally appropriate index of social capital is the ďŹ rst step to take in order to be able to examine the utility of social capital and to investigate how to add to its stock, and good knowledge of the speciďŹ c societal context is crucial for developing such an index.

3

The social capital assessment tool (SCAT), adopted for empirical work by the World Bank, adapts and builds upon these earlier investigations in Rajasthan (Krishna and Shrader, 2000).

A. Krishna / Agricultural Systems 82 (2004) 291–305

297

4. Assessing the utility of social capital Development means different things to different people. For people living in these villages, four important outcomes define development performance: Livelihood stability, ranking first, followed by employment generation, poverty reduction, and quality of basic services. It is hardly surprising that these four development outcomes outrank by far any other outcomes valued in the economic realm.

Living in a semi-arid region, where rainfall is scarce and highly variable – where most people depend on agriculture for their livelihood but most have relatively small plots – these villagers are constantly concerned about food, fodder, and firewood availability. While food crops (millets, maize and some wheat) are grown on privately owned land, fodder and fuelwood are generally collected from common lands, which in most cases comprise between a third and a half of total village area. Protecting, preserving and developing these common lands is a collective concern of villagers, and their performance in a program of common land development provided one indicator for examining the relative impact of social capital.

Employment generation provided a second indicator for comparing the utility of social capital compared to other social explanations. Nearly half (45%) of all village households depend for their subsistence on wage employment for at least one month each year. Such employment is mainly obtained from some government-sponsored construction work near their village. This work is usually of a casual nature, engaging labourers from day to day, and it is in the interest of villagers that these opportunities be located as close to home as possible. People of different villages vie with each other for such employment opportunities to be located in their particular village, and the extent of success in this collective endeavour – measured in terms of days of public works employment per capita averaged over the last five years – provided a second indicator of development performance.

Poverty reduction benefits provide a third indicator. Nearly the same proportion of households that rely on wage employment in order to make ends meet (45%) earn incomes below the official poverty line. Government programs provide assistance to such households, but some villages have been consistently successful in attracting more poverty grants per capita over several years. The five year average of poverty grants per capita constitutes the third indicator of development performance.

Villagers are also concerned about better education, health facilities and clean drinking water. An index composed of their subjective assessment of service quality constitutes the fourth development performance indicator.

4.1. High- and low-performing villages These four development indicators are related to different aspects of development performance: livelihood stability, employment generation, poverty reduction benefits, and service quality. Nevertheless, village scores of these four activities are quite closely correlated with another, and they align commonly on a single underlying factor, as shown in Table 1.

298

A. Krishna / Agricultural Systems 82 (2004) 291–305

Table 1 Development performance: factor pattern Livelihood stability

0.82

Poverty assistance Employment generation Quality of basic services

0.69 0.84 0.71

Adapted from Krishna (2002).

High performing villages do well in general across multiple and different programmes, and low performing villages do poorly overall. These data indicate that the nature of the programme does not matter so much for development success or failure. Rather, there is some specific quality of villages that makes some perform well, while others perform poorly.

In order to assess what this hidden quality might be – whether it is social capital or something else – the study uses a single Index of Development Performance that combines village scores on these four activities (DEVINDEX). Although it would be better to use a more common statistic, such as GDP, to compare economic development performance across these villages, such statistics are not available for this level of disaggregation.

These specific measures correspond to what villagers themselves value. Two of these measures – employment generation and poverty reduction – reflect villagersÕ ability to obtain higher development benefits from the government. Livelihood stability reflects how well villagers have been able to preserve assets created through investments made jointly by villagers and the government. 4 Outcomes related to protecting, preserving and developing the common lands were measured seven years after government funding had come to an end. During this period protection and preservation were managed entirely by villagers. Quality of basic services, the fourth indicator, also reflects villagersÕ collective inputs. While the state government provides these basic services, the quality of these services and villagersÕ satisfaction levels vary substantially by village. While the extent of provision is closely related to population norms, service quality can be improved through collective request to government officials and by collective participation in service delivery.

Next we investigate which factors help explain differences between high-performing and low-performing villages.

4.2. Assessing the utility of social capital and other explanations Different bodies of theory were considered to identify alternative explanations for the observed variations in village development performance scores. Several independent variables considered for this purpose are discussed briefly below.

According to some observers, differences in caste, ethnicity and wealth might limit villagersÕ potential for acting collectively for common development benefits. The fol4

Villagers contributed 20% of the direct costs of these schemes and they also contributed considerable amounts of their time.

A. Krishna / Agricultural Systems 82 (2004) 291–305

299

lowing variables were constructed to assess the impact of stratiďŹ cation and heterogeneity on development performance scores. N_CASTES is the number of dierent caste groups that reside in any village, and it measures the extent of homogeneity within the village population. Another variable, CASTE_DOM, measures the proportion of village households that belong to the most numerous caste group.

Relative modernization might also aect village performance (Inkeles and Smith, 1974). Since the impulse of modernization and commercialisation is likely to be felt less within villages that are located farther from markets, the variable DISTMKT is used to measure the distance in kilometres to the nearest market town. The relative level of infrastructure facilities might also help to explain dierences in development performance (Gaiha et al., 2001). The variable INFRASTR combines scores of the level of facilities related to transportation, communications, electriďŹ cation, and water supply.

Dreze and Sen (1997) propose that economic development in communities is expected to be closely related to educational achievement. LITERACY was calculated as the sample percentage of persons having ďŹ ve or more years of formal education.

Though all villages in the sample face shortages of fodder and fuelwood, some are more acutely aected by scarcity than others. Collective action theory based on a rational-actor premise would predict that communities whose members have a greater sense of deprivation and more acutely experience common needs will be more likely to act collectively to oset this situation (Wade, 1994). This relative needs hypothesis was tested with the help of two independent variables: DRYLAND assesses the ratio of rainfed (non-irrigated) cropped area to irrigated cropped area, and PERCPOOR measures the percentage of village households under the oďŹƒcial poverty line. 5 In addition to these variables that relate to structural dierences among villages, a number of agency variables are also considered. The following agency forms are commonly functioning in these villages, and they are each regarded by some body of theory as having a signiďŹ cant (positive or negative) impact upon village development outcomes. However, the eectiveness, utility and range of functions of each type of agency dier from village to village, and these variations are considered here to develop scales for comparing agency strength.

Traditional patrons: the corresponding variable, str_PCR, is constructed from scores on survey responses to nine questions on patron–client relations.

Leaders of dierent caste groups in a village: the corresponding variable, str_CASTE was constructed from responses to survey questions related to strength of caste leadership in each village.

Traditional village councils: the corresponding variable, str_VC, is constructed from responses to survey questions related to the strength of such councils in each village.

5

Note that the Index of Development Performance denotes the ow of beneďŹ ts that a village community received over a period of time. The percentage of people below the poverty line is a stock variable (and not a ow variable) that represents at a particular point in time the extent of poverty in a village.

300

A. Krishna / Agricultural Systems 82 (2004) 291–305

OďŹƒcial local government (panchayats): the corresponding variable, str_PANCH, is constructed from responses to survey questions related to the strength of these elected bodies in each village.

Political parties: the corresponding variable, str_PARTY, is constructed from responses to survey questions about the activities of political parties in each village.

New village leaders (village youth with some education who have arisen over the past twenty years and who help other villagers establish contacts with state agencies and market operations): 6 the corresponding variable, str_NEW, is constructed from responses to survey questions concerning their existence and utility in each village and the extent of contact that villagers have with such new leaders.

4.3. Results All the above variables – along with social capital – were considered to explain the variation of development performance scores across villages. Results from regression analysis are shown in Table 2. Model 1 includes all the structural and agency variables, but only literacy had a signiďŹ cant eect, and the explanatory power is very low. Model 2 has an improved ďŹ t, it drops most of the nonsigniďŹ cant structural and agency variables and adds the SCI, which is signiďŹ cant, along with literacy.

Model 3 adds an interaction term of the SCI and capability of new leaders. 7 The social capital variable alone loses signiďŹ cance, but the interaction term has a strong eect, and the explanatory power increases further. Overall, the only signiďŹ cant variables are social capital, literacy, and the capacity of the new young village leaders.

None of the other variables – neither commercialisation, modernization, relative need, stratiďŹ cation, nor heterogeneity – are signiďŹ cant. 8 Social capital contributes to development performance, but its impact increases where agency is also stronger. These two variables – social capital and capacity of new village leaders – have a multiplicative impact in inuencing development performance. The impact of social capital is quite low (often close to zero) when agency capacity is very low or non-existent.

Social capital is a resource that needs to be marshalled carefully by agents who can bring it to bear eectively and reliably upon incentives available in the external environment of state and market. Where such agents are present, social capital helps to produce beneďŹ cial development outcomes. On the other hand, villages that do not have such eective agents are unable to convert their stock of social capital into a

6

For a more detailed discussion related to the emergence of new village leaders, see Krishna (2002, 2003).

7 Interaction terms combining the SCI with other agency variables were also considered, but none of these terms was signiďŹ cant.

8 Multicollinearity is not a major concern as the value of the Condition Index is 24.68 for Model 1, indicating moderate collinearity, and it is less than 15 for Models 2 and 3, indicating low collinearity.

A. Krishna / Agricultural Systems 82 (2004) 291–305

301

Table 2 OLS Regressions on development performance: DEVINDEX is the dependent variable Model 1 Intercept Independent variables (a) Societal variables DRYLAND PERCPOOR DISTMKT INFRASTRUCTURE NCASTES CASTEDOM LITERACY (b) Agency variables Str_PCR Str_PANCH Str_PARTY Str_CASTE Str_VC Str_NEW (c) Social capital (SCI) (d) Interaction (SCI*Str_NEW) N R2 Adj-R2 F-ratio F-probability

22.4(15.7)

0.09 (0.19) 0.79 (3.48) 0.21 (0.39) 0.24 (1.77) 0.15 (0.97) 1.14* (0.44) 0.14 (0.68) 1.45 (3.89) 0.97 (5.39) 0.25 (4.41) 0.78 (4.82) 0.87 (2.68)

60 0.12 0.04 1.56 0.186

Model 2 60.2** (24.70)

Model 3 47.2*(22.9)

0.61 (3.21) 0.30 (0.36)

0.52 (3.24) 0.27 (0.37)

0.001 (0.05) 0.65* (0.37)

0.002 (0.04) 0.52* (0.24)

0.89 (4.90) 1.12 (2.70) 1.10* (0.34)

0.69 (4.77) 0.61 (2.64) 0.35 (0.36) 0.08*** (0.009) 60 0.43 0.37 6.27 0.0001

60 0.28 0.21 3.39 0.01

Note: Standard errors are reported in parentheses.

Adapted from Krishna (2002).

DEVINDEX: development performance index; N_CASTES: number of different caste groups that reside in any village; CASTE_DOM: proportion of village households belonging to most numerous caste group; DISTMKT: distance in kilometers to the nearest market town; INFRASTR: level of transportation, communications, electrification, and water supply facilities; LITERACY: percentage of people having five or more years of formal education; DRYLAND: ratio of rainfed cropped area to irrigated cropped area; PERCPOOR: percentage of village households under the official poverty line; str_PCR: strength of traditional patrons; str_CASTE: strength of caste leadership; str_VC: strength of traditional indigenous village council; str_PANCH: strength of official local government; str_PARTY: strength of political parties; str_NEW: strength of new village leaders; SCI: social capital index. * p 6 .05. ** p 6 .01. *** p 6 .001.

flow of development benefits. Several practical consequences emerge from these findings, as discussed below.

5. Where do we go from here?

Social capital and agency both matter for development results, and they matter in conjunction with one another. High agency capacity multiplies the beneficial effects

302

A. Krishna / Agricultural Systems 82 (2004) 291–305

of social capital; and when agency capacity is low large stocks of social capital may be relatively worthless in terms of development results.

Why agency matters in this way relates to the weakness of middle-level institutions in this context. Communities, especially rural communities, are only poorly connected to the state and to markets, and they are not able to profitably engage with state and market agencies, so even very large stocks of social capital are not easily translated into high development benefits.

Communities that possess high social capital are better equipped to act collectively for mutual benefit, but this alone does not ensure that collective action will achieve its goals. Especially in terms of democracy and development, in the cases where collective action is used to solicit services from the state – i.e., where the result is not entirely or even mainly within citizensÕ control – it can very well be that collective action might end up being a wasted effort. To succeed in achieving their goals, citizens must also, at a minimum, be well informed about the decision-making processes within the state, and able to gain access to the officials who make and implement these decisions.

Adequate information and easy access are implicitly assumed within the social capital hypothesis, but information and access are not always at hand. They are relatively easily available to citizens of some countries, and they are extremely hard to obtain by citizens of others.

Agency matters in the rural Indian contexts studied here, because information about state institutions and government programs is not easily available to most villagers. Low literacy and poor-quality physical infrastructure contribute to this gap, but villagersÕ ability to obtain high-quality information is also limited by weak middle-level institutions that link the grassroots upwards to the state and national level.

In the Indian countryside, political parties are weakly organized, and local governments serve mostly as implementing agencies of a centralized state. Thus, neither parties nor local governments help villagers effectively connect to state agencies.

To gain access to the offices of the state and to avail themselves of the benefits of government programmes, villagers must take recourse to locally available agents.

Without the support of capable agents, it is not clear what ends they will target with their collective efforts and what strategies they should adopt.

As weaknesses in middle-level institutions produce large gaps in information and access, it becomes difficult for citizens to take full advantage of the opportunities for self-development that are made available by state organizations and market operations (Ostrom, 1996; Tendler, 1998). Agency becomes important in these situations.

Since middle-level institutions are not only weak in rural India, but also in other developing and transition countries, agency should matter in a fairly large number of cases. 9 Where the formal institutions of the state have been crafted from above and mesh poorly with informal and traditional institutions at the grassroots of society, a fairly

9

See, for example, Rose (1999) for Russia, Seligson (1999) for Central America and Bates (1999) for Zambia.

A. Krishna / Agricultural Systems 82 (2004) 291–305

303

significant gap often appears in the middle. National-level institutions and community-level social capital can both be made more productive in these contexts through policies that help build up effective middle-level institutions, creating strong bridges over the existing institutional gap.

Indigenous agency types can develop in response to this gap between the demand and the supply of intermediation services. The new village leaders have acquired considerable utility in this regard. Because villagers frequently consult with them in respect to diverse dealings with state and market agencies, and because they are knowledgeable and capable of effectively bargaining with government officials and party politicians, they increasingly represent the voice of a substantial number of villagers. They provide assistance at times of great need – e.g. taking a sick person to the district hospital in the middle of the night, helping another escape lightly from the clutches of a rapacious policeman, finding old-age pension support for someone else.

Some new village leaders seek to profit economically from the work that they do on behalf of individual villagers, but more than half of the over 100 new leaders interviewed are motivated by the possibility of achieving high status and political positions, while short-term economic considerations matter relatively little in their calculation of personal profit. Respect and status are accorded more highly to leaders who do not seek to become rich at the expense of their fellows. Such leaders acquire a fund of goodwill in the village by charging relatively small fees for the services that they provide.

Alternative avenues for connecting with state and markets have become available in this manner. People in villages use these connections to make more efficient use of their stock of social capital. An important collective resource – that might have remained dormant on account of missing middle-level institutions – has been mobilized and made productive through the intervention of these village agents.

Social capital may or may not be easy to build up over the short term. The evidence in this regard is mixed and so far inconclusive. Putnam et al. (1993, p. 179)) propose that social capital is accumulated only very slowly. ‘‘History determines’’, they claimed, and ‘‘historical turning points. . . have extremely long-lived consequences’’. Analyses undertaken in other parts of the world indicate, however, that social capital may not be a historically fixed endowment and that it might be possible to build up stocks of social capital within relatively short spans of time (e.g., Hall, 1997; Schneider et al., 1997). The issue is far from resolved.

While research remains to be done on the causes of social capital, experience shows that agency strength can grow relatively rapidly. Building middle-level institutions is important, but institutional forms do not need to be imported from elsewhere, and imported institutions can quite often fail to live up to their expected performance (Firmin-Sellers, 2001; Scho¨nwa¨lder, 1997). Development results can be improved considerably even within the existing system, especially if one is willing to work alongside the forces that have developed indigenously and of their own accord. New middle-level institutions are more likely to succeed if they are linked with what villagers already have, especially if they are able to hold the leaders accountable using local knowledge and everyday understandings of right and wrong. Different

304

A. Krishna / Agricultural Systems 82 (2004) 291–305

agency types have developed indigenously, and it is worth examining whether these can be strengthened and helped to acquire value and stability. Following this bottom-up route to institutional building can considerably enhance the productivity of existing social capital.

Acknowledgements I thank Ruth Meinzen-Dick and two anonymous reviewers for providing very helpful comments and suggestions.

References Baron, J., Hannan, M., 1994. The impact of economics on contemporary sociology. J. Econ. Lit. 32 (September), 1111–1146.

Bates, R.H., 1999. Ethnicity, capital formation, and conflict. Social Capital Initiative Working Paper No. 12. The World Bank, Washington, DC, USA. Coleman, J.S., 1988. Social Capital in the creation of human capital. Am. J. Sociol. 94, S95–S120.

Dreze, J., Sen, A., 1997. Indian Development: Selected Regional Perspectives. Oxford University Press, Delhi, India.

Firmin-Sellers, K., 2001. The reconstruction of society: understanding the indigenous response to French and British rule in Cameroun. Comp. Polit. (October), 43–62.

Gaiha, R., Imai, K., Kaushik, P.D., 2001. On the targeting and cost-effectiveness of anti-poverty programmes in India. Dev. Change 32 (2), 309–342.

Hall, P., 1997. Social capital in Britain. Paper prepared for Bertelsmann Stiflung Workshop on Social Capital, Berlin, June.

Inkeles, A., 2000. Measuring social capital and its consequences. Policy Sci. 33 (3/4), 245–268.

Inkeles, A., Smith, D.E., 1974. Becoming Modern: Industrial Change in Six Developing Countries.

Harvard University Press, Cambridge, MA, USA. Krishna, A., 2002. Active Social Capital: Tracing the Roots of Development and Democracy. Columbia University Press, New York, USA. Krishna, A., 2003. What is happening to caste: a view from some North Indian villages. J. Asian Stud. 62 (4), 1171–1194.

Krishna, A., Shrader, E., 2000. Cross-cultural Measures of Social Capital: A Tool and Results from India and Panama. Social Capital Initiative Working Paper Series. The World Bank, Washington, DC, USA. Lin, N., 2001. Social Capital: A Theory of Social Structure and Action. Cambridge University Press, Cambridge, UK. Ostrom, E., 1996. Crossing the great divide: co-production, synergy and development. World Dev. 24 (6), 1073–1087.

Portes, A., 1998. Social capital: its origins and applications in modern sociology. Ann. Rev. Sociol. 24, 1– 24.

Putnam, R.D., 1995. Bowling alone: AmericaÕs declining social capital. J. Democr. (January), 65–78.

Putnam, R.D., Leonardi, R., Nanetti, R.Y., 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton University Press, Princeton, NJ, USA. Robison, L.J., Allan Schmid, A., Siles, M.E., 2002. Is social capital really capital?. Rev. Soc.l Econ. 60 (1), 1–21.

Rose, R., 1999. Getting things done in an anti-modern society: social capital and networks in Russia.

Social Capital Initiative Working Paper No. 6. The World Bank, Washington, DC, USA.

A. Krishna / Agricultural Systems 82 (2004) 291–305

305

Schneider, M., Teske, P., Marschall, M., Mintrom, M., Roch, C., 1997. Institutional arrangements and the creation of social capital: the effects of public school choice. Am. Polit. Sci. Rev. 91 (1), 82–93.

Scho¨nwa¨lder, G., 1997. New democratic spaces at the grassroots. Popular participation in Latin American local governments. Dev. Change 28 (4), 753–770.

Seligson, A.L., 1999. Civic association and democratic participation in Central America: a test of the Putnam Thesis. Comp. Polit. Stud. 32 (3), 342–362.

Sen, A.K., 1999. Development as Freedom. Knopf, New York, USA. Tendler, J., 1998. Good Government in the Tropics. Johns Hopkins University Press, Baltimore, MD, USA and London, UK. Uphoff, N., 2000. Understanding social capital: learning from the analysis and experience of participation.

In: Dasgupta, P., Serageldin, I. (Eds.), Social Capital: A Multifaceted Perspective. The World Bank, Washington, DC, USA. Wade, R., 1994. Village Republics: Economic Conditions for Collective Action in South India. Institute for Contemporary Studies, San Francisco.