The Gradient of Governance: Distance and Disengagement in Indian Villages



Download 313.25 Kb.
Page1/4
Date conversion08.01.2017
Size313.25 Kb.
  1   2   3   4
The Gradient of Governance:

Distance and Disengagement in Indian Villages
Anirudh Krishna

Professor of Public Policy and Political Science, Duke University

Durham, NC 27708, USA

(919) 960 4658 (Home)

(919) 613 7337 (Office)

ak30@duke.edu


and

Gregory Schober

PhD Candidate, Department of Political Science, Duke University

gss14@duke.edu



Abstract

National governance scores must be seen in light of large within-country variance. Not only being a rural village, but being located at a substantial distance from cities, has great importance for good governance. Analysis of household data from different parts of rural India shows how villages at greater distances to towns tend to have lower scores on multiple governance dimensions. Even after controlling for diverse influences, using both ordinary least-square and multi-level regression models, this gradient of governance remains significant, imposing a dual penalty. Already penalised by markets, which have disproportionately rewarded urban and peri-urban areas, residents of villages located further from towns also experience and expect to receive worse treatment from government.

Many in the policy community are convinced that the quality of governance in a country exerts a (if not the) principal influence upon its economic development performance.1 Countries’ past performances or their promises of future improvements in governance quality constitute a primary criterion for aid allocation, for instance, in the United States government’s Millennium Challenge Account.2 Yet, a number of doubts surround both the good governance hypothesis and the particular measures of governance upon which tests of this hypothesis have been based.3

As defined by the World Bank, governance refers to “the traditions and institutions by which authority in a country is exercised,” including “the processes by which governments are selected, monitored, and replaced; the capacity of the government to effectively formulate and implement sound policies; and the respect of citizens and the state for the institutions that govern social and economic interactions among them” (Kaufmann, Kraay and Zoido-Lobatón 1999). More generally, governance is defined in the Oxford English Dictionary as “The action or manner of governing; controlling, directing, or regulating influence; control, sway, mastery; the office, function, or power of governing; authority or permission to govern.”

Whether one considers the more general definition or the specific one proposed by the World Bank, three aspects merit closer attention. First, governance is – at least partly, if not entirely – an outcome of state-society interactions and cannot be associated uniquely with any given set of institutional designs (Andrews 2010). Institutions certainly matter, but citizens’ consent matters additionally (Portes 2006): what citizens do (and not do) affects the quality of governance as much as what governments plan and initiate. The best-designed government institutions can produce quite indifferent results in situations where citizens do not regard them as legitimate and accessible (Levi 2006). Reversing this logic, types of government institutions that few current-day proponents of good governance would regard as ideal have on occasion helped deliver quite outstanding results, implementing responsive and far-sighted policies, and helping promote equitable and high-speed economic growth (Goldsmith 2007; Rodrik 2007; Root 2006).



Second, national or macro-level institutions are not all that matter. For most citizens, interactions with the state take place much closer to home. Their perceptions and their experiences are more likely to result from interactions with government agencies at the municipal and provincial levels. In order to understand better why some particular governance outcomes are given rise, the locus of inquiry has to shift closer to the venues where interactions occur between the state and different citizen groups.

Closely related is a point about informal institutions. Not all authority is exercised by formal state institutions. Everywhere, but more particularly in countries where the central state lacks reach and penetration, effective authority in some arenas is exercised by non-state actors and informal institutions (Azari and Smith 2012; Cammett and Maclean 2011; Helmke and Levitsky 2004).



Third, governance outcomes can vary widely across different parts of the same country (Gingerich 2013; Grindle 2004). One group of citizens that interacts primarily with one level of government and some types of agencies can report feeling very satisfied. Simultaneously, another group of citizens may be very dissatisfied.

While inter-personal and inter-group differences in governance experiences are likely to arise within every country, there are reasons to believe that these differences will be larger in developing countries and in others where state-building remains incomplete. The quality of governance in such countries will be more accurately reflected by considering both the mean national score and the variance in governance outcomes across individuals and groups. If universalism is a fundamental underlying principle of good governance, as it should be (Levi 2006; Rothstein 2011; Rothstein and Teorell 2008), then it follows that reducing the variance in governance experiences is as important as raising the national average score.


The Growing Importance of Location

Individuals’ abilities to obtain better governance outcomes might vary on account of differences in income, wealth, education, age, and gender and other individual- and community-level factors, reviewed below. Differences could also arise across regions and among diverse religious and ethnic groups. As countries have become increasingly enmeshed in global economic flows, however, another source of inequality has begun to rear its head. Spatial differences, specifically, differences between cities and the countryside, and within the countryside, between communities located closer to cities and others located further away, have started becoming larger, configuring inequality in different ways.

Alongside advancing globalization, economic opportunity is increasingly concentrated within cities. A “combination of spatial dispersion and global integration has created a new strategic role for major cities… These changes in the functioning of cities have had a massive impact… Cities concentrate control over vast resources” (Sassen 2001: 3). Economic activity is becoming concentrated within city-based “clusters of highly specialised skills and knowledge, institutions, rivals, related businesses, and sophisticated customers in a particular nation or region. Proximity in geographic, cultural, and institutional terms allows special access, special relationships, better information, powerful incentives, and other advantages in productivity and growth that are difficult to tap from a distance” (Porter 2000: 32; emphasis added).

The effects of living at a distance from a city or town are experienced in terms of differences in economic opportunity. While larger cities advance economically, remote rural communities lag further behind.

Remarking upon the “spiky” nature of current-day economic growth, Florida (2008: 19) notes how “the tallest spikes – the cities and regions [concentrated around cities] – are growing ever higher, while the valleys…mostly languish.” Such spatial clustering of economic opportunity has become acute within many parts of the developing world. China’s remarkable economic growth, for example, is “a result of only a handful of…spiky centers …each of which is a world apart from its vast impoverished rural areas… In 2006, average household incomes in urban China were two and a half times those in rural areas… The prospects for bridging these gaps are weak… But all that pales in comparison with the growing pains felt by India’s poor. India’s growing economic spikes – city regions such as Bangalore, Hyderabad, Mumbai, and parts of New Delhi – are also pulling away from the rest of that crowded country” (Florida 2008: 35-36). 4

Analysts examining the rise of inequality in India have noted how income differentials are widening between urban areas (which still account for no more than 30 percent of the country’s population), and rural areas, where almost 70 percent of all Indians continue to live.5 The biggest Indian towns have the largest concentrations of productive and household assets.6 Regardless of one’s level of education or training, earnings are higher if one lives within a large compared to small town and in towns compared to a rural villages (Shukla 2010). These spatial economic differences have intensified in recent years. The largest cities have experienced the greatest economic gains; people in smaller towns have also benefited but not as much; but in rural areas, especially in more remote villages, located more than five or ten kilometers from the nearest town, inflation-adjusted per capita incomes have fallen on average during a period of rapid national growth, 1993-2005 (Krishna and Bajpai 2011).

Falling behind economically, and seeing evidence for it in 24x7 television images, residents of rural communities look to governments for support, but to what extent are experiences of governance also skewed against more remote rural locations? Paralleling the manifestations of growing economic inequality, if the quality of governance, exemplified in people’s everyday experiences and expectations, is also spatially differentiated – being better within and close to India’s cities and worse in communities located at greater distances from towns – then instead of ameliorating the growing spatial differences in economic opportunity, the activities of government may actually intensify and deepen differences, producing multiple and cumulative disabilities.

The gradient of governance – the extent of slippage in governance outcomes between more central and more remote locations – is therefore important to investigate. We report below results from investigations undertaken in rural parts of two Indian states, Karnataka and Andhra Pradesh, both of which are close to the average in India in terms of literacy, life expectancy, and other socio-economic indicators, identifying the factors, both individual-level (such as social group, gender, wealth, information, and education) and location-based that associate significantly with high and low governance scores. Individuals’ experiences with state as well as national governments are examined. The impacts made by different agencies, state- and society-based, formal and informal – such as political parties, panchayats (rural local governments), caste-based and religious associations – are considered.

Both parts of the investigation show how, even after controlling for other salient factors, distance to town matters significantly. Communities located at greater distances from the nearest government center or town have significantly inferior governance results.
Measuring and Explaining Governance Outcomes

Diverse indicators have been developed for assessing governance outcomes. Composed of different components, these indices reflect divergent understandings of the concept (Apaza 2009; Arndt and Oman 2006; Knoll and Zloczysti 2012; Rothstein 2011). While some indices look at aspects of democratic process, others speak to different concerns, such as the climate for private investment and the degree of market openness. These sets of desiderata continue to grow, dangerously stretching the concept while simultaneously overwhelming the capacity of national governments to cope with an ever-expanding list of donor demands (Grindle 2004).

The most widely used indices were developed by the World Bank governance team. Six broad areas of concern have been defined, including voice and accountability; political stability and absence of violence; government effectiveness in service delivery; regulatory quality; rule of law; and control of corruption. For each of these areas, a separate index has been constructed. Multiple sources of data are aggregated in order to arrive at individual countries’ scores on each separate index (Kaufmann, Kraay, and Mastruzzi 2008).

Critics have alleged, however, that the seemingly separate indices of the World Bank are structurally related to one another, reflecting, in reality, a single underlying dimension (Langbein and Knack 2010). Going further, it is argued that the World Bank’s calculations of countries’ scores on its governance indices are heavily biased toward the perspectives of business elites (Kurtz and Schrank 2007).

As a matter of fact, because of gaps in the data and because of how these indices are practically constructed, ordinary citizen’s voices count for relatively little and experts’ and elites’ opinions are given the greatest weight (Brewer, et al. 2007; Treisman 2007).7 Within the World Bank’s 2007 index on Voice and Accountability, for instance, ordinary citizen’s voices carry only three percent of the total weight. Similarly, ordinary citizens’ opinions count for only six percent of the total weight in the government effectiveness index, five percent in the rule of law index, three percent in the control of corruption index, and not at all in the other two indices related, respectively, to regulatory quality and to political stability and absence of violence.8

Other existing databases can help examine differences across individuals and groups in some parts of the world, notably Africa and the Americas, but such rich data are not available or easy to access for other regions, one reason why within-country variances have not been calculated and made public. The scant prior work that has examined different citizens’ experiences has relied upon collecting primary data afresh.9

For the data collection exercises associated with this research project, two separate investigations were undertaken, a smaller one, carried out in 2011 in 12 villages of two districts of the state of Karnataka, and separately, a larger investigation undertaken a few years previously in 33 villages of three districts of the adjoining state of Andhra Pradesh. The more recent investigation, in Karnataka, was carried out with the specific intent of examining the gradient of governance. The earlier investigation in Andhra Pradesh was designed to look at poverty and democratic participation, and we consulted these closely comparable data in order to verify the conclusions derived from the Karnataka investigation.

Villages within two districts of Karnataka (Gulbarga and Raichur) were selected after being initially categorized into four groups based on distance from the nearest government center (defined below): within 2 km; from 2 to 5 km; from 5 to 10 km; and beyond 10 km. Random sampling helped select 12 villages, four from each of these distance categories. Twenty percent of all households within each of these villages were selected after taking a census of all village households and undertaking random sampling. A total of 772 households selected in this manner were interviewed using a pretested questionnaire. Discussions were also held with focus groups within each village.

In the Andhra Pradesh inquiry, three districts – Khammam, East Godavari, and Nalgonda – were selected that represent high, medium, and low points within this state in terms of diverse socioeconomic indicators. We selected 11 villages in each district that represent a diverse mix of population categories, economic activity, and distance to nearest government facility. A total of 1,295 individuals were selected for interviews following a similar process of random sampling.

We focus below on five key issue areas related to governance. The first three of these areas are related, as in the World Bank’s work, to voice and accountability; government effectiveness (in the provision of basic services); and control of corruption. Our fourth issue area combines two World Bank areas: absence of violence and rule of law. We left aside the remaining area related to regulatory quality, particularly since we interviewed individuals and not business enterprises. Instead, we considered a different aspect: trust in government institutions (local, state and national).

Table 1 displays the survey questions that went into the construction of the separate indices corresponding to each of these five dimensions of governance. Individual scores in relation to each dimension (or index) were calculated by aggregating their scores on individual survey questions, giving each question equal weight. For ease of interpretation, these scores were normalised to range from zero to 100, with higher scores representing better experiences of governance. These indices constitute the dependent variables for the analyses that follow. Other related survey questions, also pertaining to aspects of governance, produced results similar to the ones discussed below, suggesting that these results are robust to alternative constructions of our governance indices.
-- Table 1 about here --

Among individual-level independent variables, we considered age, education, gender, and relative wealth. Social group is important in these rural Indian contexts, and we examined differences among individuals belonging to Scheduled Castes (SC), Scheduled Tribes (ST), and Other Backward Castes (OBC) – large social groupings that have been historically oppressed and which remain relatively poor.10 We also looked at specific castes that have been dominant, numerically and in terms of economic and political power, within particular villages (Srinivas 1987). Prior analyses of governance at the individual level have pointed to the effects, separately from education, of access to information sources (e.g., Cramer and Kaufman 2011; Kaufmann, Montorial-Garriga, and Recanatini 2008). We also examined the effects of this variable. Separately, we looked at how access to political parties, village panchayats (explained below), and other local agencies influences individuals’ and communities’ governance scores.

Towns are loci both of commerce and economic opportunity, and more to the point for this paper, also of government offices and public decision making. We measured distance to town in different ways, best explained by referring briefly to the structures of public administration in rural India.

Public administration in rural India is organised into districts, sub-divisions, and talukas (also called tahsils), of which the first and third units have the greatest historical and contemporary importance (Potter 1996). At the time of writing, India was divided into 642 districts, of which there are 23 in Andhra Pradesh and 30 in Karnataka. Districts are further divided into tahsils or talukas, which functioned in colonial times as the basic units of land administration and revenue collection. Since land was (and still is, albeit to a lower extent) the principal income-producing asset in villages, the taluka acquired great importance for villagers, particularly since other important government offices – police stations, public health centers, high schools, and since independence, also block development offices – are most often co-located at taluka headquarters. Encounters with the government have typically involved for most villagers a trip to the taluka headquarters and progressively rarely to district, state, and national capitals. Even so, these journeys have often entailed considerable distances.

A large network of institutions supporting rural development programs has also been built up in the period after national independence, spreading particularly vigorously over the previous three decades. Collectively referred to as panchayati raj (or rural local self-government) institutions, they are organised, respectively, at the village level (gram panchayat, with jurisdiction over anywhere between one to five villages, depending upon population size) and the district level (zila parishad), with an intermediate layer in some states, the “block-level,” very often coterminous with taluka jurisdictions.

Below we examine the effects separately of distance to District and Taluka HQ. Separately, we looked at distance to gram panchayat HQ, finding this variable not significant for the analysis of governance scores, a likely explanation for which we will offer below.




Exploring the Effects of Distance in Karnataka

Table 2 presents village scores (calculated as an un-weighted average of individual scores) related to each of the five dimensions of governance examined here. Higher scores represent superior governance outcomes. Since individuals within each village were selected using random sampling, the average of individual scores provides us with an unbiased estimate of village scores. Villages are organised by increasing distance to nearest town. In nine of 12 cases this distance is identical with distance to Taluka HQ; hardly surprising, since longstanding centers of public administration either became or were always centers of commerce to boot (Bayly 1983) – a point that is important to bear in mind while interpreting the results presented below.

We constructed similar tables and graphs using other distance measures – including distance to District HQ and distance to Panchayat HQ – but in none of these cases was the gradient of governance as clear. As Table 2 shows, except for one dimension (Absence of Violence), village governance scores tend to decrease with greater distance to Tahsil HQ (or what is nearly the same thing, with distance to the nearest town).
-- Table 2 about here --
The large differences in governance scores seen in Table 2 are not merely the product of statistical artifice. They represent something real and experienced, as revealed by our individual interviews and focus group discussions. As illustration, consider two villages Arejambga (21 km from the nearest town and Taluka HQ) and Karidigudda (only 2 km from the nearest town), which respectively have lower-than-average and higher-than-average scores on all five governance indicators. In Arejambga, the average respondent reported paying a bribe three times in the past year and more than half of all respondents felt that all government officials were corrupt. Arejambga residents commonly felt that it was very hard to obtain information about a government program and that a politician would not be responsive to their communication efforts. In Karadigudda, on the other hand, the average respondent did not pay a bribe in the year preceding the survey and fewer than ten percent of respondents believed that all or many government officials were involved in corruption. People here find it easy to obtain information about government programs, and they expect politicians to be responsive at least some of the time. In other near and more distant villages similarly encouraging and discouraging experiences and expectations, respectively, were narrated.

Graph 1 visually depicts these gradients of governance, showing how – no matter which dimension of governance is considered – governance scores tend to be lower among villages located at greater distances from towns.


-- Graph 1 about here --
Notice how these governance scores tend to be variously high or low for villages located closer to Taluka HQ, but tend to diminish and become almost uniformly lower in further-away villages. Explaining these village differences better requires looking simultaneously at multiple influences, including distance to district HQ and distance to paved road. We describe briefly how each of these independent variables was constructed.

Individual- (or Household)-Level Variables

Differences in wealth have been identified, following Lipset (1960), by a large swathe of the literature as a likely reason for differences in civic engagement and political participation.11 We examined two separate variables.



Asset Index: Respondents were asked whether they possessed each of the following six assets: bicycle, motorcycle or scooter, car, TV, radio, and mobile phone. The simplest asset index was calculated as the total number of assets possessed. (Mean = 2.05, SD = 1.24). Alternative index constructions, weighting each asset by its average market price – as well as other measures of relative wealth, such as land holdings or quality of home construction (brick v. mud-wattle) – were highly correlated with this simple index.

Asset Change: Change in Asset Index during the past 10 years. (Mean = 2.7, SD = 0.51).

Education can matter separately from wealth, as other strands of the literature have pointed out (Finkel 2002; Jackson 1995; Nie, et al. 1996; Seligson, et al. 1995). Prior analyses of governance at the individual level have pointed to the effects, separately from education, of access to information sources (e.g., Cramer and Kaufman 2011; Kaufmann, et al. 2008). Gender has been found to be important in some analyses (Delli Karpini and Keeter 1996; Verba, et al. 1995). We operationalised these variables in the following ways.



Education: Highest level in the household, scaled as 0= illiterate; 1= 1-4 years; 2= 5-7 years; 3= 8-10 years; 4= Higher Secondary or PUC; 5= College degree; 6= Post graduation (Mean = 2.8, SD = 1.38).

Access to Information: Individuals were asked about which among the following eight information sources they consulted regularly, i.e., at least once a month: household members, neighbors, village leaders, government officials, radio, television, newspaper, and village assembly. The score for each individual was computed by aggregating all Yes responses. (Mean = 3.6, SD = 0.51)

Gender: 1=Female; 0=Male

Social group is important in these rural Indian contexts (Jaffrelot 2003), and we examined differences among individuals belonging to Scheduled Castes (SC), Scheduled Tribes (ST), and Other Backward Castes (OBC) – large grouping which have been historically oppressed and remain relatively poor. We also looked at specific social groups that have been dominant, numerically and in terms of economic and political power, within particular villages (Srinivas 1987).



Social group and Religion: In addition to the three broad social groups listed above, we also considered two specific agriculturist caste groups (Lingayat and Kuruba), both classified officially as OBC,12 which are dominant in these villages. Overall, in all 12 villages, the population shares of different groups were as follows: SC 18%; ST 19%; Lingayat 24%; Kuruba 11%; and Muslim 7%. Zero-one dummy variables were constructed for each of these groups. High-caste Hindus serves as the comparison category. Apart from Hindus and Muslims no other religions are represented in any significant numbers.

Access to government agencies may be mediated through party officials or other intermediaries, and the nature of the mediating agency might result in significantly different governance experiences (Krishna 2002).



Interaction with gram panchayat or political parties: Respondents were asked about their frequency of interaction with each of these institutions, particularly for obtaining assistance vis-à-vis government services and program benefits. Responses were scaled from zero to five: never (coded 0), rarely (once a year or less often), occasionally (once every three or four months), often (once a month); very often (weekly); and all the time (coded 5). Whether considered individually or combined, the frequency of interaction with these agencies was quite low; few villagers consult with them for obtaining assistance (Mean = 1.5, SD = 1.2).

Village-Level Variables

Distance to Taluka HQ and Distance to District HQ were measured in kilometers. Based on focus group discussions, which revealed the likely relevance of this additional variable, we also considered Distance to Paved Road, which ranges from 0.5 to 14 kilometers.

Population: Based on our village censuses.

We estimated multivariate linear regression models to analyze the effects of the independent variables on governance. Table 3 reports the results of ordinary least-square (OLS) regressions with each separate governance indicator considered as a dependent variable. Given the large number of response categories in each dependent variable, it was determined that the linear regression model is the appropriate model choice (DeMaris 1992: 77-78). We also ran multilevel linear regression models, results from which were largely similar to the ones reported below in terms of which independent variables gained significance. For the sake of simplicity, we report only the OLS results.13


-- Table 3 about here --
Wealth (assessed by Asset Index) matters for only one of five dimensions of governance (Voice and Accountability) and is not significant for the other four dimensions. Asset Change, i.e., an increase in household assets over the previous ten years, matters for one other dimension (Trust in Government Institutions) but not for the other four dimensions.

Surprisingly, education is not significant for any of the five dimensions, probably a reflection of the overall low levels of education in these villages. However, access to information is significantly related to three of five dimensions, a point we will revisit below.

Among different social group- and religion-related variables, only two attain significance and that, too, in relation to only one of five governance dimensions each. Caste Lingayat is significantly and positively related to Voice and Accountability. Scheduled Tribes (ST) is significant in relation to Trust in Government Institutions. Similarly, the variable for Muslims is significant for only one of five dimensions. The variable for Gender is not significant in relation to any of the five indicators of good governance.

Frequency of contact with political parties or elected local councils (panchayats) is negatively associated with two of five governance dimensions: people who contact these institutions more often tend to experience worse governance outcomes. In alternative specifications of the regression models, we split this variable, examining contact with political parties and separately with panchayats. However, these results did not change.

Making sense of this particular result requires a little more background. Political parties are rarely represented in Indian villages, being organised more often at district and less often at taluka levels (Kohli 1990), part of the top-down structures that help give distance greater salience in these contexts. Panchayats represent perhaps the best hope of establishing formal institutionalised state presence at the grassroots, but these bodies have so far functioned mainly as implementing agencies for programs of the central and state governments. Few villagers consult with either political parties or panchayats in relation to their everyday demands upon government.14 Rather, their interactions are mediated, especially in the case of more distant villages, by non-formal middlemen and political brokers, a point we will revisit below.

Coming to village-level variables we find that Distance to Nearest Paved Road is significantly and negatively associated with three of five governance dimensions. Villages located further away from roads on which cars and buses ply tend to have lower scores on Voice and Accountability, Absence of Violence, and Trust in Government Institutions. Distance to District HQ is significant for two dimensions, being significantly related, first, to Absence of Violence, albeit with the wrong sign: villages located at greater distances from district headquarters tend to have more absence of violence. In substantive terms, the size of this effect is, however, quite small. Second, Trust in Government Institutions is significantly lower in villages located further from district headquarters, although, once again, these effects are small in size. Distance to Taluka HQ is significantly associated with four dimensions of governance, showing how, even after controlling for diverse influences, the quality of governance becomes progressively worse at greater distances from towns (which are coterminous with nine of 12 Taluka HQ and adjacent in the other three cases). For all except one dimension of governance – Absence of Violence, which seems hardest to explain – Distance to Taluka HQ is consistently and strongly significant and has the expected negative sign.

In sum, two variables – Distance to Taluka HQ (the nearest government center) and Access to Information – have the greatest purchase on individuals’ governance experiences. Distance to Taluka HQ (in most cases also the nearest town) is significantly and negatively associated with four of five governance dimensions. Access to information is significantly associated with three of five indicators: Voice and Accountability, Trust in Government Institutions, and Absence of Corruption. Distance to paved road matters as well in relation to two dimensions of governance.

We confirmed these results by examining data compiled separately in villages of two other Indian states, Andhra Pradesh and Rajasthan. In each case, the same result was corroborated: villages at greater distances from towns have significantly poorer governance outcomes. For the sake of brevity, and because these results are similar, we reproduce below only the analysis from Andhra Pradesh.

  1   2   3   4


The database is protected by copyright ©ininet.org 2016
send message

    Main page