How effective are gram sabhas?

Village panchayats across India are mandated to organise local public meetings called gram sabhas several times a year, wherein villagers discuss issues such as local public good provision. This column finds that gram sabhas are indeed effective in altering the composition of local public goods provided. Promoting the institution of gram sabha can make policymaking more sensitive to the preferences of discriminated groups such as women.

Many countries around the world, including India, have institutionalised local public meetings in jurisdictions such as villages or towns, where citizens gather on a regular basis to deliberate on important public issues, such as local public good provision¹.

In India these local meetings are called gram sabhas and are organised by every village panchayat multiple times in a year, as mandated by the Panchayati Raj Acts in the different states. There are about 250,000 village panchayats in India, and hence, several hundreds of thousands of such meetings are conducted every year in the country. At these local meetings, villagers gather to deliberate, debate, and press for provisioning of their preferred public goods - be it construction or repair of local roads, or irrigation canals, or provision of hand pumps, or sanitation facilities. This is the largest exercise in deliberative democracy in the history of human civilisation. The Panchayati Raj Acts specify that public good projects that receive favourable support in the gram sabha should be incorporated while preparing the annual budgets of the village panchayats and the budget must be approved by the gram sabha before it is sent to the block panchayat for grant request. Hence, the Acts do provide some power to this institution to affect public good provision. Yet we do not have much idea about whether these deliberations actually result in any change in the provision of local public goods, or they are just “talking shops”² for people who enjoy group discussions. I attempt to address this issue in my paper.

Analysing the impact of gram sabhas on provision of local public goods

There is a literature that looks at how panchayat elections and identities (in terms of gender, caste or tribe) of the elected pradhans affect the provisioning of public goods in Indian villages. For example, Foster and Rosenzweig (2004) show that having elections makes public good provision more sensitive to the preferences of lower castes, while others (Bandopadhyay and Duflo (2004), Besley, Pande and Rao (2005, 2012), Dunning and Nilekani (2013), Gajwani and Zhang (2014)) show that reserving the pradhan election for women or caste/ ethnic minorities also affects the composition of public good provided. Given that the institution of gram sabha coexists with local elections, it is difficult to isolate the potential effect of gram sabha participation from the effect of local elections and elected representatives on public good provisioning.

The other challenge is availability of data. One would require individual-level attendance information in these meetings along with other information to conduct this sort of analysis. Though it is possible to look at the attendance records of the meetings kept at the respective village panchayat’s office to know about attendance rates, these records do not always reflect true attendance and may be problematic to use.

Fortunately, there is a dataset that largely meets all the data requirements³. This is the Rural Economic and Demographic Survey (REDS) dataset, collected during a national-level household survey conducted by the National Council of Applied Economic Research (NCAER) over 2006-2008⁴. The survey asks adult members of households about their participation in the past four gram sabhas. It also asks them to report their top three choices of local public goods where they would want the village panchayat to invest resources. This (partial) preference ranking provides an extremely good measure of individual preferences. Also, the survey contains information about the provision of various public goods carried out by the village panchayats. I use this data to examine the effectiveness of gram sabhas, that is, if attendance in gram sabhas causes any change in public good provision, in a way that is consistent with the public good preferences of those that attend gram sabhas.

The data reveals that, on average, 13% of villagers attend gram sabhas⁵⁶. However, the attendance rates differ quite widely by gender. While about 21% of men, on average, show up at the gram sabhas, the attendance rate is only 7% for women⁷. On the other hand, the difference in attendance rates across Scheduled Castes (SCs) and non-SCs, or Scheduled Tribes (STs) and non-STs are minimal⁸; the attendance rates of SCs and non-SCs (STs and non-STs) are 12% and 14% (17% and 13%), respectively. The data also shows that men and women have strong differences in their preferences over local public goods (this fact is consistent with the findings of previous research such as Chattopadhyay and Duflo (2004) and Coffey el al. (2014)). Women systematically rank water, sanitation and health higher relative to men, while men rank road construction, school provision and investment in public irrigation higher relative to women. These differences in preferences across gender are strong even within households. However, the differences in preferences across SCs and non-SCs, or STs and non-STs are surprisingly not that large, except for public irrigation, which is relatively less preferred by SCs and STs, presumably because they are more likely to be landless than the rest of the population. It is also interesting to note that there is no association between meeting attendance and the landholding of the households, implying that the participation rates across rich and poor households are similar. This alleviates the concern that meeting attendance is primarily driven by incentives of the poorer households to include themselves in the beneficiary lists of different targeted government welfare schemes, which is one of the regular agendas of the gram sabhas. This is consistent with the finding by Ghatak and Ghatak (2002) who survey a sample of villages from West Bengal and report that discussions about the beneficiary list receive less importance than the deliberations about public goods projects.

Since the preference differences across gender are stark, I use men and women as the relevant groups in the population, and I examine whether higher attendance of women relative to men in the meeting results in a greater provision of public goods that women prefer more and lower provision of the ones that men prefer more. The idea is to conceptualise gram sabha as an institution that helps population groups to constrain the policy choice of the elected policymaker after he or she is elected to power. This may be necessary because elections may fail to aggregate preferences perfectly when individuals care about more than one issue⁹

However, the attendance of men vs. women may also be affected by the gender of the pradhan. This is because the incentive to attend meetings for both groups are different under men and women pradhans as, conditional on attendance, pradhans can bias the outcomes of meetings in favour of his/ her group¹⁰. Thus, the gender of the pradhan can directly influence both relative attendances of the two groups as well as the composition of public goods provided. The econometric methodology of my research is able to estimate the causal effect of relative attendance of women to men in meetings on the provision of different public goods, over and above the potential direct effect of the gender of pradhan/ reservation of pradhan chairs¹¹.

Gram sabhas can alter composition of local public goods provided by village panchayat

I find that gram sabhas are indeed effective in altering the composition of local public goods provided by the village panchayat. One standard deviation (s.d.)¹² increase in the relative attendance of women to men at the gram sabha increases the provision of sanitation and health facilities by 0.5 and 0.6 s.d., respectively, and reduces the provision of road construction by 1 s.d. This implies that if women attendance were to be reduced to zero, keeping men’s attendance the same, it would have reduced sanitation construction by about 25% and increased road construction by about 33%. This, however, assumes that individuals would not change their voting behaviour when women’s attendance in meeting is reduced. Therefore, this is probably an overestimate of the meeting effect.

The estimated effects are reasonably large, considering the fact that those public goods are three out of the four most demanded public goods in the sample. One possible reason for finding this large effect is that attendance of women on average is much lower than men, and hence the effect of women’s attendance (relative to men’s) at the margin is high. This may seem puzzling, since if the marginal effect of women’s attendance is high at the observed level of attendance, more women should attend gram sabhas. One reason for this is that women presumably face a much higher fixed cost of attending public platforms such as gram sabha meetings.

Promoting gram sabhas can enable greater political participation of women

My findings suggest that the ability of discriminated groups, such as women, to directly affect policy is underestimated if we only focus on their representation in electoral politics. Hence, an alternative suggestion to promote greater political participation of women would be to strengthen the institution of gram sabha. The state governments can begin by laying out better laws that make the meeting procedures and the exact powers of gram sabha more precise. This presumably would make the enforcement of meeting outcomes easier for the states. Also, information campaigns promoting the legal power and importance of this institution may go a long way in making policymaking more sensitive to women’s preferences and, hence, more democratic.

Notes:

The history of such regular open public meetings goes as far back as the inception of democracy itself, with the citizen assemblies in ancient Greek city-states around 500 BC. The New England region in US and Switzerland continue to have local meetings in their towns/ cantons. Many Latin American countries like Brazil have more recently adopted the institution of local meetings (known as participatory budgeting) in their municipalities.
This phrase has been borrowed from Ban, Jha and Rao (2013).
To be best of my knowledge, this is the only dataset that is available with all the necessary information (individual-level information on both meeting attendance and public good preference, and village-level public good provision and pradhan’s reservation status) required for this analysis.
This is the latest round of the REDS that NCAER has been conducting since 1968. Although it is primarily meant for collecting information on agricultural practices of farmers in rural India, the latest round asked questions about political participation, including participation in the gram sabhas.
All the results reported here relate to the adult (18 years or above) population in villages, since they constitute the electorate.
The meeting attendance rates in other societies bear striking similarity to the Indian figure. The attendance rate in Swedish town meetings (known as kommunalstämma) during 1919-1938 was 14%, in Swiss town meetings (known as Landsgemeinde) in 1988 was 18%, in Massachusetts, US town meetings (known as Town Halls) in 1996 was 12%.
These statistics match very well with the attendance rates calculated with a few other publicly available survey datasets on India that contain individual-level gram sabha attendance information, and hence provide confidence in the validity of the self-reported attendance behaviour of the survey respondents.
SCs/ STs are historically discriminated caste and ethnic minorities in India.
This relates to the well-known result in political economy about the failure of the median voter theorem in the presence of multidimensional electoral platform – that is, the “median” person’s preferences are not chosen as policy when people care about multiple issues.
This can happen if pradhan enjoys some control over the meeting procedures, such as agenda setting, deciding duration of deliberation on specific agendas etc. Given the way gram sabhas are conducted, this seems to be a reasonable assumption.
To estimate the causal effect of attendance of men vs. women on public good provision, I need some exogenous variation in the gender composition of attendees. I instrument the variation in the relative attendance of women to men with the interaction of the divergence in preference in gender across villages and a dummy variable indicating whether the pradhan position of the villages are reserved for women or not (Chattopadhyay and Duflo (2004) show that the way the pradhan elections of villages are reserved for women under the Women Reservation policy is effectively random). The assumption is that even though preference divergence and women reservation may have a direct effect on public good provision, the interaction of them would not. The interaction, however, affects the relative attendance of women. This is because, as preferences across gender diverge, the incentive to attend meetings goes up differentially in villages with men and women pradhans, as pradhans favour their own groups (as stated before).
S.d. is a measure of variation in the values of a variable.