You are on page 1of 13

Social Capital: Meaningful and Measurable at the State Level?

Author(s): Renata Serra


Source: Economic and Political Weekly, Vol. 36, No. 8 (Feb. 24 - Mar. 2, 2001), pp. 693-704
Published by: Economic and Political Weekly
Stable URL: http://www.jstor.org/stable/4410326
Accessed: 14-10-2017 08:26 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Economic and Political Weekly is collaborating with JSTOR to digitize, preserve and extend
access to Economic and Political Weekly

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
Social Capital: Meaningful
and Measurable at the State Level?
The new elements in the recent interest in 'social capital' are that, first, one can
measure the level of trust and cooperation, and compare these levels across communities/
regions/countries, and, second, that one can assess empirically the influence of these
measured variables on various micro and macro social phenomena. This paper attempts t
analyse the feasibility and the validity of both measuring social capital at the level of the
states and identifying its role in explaining the differential performance of states.
It identifies three types of problems in the application of this methodology to interstate
analysis: finding appropriate measures for social capital; locating alternative
indicators valid for interstate analysis; and interpreting the statistical association
between social capital and state performance.

RENATA SERRA

ment of the book is that features of social


personal commitments, density of social
networks, number of tertiary associations,organisation that enhance trust and coop-
Introduction
social homogeneity, political participation),eration increase societies' vwell-being by
D oes 'social capital' help explain making government institutions more
and at various levels of aggregation (the
the differential success of statescommunity, the village, the region, the democratic and more efficient in delivering
and communities in improvingstate). In particular, social scientists havepublic goods. This work has enjoyed an
their material well-being, in reducing used different conceptualisations of social
extraordinary success and spurred a conti-
capital as an independent explanatory nuing debate. Is it possible to use Putnam's
deprivation and in having a better quality
of life? Recent literature has emphasisedvariable. methodology in other countries, and ex-
the importance of mutual trust, norms ofPart of the enthusiasm for the empirical plain differentials in institutional perfor-
cooperation and previous experiences of application of the social capital notion mance among governments with identical
collective endeavours within societies to derives from dissatisfaction with existing organisational forms? This paper exam-
explain higher incomes [Narayan andexplanatory frameworks of development. ines this question in the context of Indian
Pritchett 1995], economic growth [KnackThe question of why some countries are states. The objective has been not so much
and Keefer 1996] and institutional perfor-more successful than others in managing to replicate Putnam's methodology (given
mance [Putnam 1993]. Much effort is nowto develop their economies and establish the main instruments of the enquiry were
put to collecting data on social capital froma democratic government is still a puzzle, a short field visit and existing aggregate
various communities, to test how far it iswhich decadesof exercises in cross-country data) but to analyse the feasibility and the
linked to various aspects of well-being.leconometric analysis has not been able to validity of both measuring social capital
The notion that features of social solve. It is apparent that the issue is not at Indian state level and identifying its role
organisation, level of trust within groups the amount of sheer resources pumped into in explaining the differential performance
and previous history of cooperation aaffect country, but the existence of certain of states.
social and economic outcomes is notconditions new, for resources to be channelled The present paper presents some reflec-
as a long-standing tradition withineffectivelysocio- into priority goals. Governmenttions on this research. Its structure is as
logy and political economy can attest to.2
commitment to development and the trans-follows. The next section introduces to the
However, the new elements in the parency
recent and efficiency of the bureaucracyconcept of social capital and describes
interest in 'social capital' are that,are
first,
very crucial factors in this respect. briefly Putnam's methodology. Sections
one can measure the level of trust and However, these two conditions do not III-V identify, respectively, three types of
emerge in a vacuum. The role of civil problems in the application of such
cooperation, and compare these levels
society, of citizens' awareness of their methodology to inter-state analysis in
across communities/regions/countries, and,
second, that one can assess empirically own
the rights and responsibilities, and ofIndia. These difficulties are, respectively,
influence of these measured variables on features of social organisation must also finding appropriate measures for the
be taken into account.
various micro and macro social pheno- aspects of social capital proposed in
mena. As far as measurement is concerned, Putnam's (1993) analysis of Italian re- Putnam (1993); locating alternative indi-
social scientists have constructed variables gional governments' differential perfor-
cators valid for inter-state analysis; and
of social capital including different com- interpreting the statistical association
mance is among the most influential contri-
ponents (level of trust, extent of inter-butions on social capital. The main argu-between whatever measure of social

Economic and Political Weekly February 24, 2001 693

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
capital and state performance indicators. taken to suggest that effective govern-
capital, but that of the society as a whole
Section VI presents some conclusions. (this does not need to be the state but can
ment-citizens relationship is the outcome
of successful solutions to dilemmas of
be also a region, as long as it has formal
II institutions with some governing power).collective actions. Norms of reciprocity
and networks of civic engagement favour
Putnam's (1993) enquiry originated from
Social Capital, Institutional this outcome, because they raise the costs
the aim to explain differential government
Performance and Socio-
performance (and economic growth) acrossof defection, facilitate information among
Economic Development people, reduce uncertainty and provide
the 20 regions of Italy. The institutional
The Framework performance of each region is measured models for future cooperation. Particular
by a composite index based on 12 elements,
importance is attributed to membership in
The importance of the existence which
of include promptness in approving cultural and recreational associations, as
generalised trust and cooperation within opposed to the one in business groups,
the budget by the regional assembly, extent
communities for achieving collective goals
of legislative innovation, provision of day
unions and parties. Since the former types
cannot be overemphasised. Althoughcare the centres and of family clinics, indus-
of associations do not entertain a specific
connection between trust and well-being economic and political purpose, they are
trial policy instruments, local health unit
is universally assumed, the mechanisms by
expenditures and bureaucratic responsive-
more likely to provide the context in which
which these two factors feed into one ness to citizens' queries. This composite people internalise principles of trust and
another are complex and not easily pre- cooperation.
indicator is shown to be highly consistent
dicted. Moreover, it is still unclear how
with opinions given by the region's citizens
trust and cooperative attitudes are gene-
and with the degree of satisfaction ex- Application to Indian context
rated and maintained within societies. pressed by a sample of community leaders.
In the framework proposed by Putnam, The variable that Putnam finds more India is divided in 24 states and seven
trust and cooperation generate their effects
significant in explaining the differentialunion
in territories. Indian states share many
only through the mediation of government institutional performance in Italy is the fundamental political features and have
degree of civicness in the community.
institutions. The hypothesis is that societ- similar government structures, and yet the
ies where citizens trust one another and According to the neo-Tocquevillian frame- contrast between their achievements in
cooperate give rise to more accountable work embraced by Putnam, a civic com- terms of social stability, economic perfor-
and efficient governments, which in turn munity is characterised by active partici-mance and social indicators could not be
pation in public affairs, horizontal, egreater.
deliver a higher and better quality of public g, With independence, the consti-
equalitarian, relations of reciprocity and
goods and create the conditions for society tution gave state governments the power
to develop faster. This link is not simply cooperation (as opposed to vertical rela-
and the duty to play an important part in
the overarching objectives of the inde-
tions based on authority and clientelism),
unidirectional, since a vibrant civic society
and a democratic and an efficient govern- bonds of solidarity and trust, and distinc-
pendent India: to eradicate poverty, grant
ment are two sides of the same coin. tive social structures and practices universal education and access to health
However, the link from society to the characterised
insti- care, and ensure participation of every
by thriving local associations.
tutional context is particularly emphasised
The statistical relationship between regional
citizen to collective and political life. States
in Putnam's analysis: pressure from below
variation in institutional performance andtake expenditure decisions in the sectors
through the different instruments of
differences in the degree of civicnessofishealth, education, infrastructures, agri-
found to be more significant than in the
voting, protest, expression of disagreement culture and rural development, among
induces representatives to adhere morecase of socio-economic variables. This is others. Not surprisingly, the observed
closely to their mandate.
Unlike in the interpretation where social Table 1: Average Turnout at Elections and Literacy
capital is the number and extent of social
State Turnout Difft Literacy Diffl
relationships attributable to an individual
Andhra Pradesh 63.52 3.82 32.62 -9.14
[see Coleman 1990], social capital in Assam
the 59.58 -0.12 38.47 -3.29
Putnam's sense cannot be measured at Bihar 53.45 -6.25 28.08 -13.68
either the individual or the group level.
Gujarat 55.54 -4.16 46.37 4.6
Haryana
Social capital, even when generated by a 67.28 7.58 49.59 7.83
Himachal Pradesh 61.17 1.47 57.35 15.59
specific group, is not just for the group Karnataka
but 62.15 2.45 41.1 -0.66
for the society as a whole, e g, it benefits
Kerala 74.6 14.9 72.18 30.41

every member within a given society, Madhya


also Pradesh 50.06 -9.64 30.29 -11.47
Maharashtra 58.86 -0.84 48.61 6.85
those who are neither trusting nor trust-
Orissa 45.67 -14.03 35.44 -6.32
worthy (it is a public good in Putnam and
Punjab 65.19 5.49 41.04 -0.72
a club good in Coleman). This is so because,
Rajasthan 52.05 -7.65 26.67 -15.09
Tamil Nadu 67.7 8 47.23 5.47
if there is a critical mass of people who
Uttar Pradesh 51.25 -8.45 29.52 -12.24
keep trust and cooperation high, the insti-
West Bengal 67.09 7.39 43.63 1.87
tutions that are the expression of that society
will be efficient and accountable. When Notes: Turnout: Mean turnout at Lok and V
Difft: Difference between mean state tur
focusing on this particular form of social Literacy: Mean literacy rate, 1961-91 (19
capital, one should measure not the indi- Diffl: Difference between mean state lit
vidual or the group's endowment of social
Sources: Singh and Bose (1991); Census of

694 Economic and Political Weekly February 24, 2001

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
disparities across states in terms of their trust, absence of the conditions for indi-
As mentioned above, a civic community
economic and social development, in the viduals to pursue collective endeavours, is characterised by active participation in
presence of similarly designed government lack of historical experience of coopera-public affairs, horizontal relations of reci-
institutions, have generated much debate. tion; in summary, what Putnam calls a lack
procity and cooperation, bonds of solidar-
Differences in terms of the operation of of 'social capital'. ity and trust, and distinctive social struc-
the political system, the party in power and This seems to be a legitimate researchtures and practices characterised by thriving
the type of policies pursued are undoubt- question. The problems lie with the taskslocal associations. Regional indicators used
edly crucial to explain this divergence of constructing and interpreting measures
by Putnam for Italy were turnout at national
[Weiner 1968; Kohli 1987]. The uneven of social structures of cooperation, soli-
referenda, lack of candidate preference
distribution of the British legacy, in terms darity and trust, which may impinge on voting in parliamentary elections, mem-
of the development of the administrative state-citizens relations. These issues are bership in associations and newspaper
system, of communication and infrastruc- dealt with in the next three sections. readership. As a first step, we consider the
ture, represents another complicating fac- possibility of constructing similar indi-
tor. However, there is also a perception Ill cators for Indian states, leaving aside pre-
that features such as statewide identifi- ference voting, which is peculiar to the
Putnam's Indicators:
cation, previous experience of mass proportional electoral system then present
mobilisation and high political participationProblems of Data Availabilityin Italy.
can have a profound influence, not only and Interpretation Political Participation: The extensive
on the type of policies to be chosen, butIn this section we consider the problemsparticipation of the masses in political life
also on their effectiveness. has been a notable feature in India since
of arriving at a measure of the degree of
It has been argued, for instance, thatsociety's civicness in India, a la Putnam.
independence, and contrasts with the more
states with greater administrative efficiency
and socio-political development are also Table 2: Variables from CSDS Questionnaire
those which in the post-independence
Question asked Answer Groupings of Short Name
period had an active political community Considered Variables for the
claiming a state formation on the basis of Variable

linguistic homogeneity [Weiner 1968, OpinionsNiews


1989]. States like Tamil Nadu, West BengalDo you think it is important or not important for you to Not important Vote
and Punjab are thus set against Uttar vote the same way your caste group/community votes?
Would you say that persons we elect by voting Care Election Elected
Pradesh, Rajasthan and Madhya Pradesh.
generally care about what people like you think, Variables
Shared historical roots and a negotiated or that they don't care?
unification process were important toHow much does having elections from time to time make Good deal Election role
establish, among both administration and the government pay attention to the people - good deal,
somewhat or not much?
citizens in the former group of states, a
How much trust/confidence do you have in the state
greater civic sense, to use Putnam's words. government - a great deal, somewhat or no trust at all? Great deal Trust in SG
This does not imply, however, a lowerHow much trust do you have in local government/ Great deal Trust Trust in LG
level of conflict within society or thepanchayat/municipality
or no trust at all?
- a great deal, somewhat

absence of other kinds of divisions alongHow much trust do you have in government
caste, religious or political lines (a pointofficials - a great deal, somewhat or no trust at all? Great deal Trust official
'Action'/Personal Involvement
to which we come back later); Punjab, for
instance, has been most severely afflictedLeaving aside the period of elections, how much
interest would you say you have in politics and public Great deal Interest
by conflict and violence. affairs, a great deal of interest, some interest,
Worst performances have been explainedor no interest at all?
in terms of previous historical experiencesLet us talk about associations and organisations
other than political parties: are you a member of
(more exploitative colonial domination),
any religious or caste organisation? Yes Membership Rel/caste
archaic productive relations in the coun- in association
Aside from caste and religious organisation, do you Yes Other ass
tryside, lack of social change, ideology of
the state in power, type of family systems,belong to any other associations and organisations
like the cooperatives, farmers' association,
etc. All these elements; although some-trade unions, welfare organisations,
times studied separately, are inextricablycultural and sports organisations, etc.?
linked to one another. In particular, it seems
that political failures at the top are asHave you ever contacted any government official
for any need or problem? Yes Contact
important causes as corruption of the local Do you read newspaper? (If yes) How often - reg
bureaucracy, and malfeasance or inertia on sometimes or rarely? Regularly Media Paper
variables
the part of the population. There might be
Do you listen to radio? (If yes) How often - regularly,
a common root for the observed govern-
sometimes or rarely? Regularly Radio
ment inefficiency, bureaucratic corruptionOf those, on which source did you depend most for Any answer Information
and inertia of the people in some Indiangetting information about elections, parties and apart from NO
candidates? None, newspaper, radio, TV, is considered and
states. The question is whether this is at
all similar to the one Putnam has found any combination of two, all three. the sum is derived

for southern Italy, namely, low mutual


Source: National Election Study 1996, Centre for the Study of Developing Societies, Delhi

Economic and Political Weekly February 24, 2001 695

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
limited participation in many new states Vidhan Sabha (respectively national part and of the larger Hindi linguistic area
in Africa and Asia [Weiner 1989]. The state assemblies) for the 16 major states[Weiner 1968]. But one factor may be very
electoral process is among the most from the first elections in 1952 until 1991
crucial, namely, literacy.
important and widespread forms of partici- do not exhibit any particular trend over Average literacy across the four decen-
pation. It has remained fundamental despite time (turnout rises during 1952-67 nial butcensuses 1961-91 has been calculated
recurrent intense political violence and the fluctuates in almost all states afterwards).
for the 16 major states (Table 1). The states
suspension of democracy in 1975-77. Spatial comparisons averaged over time
with higher than average literacy (the mean
However, casting a vote is not necessa- are highlighted in Table 1, which reports
for all states is 40 per cent) are, in order,
rily dictated by civic-minded motives. In the mean turnout at the Lok Sabha and Kerala, Himachal, Haryana, Maharashtra,
the complex and internally divided Indian Vidhan Sabha elections over the period
Tamil Nadu, Gujarat, and West Bengal
society, political parties tend to align 1952-1991, along with the corresponding
(note that Himachal and Haryana are at the
themselves, at least at local level, on caste top because literacy data are available only
state ranking.3 Average turnout ranges from
and ethnic lines, with the consequence that over 75 per cent in Kerala to 45.7 per cent
from the two last census, thus leading to
voting patterns often reflect not the indi- a higher average than would be otherwise).
in Orissa (the latter is affected by the very
vidual conscious choice but the dictates of The two state rankings with respect to
low participation during the first few elec-
one's group's chefs [Nuna 1989]. tions). Besides Kerala, the other states turnout and literacy exhibit only few dif-
In the Italian context, Putnam selected with above average turnout are, in order, ferences: the two western states of Gujarat
turnout at national referenda, as opposed Tamil Nadu, Haryana, Himachal, West and Maharashtra have a turnout much
to turnout data for political elections, as Bengal, Punjab, Andhra Pradesh, lower than what predicted by their mean
more pertinent indicator of a behaviourKarataka and Haryana, that is the south-literacy levels, whereas the southern state
driven by social and moral concerns, ratherern states, the three northern states andof Andhra Pradesh has a low literacy but
than by the game of politics (as topics ofWest Bengal. high participation at elections; moreover,
referenda ranged from divorce to abor- What leads to greater participation inPunjab and Karnataka are slightly below
tion). Unfortunately, there are no analogousthese states? Certainly, explanationsthe average literacy but above the average
consultations in India, or features for whichfocussed on the contrasting patterns of turnout.
similar indicators are available. Partici- kinship between the south and the centre- The correlation coefficient between the
pation in the elections of the recently north (whereby prevalent matrilineal and average turnout at the election and average
revived panchayati raj bodies (local levels endogamous family structures in the southliteracy is positive and high (equal to 0.73,
of government), which have been hailed lead to lower conflicts within the extended different from zero at 0.001 per cent sig-
as an opportunity at last for people to takefamily, higher female status and a more nificance level). This high value implies
a role in the government of local issues, 'progressive' and open culture) come tothat either greater literacy in a state induces
is too recent to generate significant indi- mind;4 as well as those arguing that higher turnout, by empowering people not
favourable agro-climatic conditions have
cators for all states. Elections to panchayats only with the mere technical ability to vote
previous to the recent constitutional amend- induced more flexible and dynamic socialbut also with a heightened political cons-
ment were too often plagued by a politics structures, for example, in Punjab and ciousness and a sense of collective iden-
of patronage and exchange of votes to be Haryana [Banerjee 1997]. The role of tity; or a greater political consciousness of
of any use for our purpose. historical factors has been mentioned the people lead to better public services,
In the absence of alternative measures earlier. Areas which were formerly part ofmore functioning schools, hence higher
e g,
of political participation, figures on voterBritish India and/or states with established literacy levels. The latter, if true, would
turnout at assemblies elections will be takenmovements for independence have expe- confirm the social capital hypothesis (for
as the best available indicators. Data on rienced higher political participation than the empirical problems in disentangling
election turnout at the Lok Sabha and those ruled by Maharajas or constituting this relationship see Section V) but it cannot

Table 3: State-Level Values for Different Variables from 1996 CSDS Survey
State Vote Elected Elections Trust SG Trust LG Trust Off Interest Rel/caste Other Ass Contact Papers Radio Media
Andhra Pradesh 53.3 32.3 11.5 31.2 31.8 18.8 4.8 17.5 12.1 13.9 49.5
Assam 43 35.5 44.3 64.5 64.9 35.1 7 4.4 14 20.6 9.6 14.5 49.1
Bihar 27.2 14.2 11.9 30 29.8 16 9.5 3.1 3.5 15.5 5.8 10 40
Gujarat 72.5 9.1 7 22.2 39.8 11.6 3.3 1 3.5 11 16.6 3.9 43.7
Haryana 45.2 25 16.1 32.7 28 7.7 7.1 0 3 14.9 7.1 7.1 24.4
Himachal Pradesh 24.2 19.4 4.8 56.5 77.4 24.2 4.8 4.8 9.7 8.1 6.5 19.4 30.6
Karnataka 47.9 13.7 3.6 29.3 34.6 17.3 9.6 3 7.3 13.7 11.3 17.3 40.4
Kerala 74.9 45.2 40.8 54.2 58.6 18.7 19.5 12 23.6 35 29.7 35.9 83.4
Madhya Pradesh 36.8 22.9 7.1 37.8 35.2 27.3 6.6 2.5 3.5 15.6 3.7 5.9 33.5
Maharashtra 52.5 22.2 8.8 34.1 40.8 14.5 6.3 3.7 9.1 19.7 14.4 15.1 37.7
Orissa 48.3 20.1 2.4 49 51.5 34.3 5.6 2.1 7 16.1 5.6 7.5 29.8
Punjab 29.2 10.3 8.2 15.9 13.8 5.6 4.1 2.1 1 10.8 11.3 8.2 24.1
Rajasthan 39.3 26.9 11.7 50.3 36.4 11.1 2.8 1.9 2.2 14.3 11.3 12.1 32.8
Tamil Nadu 73.1 21.8 6.1 36.5 40.3 14.1 7.5 7.4 9.1 11.7 16.1 17.8 66.1
Uttar Pradesh 44.8 22.6 14 37.7 42.1 18.6 7.1 2.9 9.6 27.5 9.3 12.5 51.1
West Bengal 52 29.6 13.8 41 50.6 15.6 11.7 3.6 16.5 32 11.6 15.3 42.1
Mean 47.76 23.18 13.26 38.93 42.23 18.16 7.33 3.63 8.17 17.75 11.38 13.53 42.39

Source: National Election Study 1996, Centre for the Study of Developing

696 Economic and Political Weekly February 24, 2001

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
be denied that the causal process from expression have shown their strength not ciational activity, and interest in public
literacy to turnout is also at work. States only in achieving important material goals affairs. There was only one survey, to the
with above average literacy in 1991 are in but also in giving opportunities to people best of my knowledge, which could allow
general the same as those with greater to connect with one another and learn one to compare some aspects of social
literacy at the time of independence. The methods of cooperation and organised capital across states, e g, the 1996 National
legacy from the pre-colonial and post- Election Study Post-Poll Survey, con-
work. This enhances democratic participa-
colonial past, in terms of both the type of tion in society at large by empowering ducted by the Centre for the Study of
government in power and the policies Developing Societies (CSDS).5 Although
individuals and raising their social under-
pursued, appears as important as the standing. However, data on the numbers, the survey is conceived as post-poll sur-
actions undertaken by state governments vey, aimed at analysing voting patterns and
membership and activity of voluntary asso-
since independence. changing political allegiance of the elec-
ciations were, at the moment of the study,
Membership in Associations: The only neither available nor comparable at torate, it also asks questions on social
data on membership in associations at state state level. structures of participation and cooperation.
level are those available from the ministry Newspaper readership: The data situa- In particular, 13 questions are relevant to
of rural development relative to a number the social capital concept.6 They identify
tion on aspects other than participation is
of associations constituted under govern- broadly two areas, the first on electors'
not necessarily better. There is no compre-
ment initiatives. In its emphasis on partici- hensive survey on newspaper readership views about their community, the role of
patory rural development, Indian govern- in India nor to access to media in general, elections and about their representatives
ment has tried to establish, since 1952 with which would indicate to what extent people and officials; the second area covering
the launch of the community development are able to get information on issues rele- electors' participation in community and
programme, various rural associations, for vant at local or state level. Only data on public affairs: membership in associa-
the promotion of members' well-being and newspaper circulation and of number of tions, extent of contact with public offi-
also as a link to government programmes. TV sets per capita are available, which are cials and media usage. State aggregate
Data on the number and membership of clearly insufficient given the often public data have been considered, since statewise
women's associations ('mahila mandals') nature of media - one paper can pass comparison is the focus here.
have been used elsewhere as state-level indi- through many hands and people often gather Responses are typically qualitative cate-
cators of social capital [Morris 1998]. together at public places to watch TV. gories of the type yes/no or a great deal/
However, these data cannot be considered On the basis of what we have considered somewhat/never. For each question, we
as satisfactory measures, since many of up to now, aggregate data (with exception 'have selected the type of response that is
these associations exist merely on paper, possibly for voter turnout) appear inappro- most consistent with the notion of trust,
or their members' list is arbitrarily inflated priate for measuring three of the four aspects adherence to democratic politics and active
in orderto qualify for government funding. studied by Putnam, as such measures fail participation in public affairs. For instance,
Voluntary associations is another mat- to reflect closely people's disposition to. to the question on whether one has trust
ter, and their vibrancy 'in India is quite cooperate in order to attain collective goals. in state governments, the percentage of
extraordinary in many respects, both for Data from detailed surveys may represent people who answered "a great deal" has
its geographical incidence and its capacity more valid means for getting information been considered. Table 2 lists the 13 ques-
to involve poor and otherwise powerless on people's views of politics, their level tions, together with, the answers we are
groups. These various forms of collective of trust in governments, types of asso- most interested, and Table 3 shows the

Table 4: Matrix of Correlation of CSDS Data

Vote Elected Election Trust SG Trust LG TrustOff Rel/Caste OtherAss Interest Contact Papers Radio

Vote 1.00
1.00
Elected 0.32 1.00
0.22 0.77*
Election 0.22 0.00 0.55* 1.00
0.41 0.67* 0.03
Trust SG -0.03 0.00 0.38 0.85* 1.00
0.90 0.45* 0.15 0.00
Trust LG 0.09 0.08 0.26 0.68* 0.65* 1.00
0.74 0.32 0.33 0.00 0.01
Trust Off -0.12 0.23 0.52* 0.44* 0.48* 0.18 1.00
0.67 0.64* 0.05 0.10 0.07 0.53
Rel/caste 0.47* 0.01 0.65* 0.57* 0.67 0.34 0.81* 1.00
0.08 0.78* 0.01 0.02 0.01 0.22 0.00
Other ass 0.47* 0.00 0.55* 0.26 0.26 0.07 0.75* 0.80* 1.00
0.08 0.57* 0.03 0.34 0.33 0.79 0.00 0.00
Interest 0.38 0.02 0.61* 0.33 0.26 0.14 0.50' 0.80* 0.75 1.00
0.14 0.71* 0.01 0.22 0.32 0.62 0.06 0.00 0.00
Contact 0.36 0.00 0.46* 0.05 0.13 -0.28 0.73* 0.62 0.59* 0.47* 1.00
0.18 0.45* 0.07 0.85 0.62 0.29 0.00 0.01 0.02 0.06
Papers 0.77* 0.08 0.52* 0.48* 0.51* 0.09 0.92* 0.84* 0.76* 0.54* 0.72* 1.00
0.00 0.65* 0.04 0.06 0.05 0.75 0.00 0.00 0.00 0.03 0.00
Radio 0.36 0.01 0.55* 0.27 0.31 0.10 0.87* 0.75' 0.68* 0.58* 0.79* 0.75*
0.17 0.59* 0.03 0.31 0.24 0.71 0.00 0.00 0.00 0.02 0.00 0.00

Economic and Political Weekly February 24, 2001 697

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
correspondent percentages of respondents (coefficient 0.77). Moreover, it is signifi-Moreover, trust cannot be discarded here
in each of the main 16 states. cant that in those states where this view as a low key variable, as it is the case for
All-India values for the social capitalof elections prevails, people place greaterassociation membership: the majority of
variables so generated are quite low (nonetrust in state government (coefficients arepeople in India have some trust in their
is over 50 per cent). The highest score are0.67 and 0.55 respectively), are more in- state and local government (only 19.2 per
obtained on the issue of caste-related vote terested in politics (0.57 and 0.55) and also
cent and 23.2 per cent do not trust their
(47.7 per cent of the entire sample believesmore likely to have contacted some offi-state and local governments, respectively).
vote should not be influenced by castecials for some problems (coefficients areWhat does then trust reflect in this case?
members) and on trust in state and local0.71 and 0.61). Moreover, exposure to Is it possible that, when governments and
governments (around 40 per cent). But,media is significantly positively correlatedofficials are&themselves not too deserving,
overall, people in India appear, on average,with interest in politics and with contactthe presence of trust on the part of citizens
to have little interest in politics, to haveof officials. should be taken as a negative rather than
little confidence that elections make gov- On the other hand, one notes the lack a positive element? That is, admission of
ernments aware of ordinary people (just of correlation between the election vari- one's trust may reflect not always a con-
over 13 per cent of the sample believesables mentioned above and the belief in scious knowledge of government actions
they do), and to make scarce use of media.a vote unfettered by the influence of caste and trustworthiness, but sometimes also a
Particularly puzzling appear the lowor religious leaders. Also, the three trust mechanic and passive compliance, and
percentages of respondents who declarevariables are predictably highly correlated maybe, even the existence of collusion
membership in any organisation/associa-with one another, but there seems to with be corrupt behaviour. The problem is
tion: membership in caste or religiousno relation with interest in politics, media how to distinguish these two meanings
associations is below 6 per cent in mostusage and the belief that votes should not across different contexts. Here, the lack of
states and greater than 10 per cent only infollow suggestion from one's own caste significant relationship between trust in
Kerala, whereas membership in other or community group; the correlation coeffi- governments/officials, on the one hand,
organisations (recreational, cultural, co-cient with the feature 'contacts with 6f- and most other variables, in particular
operatives) is greater than 10 per cent just ficials' is also low (coefficients are, re- interest in politics, contact of officials and
in Kerala, Assam and West Bengal. Thesespectively, 0.33, 0.26 and 0.14).7 Positive newspaper readership, on the other, may
figures seem too low for India and oneand significant relationships are evident represent a warning as to the interpretation
possible explanation is that they capturewith the two association variables. This to be given to trust.
only membership in formal associationswould support the notion that trust inIf this is the case, it means that some of
and organisations, but exclude, for instance,members of societies and trust in govern- the aspects measured by the CSDS survey
informal yet locally recognised collectivement/representatives do feed into one are not unambiguous indicators of civic
structures that are very important, espe-another, especially because the coefficient behaviour, as one might have thought.
cially in villages. That is, features of asso-attached to 'other associations' is greaterThis is a similar problem to the one iden-
ciational life may prove to be more elusive than to those identified by caste and reli- tified earlier with other data, for instance
than what one might think. gious affiliation. However, as mentioned those on turnout at elections. One way of
One feature, which the questionnaireabove, the membership variables must separating be out the independent factors
has been able to capture quite well is the taken with caution in this survey. among a set of variables is to apply factor
exceptionality of Kerala. This state exhi- The absence of relationship between trust analysis, more precisely principal coinpo-
bits the highest scores on almost all vari-in governments/officials and most other nent analysis. The latter aims to find a
ables, and exhibits a quite unique profilefeatures is a problematic element from the small number of orthogonal components
with respect to all other states. Such find-viewpoint of social capital theory, accor- that can linearly reconstruct a larger set of
ings attest the marked political cons- ding to which trust in institutions is funda- variables, and is often resorted to when the
ciousness of Kerala people, which has led,mental in maintaining effective relation- latter are highly collinear, as this makes
as it is widely recognised, to a highly ship between government and citizens. them unsuitable to regression analysis.
effective state-citizens interaction and to
Table 5: Factor Analysis: Social Capital Variables
a very democratic society. Kerala appears
as the best Indian example of a society with Factor Eigenvalue Difference Proportion Cumulative
high social capital. Uncovering why-this 1 5.47137 3.16139 0.5746 0.5746
is so is key to understanding social capital. 2 2.30998 1.52807 0.2426 0.8172

In order to comprehend the relationship 3 0.78191 0.27107 0.0821 0.8993

between the features measured by the


Scoring Variable Factorl Factor2
survey, the pairwise correlation coefficients
Vote 0.13167 -0.14906
were calculated (Table 4). Out of the three Elected 0.15081 0.01907
variables that we have named election Election 0.13678 0.08446
variables, two seem to be highly related
Trust SG 0.12751 0.45413
Trust LG 0.06523 0.26091
to one another (those named 'elected' and
Trust Off 0.05158 0.08342
'elections'). The perception that elected
Interest 0.12604 -0.06586
representatives care about what lay elec-
Contact 0.1149 -0.01467
Papers 0.1 658 -0.47635
tors think is highly correlated across states
Radio 0.22308 -0.02856
with the belief that elections induce gov-
Media 0.10616 -0.01813
ernments to pay more attention to the people

698 Economic. and Political Weekly February 24, 2001

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
Principal components are identified for cent is interested in politics and less than responses are unambiguous indicators of
eleven of the original variables, after ex- 10 per cent reads regularly newspapers. civic virtue.
cluding the two association variables, due Punjab, on the other hand, scores very
to their low significance and to the lack low on most questions, of the first and IV
of observations for Andhra Pradesh. second 'type'; government institutions have
Interpretation and
Table 5 shows the results of factor little credit in the eyes of citizens, who Quantification of Other
analysis. The first two components, being might react to such disaffection by retrea-
the only ones that are highly relevant ting in private and closed forms of collec-
Aspects of Social Capital
(eigenvalues greater than one), andtive ex-life. This does not necessarily imply The problems identified in the previous
plaining together over 80 per cent ofthat thePunjab has a low value of social capital section, one might argue, result from
total variance, are retained for the estima- in its more extensive connotation, but only pretending to find measures of political
tion of the score coefficients. On the first with respect to the notion captured by the engagement and mutual solidarity in India
factor, all loadings are positive (Table 5). CSDS questionnaire, which may well similarly to those found in Italy. When
Factor can be easily interpreted asageneral denote a restricted version. Given the highly applying a theory to a very different con-
'social capital' variable, in the sense that conflictual political situation in particular text from the original one, it is important
states that score high on these eleven during the mid-1990s in Punjab, respon- to be able to adapt and modify concepts,
questions can be expected to have citizens dents to the questionnaire are likely to have while maintaining the basic idea. But to
who participate to a greater extent in been conditioned, even more than usual, what extent can one stretch the original
community issues at large and accept the by current events, and conveyed all their formulation to search for genuine and
rules of the existing democratic system. anxieties and disaffection with politics. meaningful forms of social capital in India?
The second factor loads negatively on In conclusion, the set of variables de- And in what measure can one thus over-
'vote', 'interest', 'contact' and the three rived from the CSDS 1996 Survey shows come the criticisms of those who raise
media variables, and positively on all others. that there is some degree of relationship doubts as to whether the concept of social
It identifies a sub-set of states where citi- among some of the factors that the litera- capital, in the Putnam sense, is at all
zens are ready to answer positively to ture groups under the heading 'social appropriate for Indian society?
questions about attitudes and trust (in the capital', but there is also lack of statistical Among Indian scholars and develop-
electoral mechanism, in governments repre- correlation between some crucial variables. ment practitioners the term social capital
sentatives and in officials), but they are Rather than implying the failure of Indian is intended mostly as social networks based
themselves not interested in politics, they social features to meet theoretical predic- on kinship ties or on caste and religion,
contact rarely officials, they are exposed tions about the mutual reinforcement of which provide vital support to individuals.
to media to a very limited extent, and, engagement and trust, this may be due toAbsence of social capital in this sense
especially, they believe it is important that a combination of two reasons. First, the means inability to face life contingencies,
votes follow the opinions of caste or reli- measured variables are unable to capture poverty, exclusion and powerlessness.
gious leaders (this is where the highest what we mean by generalised trust and Structures of cooperation are regarded as
negative loading is). This second factor collective structures of cooperation. Sec- alternative, and sometimes antithetic, to
could be interpreted as conservative com- ond, forms of participation apparently in formal institutions, as they arise to provide
pliance to existing order, implying no direct line with the notion of civic virtue may not those goods and services that the latter are
personal commitment; it may capture a be so. Situations whereby trust in govern- too weak or too corrupt to ensure. Govern-
.passive attitude on the part of citizens, ment institutions is high but forms of ments and officials are regard6d as distant
possibly due to unresponsive and dis- and often hostile by ordinary Indian
participation in collective life are scarce
interested governments, so that there is a cannot be taken as exhibiting high level people. All this implies a certain resis-
lack of a predisposition to cooperate and of social capital in the Putnam sense. Thesetance to conceptualising social capital
engage for better change. as an ingredient for effective and fruitful
two possibilities suggest the need for further
The ranking of states with respect to research using more refined methodologiesinteraction with governments. It is
these two components is quite telling that are able to probe whether surveyinstead more often conceived as a source
(Table 6). Regarding the firstfactor, Kerala
appears to be well ahead of other states, Table 6: State Rankings with Respect to First Two Factors (Social Capital)
followed by Assam and West Bengal. State Factor 1 State Factor 2
Gujarat and Punjab are instead at the
Kerala 3.05 Assam 1.85
bottom. However, the scores on the second Assam 1.03 Himachal Pradesh 1.85
component modify radically the state rank- West Bengal 0.59 Orissa 1.1
ing with Kerala just ahead of Punjab and Tamil Nadu 0.35 Madhya Pradesh 0.64
Uttar Pradesh 0.09 Rajasthan 0.41
Gujarat, but at the opposite end with Andhra Pradesh 0.05 Uttar Pradesh 0.13
respect to Assam. The reason why Kerala Maharashtra -0.03 Bihar 0.04
Himachal Pradesh -0.22 West Bengal 0.04
and Assam are so different in terms of the
Karnataka -0.28 Haryana -0.13
second factor can be gauged by looking Rajasthan -0.29 Andhra Pradesh -0.51
again at Table 3. While Kerala scores Orissa -0.39 Maharashtra -0.53

high on all questions, Assam's citizen- Haryana -0.59 Karnataka -0.6


Madhya Pradesh -0.64 Tamil Nadu -0.85
government relation appears problematic: Bihar -0.68 Kerala -0.99
around 65 per cent of the population Gujarat -0.71 Punjab -1.17
declares to trust governments but only 7 per Punjab -1.33 Gujarat -1.29

Economic and Political Weekly February 24, 2001 699

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
of support for an identifiable and con- growth, schooling attainment, availabilityapproximately 20 per cent of the popula-
nected set of people. of infrastructures and sound policies in ation each.
Nonetheless, various strands within the number of African countries. Can any measure based on states' lin-
current literature do provide good examples The basic tenet is that people similar toguistic or religious composition of the
that attest the validity, in the Indian con- population bear any statistical relationship
one another, in terms of language, ethnicity
*text, of a concept of social capital a la and culture, manage to cooperate morewith indicators of state performance? I
Putnam. Studies on educational structures successfully, because of easiness in com-found no evidence of such relationship
and facilities suggest that the source of themunication, possibility of building easilywhen using various combinations of
higher quality of services in Kerala as reputation, and the presence of commonfragmentation measures and performance
opposed to Uttar Pradesh lies with the values and conventions. To be sure, his- indicators. One important reason for the
attitudes of parents, and their willingnesstory has abundantly shown that this notionfailure of either religious or linguistic
and capability to voice their protest andis much more difficult to substantiate in fragmentation to account for differences
take appropriate action if needed, whenpractice, as many other factors intervenein state performance may be the multi-
teachers are absent, infrastructures are in determining whether trust can ensue, toplicity of the divisive lines running along
decadent, and promised funds do notthe point that the factor of diversity canIndian societies. The coexistence of these
arrive.8 Mobilisation is possible when become marginal. many and non-overlapping dimensions has
people share their understanding of the Indian society is characterised by a great prevented a single element ever to prevail
common good and trust one another to degree of fragmentation along ethnicity,and factions of permanent nature to
achieve this objective. Citizens' voice andcaste, religion and class. The importancecrystallise, so that Indian societies witness
reaction might be what makes the differ-of linguistic identity has already been noted. constant fluidity and frequent changes in
ence between malfunctioning and efficient Basically India is divided between statesalliances [Manor 1996].
schools in India, providing an appropriatewhich were formed out of linguistic Jn conclusion, therefore, whether the co-
example of the notion that greater social homogeneous areas, and those whichexistence of several linguistic and ethnic
capital makes existing physical and humanbelong to the larger Hindi speaking area. groups is an obstacle to developmental
capital more effective [Ostrom 1996]. The former group includes, out the majoroutcomes is not an issue that can be resolved
But what are the elements that enable states, Andhra Pradesh, Assam, Gujarat,on the basis of aggregate data on the degree
poor people to give voice to their dissent Karataka, Kerala, Maharashtra, Orissa, of fragmentation in different religious or
and defend their own right? Qualitative Punjab, Tamil Nadu and West Bengal. linguistic groups. It requires instead a
evidence has shown that previous expe- Leaving aside caste composition,10 re-different type and more thorough analysis.
rience of collective action, local presence ligion is a very divisive and complex issue
of NGOs, and rates of literacy are impor- at the state level. In our 16 states Hinduism V
tant factors in accounting for local com- (or its Sikh version in Punjab) represents
Social Capital and State
munities' capacity to interact with insti- the main religion, which coexists with
Performance: Problem of
tutions for achieving better outcomes. various others, the most important of which,
However, these phenomena are hard to by far in most states, is Islam. Kerala is Intervening Variables
capture by means of quantitative indicators, the only state with an equal and significant In the previous two sections we dis-
for the lack of comprehensive figures on proportion of two minority religions, as cussed the problems associated with iden-
these movements and activities concerned.Muslim and Christian communities count tifying aspects of social capital, which are
It would also be difficult to gauge their
presence in terms of states as units of Table 7: Selected Socio-economic Indicators, State Ranking
analysis. Despite the state-wide resonance State Growth HDC SDP IMR LEB FEMLIT HDI
of some of these movements, the fact that
Andhra Pradesh 12 7 5 8 9 12 10
some Indian states are as big and diverse Assam 11 11 12 13 15 9 11
as the largest world nations makes state Bihar 16 16 16 14 11 15 14
comparison quite meaningless. Gujarat 5 14 4 7 10 6 5
Haryana 7 3 2 10 5 10 4
In looking for other sources of trust and Himachal Pradesh 8 1 8 4 4 3 7
cooperation, one might wonder whether Karnataka 9 12 6 5 7 8 8
some of the theories that have been dis- Kerala 3 5 10 1 1 1 1
Madhya Pradesh 14 15 13 12 16 13 15
carded by Putnam as irrelevant for Italy Maharashtra 2 13 3 2 3 2 3
may be more appropriate in the case ofOrissa 15 4 14 16 14 11 12

India.9 In particular, the notions of social Punjab 6 2 1 3 2 5 '2


Rajasthan 4 8 11 9 12 16 13
cohesion and political harmony need to be Tamil Nadu 1 10 9 6 6 4 9
considered, since they are increasingly heldUttar Pradesh 13 6 15 15 13 14 16

up as very crucial in other contexts West Bengal 10 9 7 11 8 7 6


[Klitgaard and Fedderke 1995; EasterlyNotes: Growth: Annual rate of grow
and Levine 1995]. Ethnic and linguistic HDC: Headcount index of povert
SDP: Per capita net state domesti
fragmentation has been found to be empi- IMR: Infant mortality rates, 1992-
rically important for economic and demo- LEB: Life expectancy at birth, 19
cratic success. Easterly and Levine (1995) FEMLIT: Female literacy, 1991 (p
HDI: Human development index,
find that a measure of ethnic diversity is Sources: Shenggen et al (1998); Dre
negatively correlated with economic 1991; Government of India, Minist

700 Economic and Political Weekly February 24, 2001

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
both meaningful in the Indian context and extent also in Tamil Nadu, involving institutional performance, that of health.
measurable. Some indicators have been massive investment in human develop- This particular choice reflects the great
retained, namely, turnout at the elections ment and an emphasis on redistribution importance attributed to health services
and the two composite variables derived (land reforms, minimum wage legislation). provision for the well-being of people at
from the CSDS Survey (Table 6), yet doubts Although Kerala and Punjab have the large, the richness of studies on determi-
highest HDI in India, they differ signi- nants of health indicators in India [Dasgupta
have been raised as to their ability to reflect
appropriately the elements of cooperation, ficantly with respect to the elements com- et al 1995, 1997] and the fact that health
generalised trust and solidarity. This con- posing this indicator. Kerala's life expect- provision is only weakly related with
jecture cannot be confirmed, however, until ancy at birth, infant mortality rates and per capita income [Caldwell 1986]. Our
a test of their empirical significance is
education, particularly of women, are at selected indicators of state health perfor-
carried out. This section presents a preli- the level of world's middle-income coun- mance relate both to the level of inputs
minary analysis of the role of the social tries.13 However, growth and employmentensured by the government (the level of
are stagnating, and therefore per capita
capital indicators identified so far in explai- expenditures in the state budget, the avail-
ning state differences in performance and SDP is low.14 On the other hand, Punjabability of infrastructures and medical
socio-economic outcomes. has achieved the fastest growth and ensured personnel) and to outcome measures, such
the highest per capita income in India, butas life expectancy and human develop-
A problem with inter-state analysis is the
extreme complexity underlying regional it occupies only the 8th position in the state
economic, political and social variation, ranking with respect to mean literacy in Table 9: Factor Analysis: Performance
Variables
because of the multiple ways in which all 1961-91 (see Table 1). Another interesting
these factors combine together locally. case is that of Haryana, which has secondFactor Eigen- Difference Proportion Cumula-
value tive
Typically, 'good' performance in one aspect highest per capita SDP, but ranked 9th in
terms of literacy during the last census, and1
is often accompanied by poorer performance 2.58 1.32 0.68 0.68
2 1.26 1.10 0.33 1.01
10th in terms of female literacy.
in another yet important dimension, making
Score Coefficients
the task of drawing lessons for develop- The complex multi-dimensionality ofState Performance
ment more difficult than ever. Indian development implies the difficulty, Kerala 2.09
This is particularly evident when com- here more than ever, of explaining differ-Himachal 1.63
paring socio-economic indicators. Table 7 ential success in terms of a single indica-Punjab 1.41
Maharashtra 0.35
shows state ranking with respect to: tor. On the contrary, Putnam's analysis of
Haryana 0.21
growth in national income during the period Italian regions shows there is a rough Tamil 0.17
1980-93; 1992-93 headcount poverty in- correspondence between regions with Karnataka 0.16
dex; 1991-92 per-capita state domestic greater economic growth and those with Gujarat -0.17
WB -0.27
product; 1992-93 infant mortality rates; better performing democratic institutions
Rajasthan -0.52
1989-91 life expectancy at birth; 1991 and higher quality of public services. Andhra -0.58
female literacy rates, and 1991 human Putnam finds that his constructed index of Assam -0.61
development index (HDI). The latter is 'social capital' can explain both growthBihar -0.65
Orissa -0.91
calculated on the basis of three indicators: and democracy. UP -1.13
per-capita consumption, life expectancy We thus confine the analysis to one Madhya -1.16
and literacy rates. specific dimension of state governments'
Only states at the bottom maintain simi-
Table 8: Differences between State Indicators of Perform
larrankings, in particular Bihar and Madhya
Pradesh. Rajasthan, the other BIMARU State Health Exp Health Centres Doctors LEB HDI
state,ll although ranking low on human Andhra Pradesh -12.07 -5.86 -0.63 -0.18 -0.04
development variables, has, on its positive Assam -9.56 5.28 -0.36 -5.88 -0.06
Bihar -23.49 -0.17 -0.51 -2.28 -0.08
side, a good record in terms of growth.12
Orissa, another backward state in many Gujarat 2.43 -2.12 -1.48 -0.68 0.03
Haryana -2.48 -3.29 -0.16 2.12 0.05
other respects, has exhibited recently a Himachal Pradesh 74.08 13.88 1.85 2.82 0.02
good performance in terms of poverty. Karnataka -2.19 0.44 -1.16 1.12 0.01
Kerala 13.27 1.01 1.12 11.22 0.17
For the other states, however, the picture
Madhya Pradesh -7.68 0.31 -0.93 -6.78 -0.09
is less consistent. There are not best per- Maharashtra -5.53 -5.57 0.65 3.42 0.09
forming states as such, and states that rank Orissa -9.82 0.59 -1.65 -5,28 -0.06
high at some indicators are clearly less Punjab -0.87 3.38 4.99 5.62 0.09
Rajasthan 9.72 1.25 -0.8 -2.78 -0.08
good at others. In fact, Table 7 provides
Tamil Nadu 0.11 -1.58 1.55 1.62 0
a snapshot of different types of develop- Uttar Pradesh -21.97 -1.78 -1.34 -4.88 -0.09
ment achieved by Indian states, a topic West Bengal -3.98 -5.77 -1.2 0.72 0.02
extensively analysed in the literature. On
Notes: Health Exp: Per capita state government
the one hand, there is the model of agri- Health centres: Number of PHC centres, sub
cultural growth, increases in per capita 10,000 population).
income and investment in infrastructures, Doctors: Doctors at primary health care cent
LEB: Life expectancy at birth, 1989-91 (numb
represented by Punjab and Haryana. On HDI: Human development index, 1991.
the other hand, there is the Kerala model Sources: Reddy and Selvaraju (1995); Governmen
of development, which is evident to a less Kumar (1996).

Economic and Political Weekly February 24, 2001 701

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
ment index. The latter is 'itself both an is even more pervasive, as it impinges
ling forsocial capital, is still very significant.
outcome and an indicator of the extent to This suggests that political participation
directly on social capital and democratic per-
which people are able to benefit from does affect state performance, but merely formance. In a country where the majority
existing facilities. More precisely five because it implies higher literacy levels.
of population is still illiterate, the access
to the main instruments of democratic
indicators are considered, e g, per capita Education appears as a powerful element
health expenditures (which include five determining whether citizens are able toparticipation (voting, newspapers, but also
self-confidence and assertive attitudes in
items, namely medical and public health, participate in society at large, interact effec-
the face of public officials) may be severely
family welfare, nutrition, water supply and tively with government, and promote demo-
sanitation, and child welfare and handi- cracy. The line of causation from social limited. This issue has been understand-
capped welfare), per capita number of capital to state performance is not proved ably underplayed in Putnam's analysis of
centres in rural areas (including primary in the context of Indian states, since, Italian
if regions, but it may be the most
health centres, subcentres and community literacy is a fundamental intervening vari- crucial element to take into account in India,
health centres), per capita number of able, it is itself a product of public policy. if democracy and people's well-being are
doctors, life expectancy at birth and HDI, Kerala presents an interesting case in to be promoted. Widespread illiteracy
all measured in or around the year 1991. this respect. Its highly recognised high represents the main paradox in the largest
democracy of the world [Weiner 1991].
There is a rough consistency between the social capital, whereby trust in institutions,
state rankings with respect to these five interest in public affairs and high degree
indicators. Table 8 shows the difference of mobilisation go hand in hand, might be VI
between the state value for each indicator regarded as a product as well as a cause of
and the average value. The best perform- the observed good state performance in
Concluding Remarks
ing states are Kerala and Himachal, which delivering public services [Heller 1996;This paper has aimed to contribute to
are always above the average, followed by Krishnan 1998]. It is an effect to the extent current debates on social capital by exam-
Tamil Nadu and Punjab, each with just one that initial high levels of literacy, which ining the applicability to the Indian context
negative difference. The worst performance are in turn a legacy of last century's enligh- of a definition, used in Putnam (1993),
states are instead Andhra, Bihar and Uttar tened policies of the princes of Travancore based on the Tocquevillian notion of the
Pradesh (always below the average). and Cochin, have enabled this society after civic community. There is plentiful quali-
Given the highly collinearity between independence to give rise to democratic tative evidence from India that features
these indicators of performance, factor governments, which have ensured effec- such as the level of trust and norms of
analysis is calculated to derive a general tive access of public services to the popu- cooperation crucially impinge on social
performance measure. The first factor is lation at large (including raising the living well-being and on the effectiveness of
highly significant, explaining 68 per cent standards of the Malabar region, which, government actions. However, there has
of the total variance (Table 9), and is thus not being formerly under the mentioned been so far no study attempting to assess
retained for the calculation of the scoring the role of measurable aspects of social
principates, had very low literacy to start with).
coefficients. Above average performance Citizens' level of education is a good capital in accounting for the observed
states (that is those with a score coefficient predictor of the extent to which various differential performances across India. This
greater than zero) turn out to be, in order, kinds of public services are effectively put paper discusses the limits involved in such
Kerala, Himachal, Punjab, Maharashtra, to use in India [Minhas 1991 ]. The analysis exercises and, by doing so, provides sug-
Haryana, Tamil Nadu and Kamataka. gestion as to the direction further analysis
in this paper suggests that the role of literacy
Table 10 (top half) shows the simple
correlation coefficients between the per- Table 10: Simple and Partial Correlation Coefficient Matrices
formance indicator just obtained and the between Performance and Social Capital Variables
measures of social capital so far identified, Matrix of Simple (Pearson) Correlation Coefficients
namely turnout at election, and the two Perform Turnout Factor 1 Factor 2 Literacy

components derived from the set of eleven Perform 1

variables from the CSDS Survey. It is Turnout 0.7185 1


(0.0017)
evident that only turnout at elections Factor 1 0.3402 0.5156 1
exhibits a significant correlation with the (0.1973) (0.0409)
Factor 2 -0.2745 -0.4328 -0.0042 1
performance- indicator.
However, in Section III we noted a (0.3036) (0.0941) (0.9877)
Literacy 0.8477 0.7283 0.5829 -0.223 1
remarkable positive correlation between (0) (0.0014) (0.0178) (0.4064)
turnout and literacy. In order to control for
Partial Correlation Coefficients of the Performance Variable with:
this intervening factor, we also measure Variable Corr Sig
Turnout
the partial correlation coefficients. Table 10 0.2782 0.315
Literacy 0.6806 0.005
(bottom half) shows that the partial con- Factor 1 -0.3398 0.235
tribution of turnout to the explanation of Factor 2 -0.1176 0.689
performance variation across states, once Literacy 0.8372 0
Turnout 0.3517 0.239
the influence of literacy alone or with the
Literacy 0.7468 0.003
other social capital indicators is taken into
Factor 1 -0.4241 0.149
account, is no more significant. The reverse Factor 2 0.0527 0.864
(Obs=16)
is instead true for literacy, e g, the latter's
correlation with performance, after control- Significance level in brackets

702 Economic and Political Weekly February 24, 2001

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
in this area should take. Three types of social capital cannot be measured or funded
is not programme on 'Public Policy for Poverty
Reduction'. I would like to thank all members of
conclusions and implications emerge from relevant for socio-economic development
the IDS Poverty Programme, especially John Toye,
our analysis. in India, but rather that it may be bettet
head of the research project on 'Social Capital in
First, Indian states' data for the indica- located at a different level than the state. India', Stephen Devereux, Anu Joshi, Jim Manor
tors that are normally used to proxy civic One possible direction for future research and Mick Moore. Invaluable help and support was
behaviour are hard to identify, or the data would be to carry out the analysis at two offered by all the people I met during fieldwork
that already exist cannot be interpreted levels, respectively larger and smaller than in Delhi, especially Jean Dreze, Ajay Mahal, Mark
Robinson and A K Shiva Kumar. Earlier versions
unambiguously as measures of civicness. the state. Clusters of states within India can
of this work have been presented at the IDS
Not only aggregate data, but also survey be fruitfully compared, for instance the Poverty Conference, at a seminar at the Ford
data may not be able to capture relevant north versus south, or Hindi versus non- Foundation, Delhi, and at the IARNIW Golden
aspects, even when they are of high quality. Hindi states, or 'Bimaru' states versus the Jubilee Conference in Delhi, all in 1999. I am
We have noted for instance problems rest. At the same time, a number of smallgrateful to the many participants who made very
helpful contributions.]
relative to the very important questions of areas within each cluster may be identified
1 The World Bank is particularly involved in
participation in associations and trust in for carrying out fieldwork and derive first-
this process: for the numerous projects on
governments' institutions. The percentages hand information on the different aspects social capital see for instance under
of respondents who are members of asso- of social capital. Differences in terms of www.worldbank.org.
ciations seem too low in the light of a institutional performance might then be 2 Forreferences seeToye (1998) and Woolcock
perception of a more vibrant associational more clearly grounded in prevailing cultural (1998).
3 The dates for national and state assemblies
life derived from direct field observations. patterns and types of social relationships,
coincide during 1952-1967 and then start to
With regard to trust variables, there is the whose origin in India rests in deep-rooted diverge most of the time. For the year when
difficulty of interpreting whether low trust factors that are only weakly reflected by both a state and a national election were held,
is due to malfeasance of representatives recently drawn state borders. the average turnout is obtained. Data for
or lack of interest from the people, and Third and most important implication of Vidhan Sabha are obtained only up to 1989
rather than 1991.
whether it is restricted to existing govern- our research is that the relationship between
4 For a discussion of the contrasting kinship
ments or extend to all previous ones. social capital and institutional performance patterns between the south and the north see
These dilemmas suggest the need for is characterised by a complex dynamics. Sopher (1980), and also Dyson and Moore
survey questions to be very carefully Whereas people's involvement in public (1983), who analyse the consequences in
designed, and for future research in this affairs and their civic character might be terms of female status and demographic
indicators.
area to accompany the use of aggregate and important to explain current government 5 The CSDS has conducted various other
survey data with other methods of data performance, these elements are also af- surveys, but only the 1996 National Electio
collection, such as directfieldebservation, fected by education, which is itself a product Study and a similar one in 1972 cover a
in order to capture more effectively the of previous policies. Future research must India territory (a third, conducted after t
desired characteristics. Possibly, for the therefore address this issue and disentangle 1998 elections, has yet to generate fin
results). Being the 1972 survey based on
case of India, one would require an even the elements of this relationship. This might
smaller sample and on a much shorter
richer combination and analysis of data imply to re-qualify the frequently made questionnaire, it was not taken into accoun
sources than the one relied upon by Putnam policy recommendation that governments The 1996 National Election Study cove
for Italy, because of the lower data quality should invest in social capital. A more constituencies in 21 of the major states an
and the greater complexity of features. relevant point could be that public policy, union territories of India, on the basis of
representative sample of the Indian electorat
Second, features that supposedly reflect insofar as it promotes universal education,
stratified by gender, age, education,
collective structures of cooperation in the might have a great impact on social capital. occupation, caste, religion, economic class
Indian society and culture are elusive to Putnam himself recognised the complex and rural-urban location. The sample
measurement and meaningless to frame interaction between social capital and consisted of slightly less than 10,000 people.
within states as units of analysis. Indian institutional practices, and suggested that 6 I would like to thank V B Singh, director
of CSDS, for making available to me the
states are complex and internally differ- the experience of Italian regional govern-
state-level data relative to these 13 questions.
entiated entities; moreover, there is no ments, in the south as well as in the north,
7 Against some possible arguments thatahigher
consistency between their performance does indicate that "changing formal insti- prevalence of contact with officials could
tutions can change political practice"
indicators. Whereas the neatness of indicate greater scope for corruption, the
[Putnam 1993:184]. However, he did not
Putnam's theory as applied to regional interpretation is endorsed here that it is a
manifestation of a personal and direct
explore
development in Italy is due to the fact thatthe implications of this fact in his
involvement with public services providers,
work on Italy. The analysis of the'process
social capital explains not only differential and an opportunity to exercise rights of protest
by which institutional change can take
democratic performance, but also delivery and criticism.
of public goods and long-run growth,societies
in out of their vicious circle of 8 See Dreze and Gazdar (1997), Krishnan
distrust,
India such a clean theory is ruled out by defection and exploitation is (1998);
a see also Probe Team (1999).
9 See Putnam (1993, Ch: 3). Putnam states that
challenge mostly lying ahead. Education
the fact that states rank very differently
"the success or failure of Italy's regional
according to the particular performance
policies may provide an essential elementgovernments was wholly uncorrelated with
aspect under consideration. One cannot of ex-
this process in the Indian case. [E virtually all measures of political frag-
pect a single indicator, however composite, mentation, ideological polarisation and social
to be able to account for neither the- Notes conflict" (p 117).
10 Note that caste fragmentation cannot be
diversity of Indian regional development,
[This paper was written when I was a Research considered, for lack of census data on caste
nor state institutional performance asOfficer
such. working at the Institute of Developmentcomposition of the population since 1931
This does not amount to the conclusion that at the University of Sussex for a DflD
Studies (the 1941 Census suffered from, serious

Economic and Political Weekly February 24, 2001 703

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms
disruption due to involvement in the second Regional 'Perspectives, Clarendon Press, Development Centre, January.
world war). After independence, the abolition Oxford. Narayan, Deepa and Lant Pritchett Lant
of questions relative to caste was decided for Dreze, Jean and Haris Gazdar (1997): 'Uttar (forthcoming) 'Cents and Sociability:
political reasons (commitment to a united Pradesh: The Burden of Inertia' in Dreze J and Household Incbme and Social Capital in Rural
India where social divisions of all kind would Sen A, (eds). Tanzania' in Partha Dasgupta, and Ismail
be underplayed). Dyson, Tim and Mick Moore (1983): 'On Kinship Serageldin, (eds) Social Capital: Integrating
11 'Bimaru' is an acronym derived from the Structure, Female Autonomy, and Demo- the Economists' and the Sociologists
initials of Bihar, Madhya Pradesh,-Rajasthan graphic Behaviour in India', Population and Perspectives, Vice-Presidency for the
and Uttar Pradesh, but it also means 'ailing/ Development Review, 9 (1), March. Environmentally Sustainable Development,
sick' in Hindi. Easterly, William and Levine Ross (1995): Africa's The World Bank.
12 See also Nagaraj et al (1998) for an account Growth Tragedy, A Retrospective 1960-89,Nuna, Sheel C (1989): Spatial Fragmentation of
of Rajasthan's growth performance lately. Policy Research Working Paper 1503, The Political Behaviour in India, Concept
13 The record in terms of poverty reduction is World Bank. Publishing Company, New Delhi.
Nussbaum, N and Amartya Sen (eds) (1995): The
Esman, M J and N T Uphoff(1984): Local Organi-
also impressive: Kerala in 1972-73 was only
10th among Indian states in terms of head- sations: Intermediaries in Rural Development, Quality of Life, Clarendon Press, Oxford.
count poverty index (which was then greater Cornell University Press, Ithaca (NY). Ostrom, Elinor (1996): 'Social Capital, Self-
than 60 per cent) but managed to pass toEvans,
a Peter (1995): Embedded Autonomy, Organisation and Development', paper pre-
fourth position in 1987-88, by almost halving Princeton University Press, Princeton. sented at a Workshop in Political Theory and
- (1996):
the proportion ofpeople below the poverty line. 'Government Action, Social Capital Policy Analysis 3/6/1996, Indiana University
14 Increases in per capita consumer expen- and Development: Reviewing the Evidence Probe, Team (1999): Public Report on Education
ditures (which are reflected in a decrease of on Synergy', World Development 24 (6): in India, OUP, New Delhi.
headcount poverty) were achieved not by 1119-1132. Putnam, Robert D (1993): Making Democracy
means of national income growth (SDP has Filmer, Deon and Lant Pritchett (1997): Child
Work: Civic Tradition in Modern Italy,
grown very slowly during the past decades) Mortality and Public Spending on Health: Princeton University Press, Princeton.
but thanks to the remittances from migrants How Mucli Does Money Matter?', Policy Reddy, K N and V Selvaraju (1994) Health Care
Expenditure by Government in India, Seven
abroad, especially from those working in the Research Working Paper No 1864 The World
Gulf countries. Bank. Hills Publication, New Delhi.
Government of India, Ministry of Health (1992): Sen, Amartya( 1995): 'Capabilities and Well Being'
References Health Information of India, Central Bureau in Nussbaum M and Sen A, (eds).
of Health Intelligence, New Delhi. Shenggen, Fan, Peter Hazell and Sukhadeo Thorat,
Ahuja, M L (1998): Electoral Politics and General
Heller, Patrick (1996): 'Social Capital as a Product Government Spending, Growth and Poverty:
Elections in India 1952-98, Mittal Publishing, of Class Mobilisation and State Intervention: An Analysis of Interlinkages in Rural India,
New Delhi. Industrial Workers in Kerala, India', World EPTD Discussion Papers No 33, Environment
Berman, Peter (1996): 'Health Care ExpendituresDevelopment 24(6): 1055-71. and Production Technology Division, IFPRI,
in India' in Dasgupta et al, (eds). International Institute of Population Sciences March.
Banerjee, Debdas (1997): The Political Economy(1995): National Family Health Survey: IndiaShiva, Kumar A K (1995): 'Women's Capabilities
of Imbalances Across Indian States: Some1992-93, IIPS, Mumbai. and InfantMortality: Lessons from Manipur'
Observations on 50 Years of Independence,Klitgaard, Robert and Johannes Fedderke (1995): in Dasgupta et a:, (eds).
Occasional Papers No 166, Centre for Studies 'Social Integration and Disintegration: An Shiva, Kumar (1996): UNDP's Gender-related
in Social Sciences, Kolkata. Exploratory Analysis of Cross-Country Data', Development Index: A Computation for Indian
Butler, D, et al (1991): India Decides Elections World Development 23 (3): 357-69. States, Unicef, New Delhi.
1952-1991, Living Media, New Delhi. Knack, Stephen and Keefer Philip (1996): 'Does Singh, V B and S Bose (1984): Elections in India
Chhibber, P (1995): 'Political Parties, Electoral Social Capital have An Economic Pay-off? Data Handbook on Lok Sabha Elections 1952-
Competition, Government Expenditures and A Cross-Country Investigation IRIS Working 80, Sage Publications, New Delhi.
Economic Reform in India', Journal of Paper No 197, University of Mariland, College Singh, V B (1994): Elections in India Volume 2:
Development Studies, Vol 32, No 1. Park. Data Handbook on Lok Sabha Elections
Caldwell, John (1986): 'Routes to Low Mortality Kohli, Atul (1987): The State and Poverty in 1986-91, Sage Publications, New Delhi.
in Poor Countries', Population and Develop- India: Politics of Reform, Cambridge Sopher, D (1980): An Exploration of Indi
mnent Review, 12(2): 171-220. University Press, New York. Geographical Perspectives on Culture an
Coleman, James S (1990): Foundations of Social - (1994): 'Centralisation and Powerlessness: Society, Longman, London.
Theory, Harvard University Press. Cambridge, India's Democracy in a Comparative Sundar, Ramamani (1995): 'Household Survey
Mass. Perspective' in Migdal Joel S et al, (eds), State Health Care Utilisation and Expenditures
Dasgupta, Monica, Lincoln C Chen and Power and Social Forces, Cambridge NCAER Working Paper No 53, Nationa
T N Krishnan (eds) (1995): Women's Health University Press. Council of Applied Economic Research, Ne
in India Risk and Vulnerability, Oxford Krishnan, T N (1998): 'The Route to Social Delhi.
University Press, Mumbai. Development in Kerala: Social Intermediation Toye, John (1998): The Concept of Social Capital,
- (1997): Health. Poverty and Development in and Public Action' in Mehrotra Santosh and Institute of Development Studies, University
India, Oxford University Press, Delhi. Jolly Richard, (eds) Development with a of Sussex, England.
Dasgupta, Partha (1998): 'Economic Development Human Face, Clarendon Press, Oxford. Warren (1985): Halsteadetal (eds) Good Health at
and the Idea of Social Capital', Paper presentedManor, James (1996): 'Ethnicity and Politics in Low Cost, Rockfeller Foundation, New York.
at a seminar at the Faculty of Social and India'; International Affairs, 72 (3): 459-75.
Weiner, Myron (1968): 'Political Development in
Political Sciences, University of Cambridge,Minhas, B S (1991) 'Education Deprivation and the Indian States' in Weiner, M, editor, State
2/2/1998. Its Role as a Spoiler of Access to Better Life Politics in India, Princeton University Press,
Devi, Radha D (1993): 'Status of Women in India: in India' in Dutta A and Agrawal M M, (eds) Princeton.
A Comparison by State', Asia-Pacific The Quality of Life, Indian Institute of
- (1989): The Indian Paradox. Essays in Indian
Population Journal, 8 (4): 59-76. Advanced Studies, Shimla. Politics, edited by Ashutosh Varshney, Sage,
Dreze, Jean and Amartya Sen (1989): Hunger and Morris, M (1998): Social Capital and Poverty in New Delhi.
Public Action, Wider Studies in Development India, IDS Working Paper 61, Institute of - (1991): The Child and the State in India,
Economics, Clarendon Press, Oxford. Development Studies, Brighton. Oxford University Press, Delhi.
-(1995): India: Economic Development andSocial Nagarj, Rayaprolu, Aristomene Varoudakis and Woolcock, Michael (1998): 'Social Capital and
Opportunity, Oxford University Press, Marie-Anne Veganzones (1998): Long-run Economic Development: Toward a Theoretical
New Delhi. Growth Trends and Convergence across Synthesis and Policy Framework', Theory and
- (eds) (1997): Indian Development. Selected Society, 27:151-208.
Indian States Technical Papers No 133, OECD

704 Economic and Political Weekly February 24, 2001

This content downloaded from 203.200.225.155 on Sat, 14 Oct 2017 08:26:30 UTC
All use subject to http://about.jstor.org/terms