Demographic Change, Natural Resources and Violent Conflict: The Case of India

Transcription

1 Demographic Change, Natural Resources and Violent Conflict: The Case of India Paper prepared for the IUSSP Seminar on the Demography of Conflict and Violence Oslo, Norway, 8-11 November 2003 Henrik Urdal Centre for the Study of Civil War (CSCW) The International Peace Research Institute, Oslo (PRIO) This is a paper in progress Please do not cite without permission Abstract: This paper attempts to go beyond cross-national studies of relationships between population pressure and violent political conflict. Propositions that violent conflict may be associated with high population pressure on renewable natural resources, with youth bulges and with ethnic heterogeneity and differential growth rates between ethnic groups are tested in a time-series study of Indian regions for the period The endeavor has so far produced rather inconclusive results. A future avenue will be to use different data for violent conflict. So far no support is found for traditional neo-malthusian claims, or for the youth bulge hypothesis. Some support is found in favor of the proposition that differential growth rates are associated with conflict. This paper is in a very early stage, all results and conclusions are tentative.

2 1. Introduction Although there has been a marked decrease in the number of both interstate and internal armed conflicts since the end of the Cold War, the number of internal armed conflicts is still high. In 2002, 30 internal armed conflicts were active in 23 different countries worldwide (Eriksson et al., 2003). Economic development and environmental factors are frequently used to explain the occurrence of armed conflict. But conflict also sustains poverty in developing countries. Perpetual violent conflict is one of the greatest obstacles to economic and social development (Collier, 1999), and this in turn increases the vulnerability to new violent conflict. There has also been a rising concern about the detrimental effects that armed conflict has on the environment (Austin & Bruch, 2000). After the end of the Cold War, an increased interest has emerged in demographic factors as potential causes of armed conflict. The debate about demography and conflict originates from a position often referred to as neomalthusian. Neo-malthusian scholars voice concerns that rapidly growing populations, especially in developing countries, outpace the local natural resource base, eventually forcing groups to fight over resource access. While some of the contributions have been of the somewhat sensationalist kind and have clearly overstated the significance of population and environmental factors (Kaplan, 1994; Myers, 1993), more cautious approaches have examined the subtle mechanisms that may link population growth and environmental change to armed conflict (Bächler, 1999; Homer-Dixon & Blitt, 1998). More recently, a much broader debate on security implications of demographic factors has emerged (Goldstone, 2001; JIA, 2002; Weiner & Russel, 2001; Weiner & Teitelbaum, 2001). Common to these approaches is that they see demographic change as processes that fundamentally and continuously influence and change society, and they assume that such change may stimulate violent conflict under certain conditions. This paper derives from a larger effort to address and analyze demographic characteristics as potential causes of domestic armed conflict. Previous cross-national studies have focused on effects of neo-malthusian factors (Urdal, 2003) and of youth bulges (Urdal, 2002). These studies have found substantial support for the proposition that large youth bulges are associated with violent conflict, but not for neo-malthusian claims that high population growth rates or density relative to productive land are driving internal conflicts. However, cross-national studies using national aggregates to measure population pressure may not capture local population dynamics very well, processes that may be argued to cause local conflicts. This study is an attempt to go beyond national aggregates to see if regional population pressure may influence on the distribution of violent conflict between regions in India. The rationale for studying India is the great demographic variety, the extent of political violence and the general good data availability. 2. Population pressure and violent conflict: theoretical perspectives 2.1 Neo-malthusian population pressure Neo-malthusian scholars see population growth and density closely linked to the potential scarcity of renewable resources. They assume that resource scarcity leads to increased inter-group competition, and under unfavorable economic and political conditions, such competition can take the form of violent conflict. Poor countries are 1

3 seen as especially vulnerable to resource conflicts as they often lack abilities of adaptation. A major reference point in the debate over population pressure and resource scarcity is Thomas Homer-Dixon (1991; 1994; 1999; Homer-Dixon & Blitt 1998). Like other neo-malthusian scholars, Homer-Dixon sees population growth closely linked to the potential scarcity of renewable resources. He further argues that resource scarcity leads to increased inter-group competition, and under unfavourable economic and political conditions, such competition can take the form of violent conflict. Poor countries are especially vulnerable to resource conflicts as they often lack the ability to adapt to scarcity (Homer-Dixon, 1999: 181). Homer-Dixon distinguishes between three main sources of resource scarcity (Homer-Dixon & Blitt, 1998: 6) (Figure 1). Supply-induced scarcity results from degradation or depletion of natural resources. It simply becomes less of a resource as a result of non-sustainable use that does not allow the resource to regenerate. 1 Demand-induced scarcity is primarily caused by population growth. If a resource base is constant, the availability of resources per person will diminish with the increasing number of people that have to share it. Such scarcity can also arise from an increase in demand per capita. A third form is structural scarcity. This is a form of scarcity that only applies to certain groups that, relative to other groups, are excluded from equal access to particular resources. Such unequal social distribution of a resource does not presuppose actual scarcity if the resource was distributed evenly. While the traditional neo-malthusian focus has been on overall population growth and population density, Homer-Dixon points to a number of more specific forms of pressure that may trigger conflict. High population growth in marginalized rural areas, and increasing pressures on urban areas are likely hotspots for conflict (Homer-Dixon, 1999: 160). Resource scarcity may trigger migration which again may lead to identity conflicts (Homer-Dixon & Blitt, 1998: 9, see Figure 1). Figure 1 Some Sources and Consequences of Environmental Scarcity Sources of Environmental Scarcity Social Effects Decrease in Quality and Quantity of Renewable Resources Migration Expulsion Ethnic Conflicts Population Growth Increased Environmental Scarcity Weakened States Coups d état Unequal Resource Access Decreased Economic Productivity Deprivation Conflicts Source: Homer-Dixon (1994: 31). 1 Homer-Dixon focuses mainly on degradation of natural resources resulting from human activity. But natural resources can be both degraded and depleted from causes that are not human-induced, such as natural disasters or less dramatic natural variation. 2

4 The resource pessimist neo-malthusian perspective is challenged by a resource optimistic or cornucopian view. Cornucopians concede the neo-malthusians premise that more people means less resources per person. They believe, however, that an increased pressure on resources leads to innovation and implementation of new technology that make resource scarcity and resource dependency increasingly less likely. Population pressure is thus believed to be either a neutral factor among determinants of armed conflict, or even a possible contributor to economic growth that can reduce conflict propensity in the longer run (Boserup, 1981; Simon, 1989; Boserup & Schultz, 1990). Optimists also claim that population pressure on natural resources will be less of a problem in the future as world population growth is slowing down (Lomborg 2001: 45 49). Most of the empirical work on the population-resource-conflict nexus has been conducted through case studies. Many of these, including Homer-Dixon s own empirical research, have been criticized for methodological deficiencies related to the careful selection exclusively of cases where both resource scarcity and armed conflict are present (Gleditsch, 2001). More rigorous empirical research has so far found ambiguous evidence for a neo-malthusian conflict scenario. Both Hauge & Ellingsen (2001) and Esty et al. (1998) found some limited evidence for a connection between resource degradation and violent conflict at a national level. Previously I have found no support for hypotheses assuming that countries with high population growth or high population density relative to potentially productive land are more conflict prone (Urdal, 2003). The aim of this paper is to investigate whether neglected dimensions of the population-environment-conflict nexus may yield more support for the neomalthusian concern than previous cross-national research has provided. This goal will be pursued through the utilization of improved measures of population pressure on natural resources (cropland and freshwater) 2, the addition of new dimensions (migration and urbanization) and the investigation of population pressure on natural resources on a sub-national level. In addition to testing neo-malthusian claims of population pressure in the form of population growth and density, the size of the rural population, rural and urban growth, I will also include a measure of human development, literacy. According to Homer-Dixon, human knowledge ( ingenuity ) is the key to adaptability to resource scarcity. From this perspective one should expect that societies with greater human knowledge should be less susceptible to violent conflict, both because they are more capable of dealing with and overcoming scarcities, and because education decreases a persons dependence of the agricultural sector. The debate over resources and conflict has been divided into two quite distinct camps, between those who emphasize scarcity and grievance and those who primarily worry about resource abundance or greed. This divide has not been very helpful, as they too often have been promoted as opposite and competing views. Renner (2002) argues rightfully that the two perspectives are complimentary rather than mutually exclusive, and that scarcity and abundance may very well coexist. This paper will briefly address also the issue of resource abundance and conflict. 2 The project aims to develop an index of carrying capacity by using GIS tools and data on climate zones, soil quality, and freswater availability/rainfall. 3

5 Hypotheses: H 1 : Regions experiencing strong population growth are more likely to experience violent political conflict. H 2 : Regions with high population density are more likely to experience violent political conflict. H 3 : High population growth and high population density combined make regions particularly conflict prone. H 4a : Regions experiencing high rural population growth are more likely to experience violent political conflict. H 4b : Regions experiencing high urban population growth are more likely to experience violent political conflict. H 5 : Regions with a large rural population are more susceptible to violent political conflict. H 6 : Increased literacy reduces the likelihood of violent political conflict. H 7 : Regions with valuable mineral resources are more likely to experience violent political conflict. 2.2 Youth bulges Many developing countries have recently experienced a substantial mortality decline, while fertility has remained relatively high. This has resulted in particularly youthful populations in many countries, often referred to as youth bulges. The security concern over youth bulges is not primarily linked to natural resource scarcity, but rather to the strain they put on social institutions such as the labour market and the educational system, thereby causing grievances that may result in violent conflict (Choucri, 1974; Braungart, 1984; Goldstone, 2001). But some scholars have also argued that the existence of youth bulges may aggravate already existing conflicts over natural resources (Homer-Dixon, 1999: 166; Mesquida & Wiener, 1996). After 11 September 2001, the youth bulge hypothesis has been presented as an important explanation of international terrorism (Sciolino, 2001; Zakaria, 2001), and Samuel P. Huntington now claims that his clash of civilization argument is more about demographic factors in the Muslim world than about culture (Steinberger, 2001). The assumed relationship between youth bulges and armed conflict is illustrated in Figure 2. The model assumes that youth of large cohorts are likely to experience unemployment because they increase the supply of labour substantially when entering the labour market (Braungart, 1984: 16). Unemployment is believed to cause grievances, and especially so if expectations are raised through expansions in education (Choucri, 1974: 73; Weiner & Teitelbaum, 2001: 18). Particularly critical situations arise when youth bulges coincide with times of economic depression (Choucri, 1974: 73). A further assumption is that grievances arise if possibilities to influence the political system and attain elite positions are limited (Goldstone, 2001: 95). 4

6 Figure 2 Youth Bulges as a Source of Armed Conflict Youth Bulges Expansion of Education Lack of Employment Opportunities Youth Grievances Deprivation Conflicts Limited Political Recruitment Source: Urdal (2002). Previous quantitative work in this field is limited. Earlier approaches suffer either from unclear concepts of armed conflict (Esty et al., 1998) or from erroneous operationalization of youth bulges (Collier, 2000). Previously I have found substantial support for a link between youth bulges and the risk of armed conflict onset, and some support for the assumption that this relationship is strengthened during times of economic crises (Urdal, 2002). However, this field remains largely unexplored. This paper will draw on my previous research and eventually investigate more thoroughly under what conditions youth bulges may increase the risk of armed conflict (economic stagnation, underemployment, the role of education), and to test whether such hypotheses hold at the sub-national level. So far, I have not compiled data on economic growth, unemployment or education. Some of the contributions above focus on young male bulges rather than youth bulges, arguing that it is young males who are the primary actors of all kinds of violent conflict. In this paper I will be analyzing whether regions with male dominance in numbers are more conflict prone, and whether young male bulges make predominantly male regions particularly susceptible to conflict. Hypotheses: H 8a : Regions with large proportions of youth relative to the total adult population are more likely to experience violent political conflict. H 8b : Regions with large proportions of young men relative to the total adult male population are more likely to experience violent political conflict. H 9 : The larger the relative size of the male population, the stronger is the conflictinducing effect of large shares of young men. Ethnic heterogeneity and differential growth rates There are a number of rigorous studies investigating the relationship between ethnic competition 3 and armed conflict (Ellingsen, 2000; Henderson & Singer, 2000; Fearon & Laitin, 2001; Sambanis, 2001). A general finding of these studies is that ethnicity matters, although not as much as portrayed in popular media. But a largely neglected 3 Here, the term ethnic competition is used in a very crude way to describe societies and countries shared between two or more ethnic, religious or linguistic groups of some size. While not always true, important faultlines (economic, political, educational) tend to follow such communal divisions in heterogeneous countries. These faultlines may be sources of violent conflict. 5

7 aspect of these quantitative studies is the dynamics of inter-ethnic conflict; how the relative strength of groups changes over time. My interest in this subject derives from my work on Bosnia-Herzegovina (Urdal, 2001) through which I came across a number of statements by Bosnian Serb leaders over the fear of being outnumbered by Bosnian Muslims. Such claims were used to legitimize the demand for a separate Bosnian Serb state. Similar concerns are found in other heterogeneous societies. Krebs & Levy (2001: 82) note that the changing Christian-Muslim balance in Lebanon has been blamed for upsetting that country s finely tuned political system in the 1970s and throwing it into decades of turmoil. Bookman (1997) argues that political leaders attempt to increase the size of their ethnic group in order to gain or consolidate power, and that this undermines prospects for inter-group cooperation. She states that the overriding goal of the demographic struggle for power is to alter the relative size of ethnic groups, and since this goal is often achieved, the consociational solution suffers limitations in its applicability (Bookman, 1997: 204). Weiner & Teitelbaum (2001: 22) claim that over the past years, many episodes of violent conflict seem to be driven in part by competitive fears resulting from [ ] compositional shifts. Toft (2002) provides support for hypotheses stating that democratic states are more likely than autocratic states to be destabilized by differential growth, and that the greater the difference in growth and the closer in magnitude the two growing groups are, the greater is the risk that democratic states are destabilized. In this paper I will be looking at different forms of differential growth in order to investigate the claim that dynamic processes of ethnic identity may fuel conflicts. But unlike the assumed effects of overall, rural and urban growth rates on conflict, differential growth is assumed to influence conflict behavior by shaping the perception 4 of individuals and groups of being losers or winners of a demographic battle. As Horowitz (1985: 194, 196) notes: Numbers are an indicator of whose country it is [...] it is clear that a census needs to be won. So the election is a census, and the census is an election. Hypotheses: H 10 : The greater the religious heterogeneity, the greater the risk of violent political conflict. H 11 : Regions with a non-hindu majority are more likely to experience violent political conflict. H 12a : The greater the growth of the second largest religious group relative to that of the largest religious group in a region, the greater the risk of violent political conflict. H 12b : The greater the growth of the largest non-hindu group relative to that of the Hindu population in a region, the greater the risk of violent political conflict. H 12c : The greater the growth of the Muslim population relative to that of the Hindu population in a region, the greater the risk of violent political conflict. 4 Such an approach also allows for factors such as changes in census categories and changes in people s own perception of national/ethnic identity to influence the magnitude of differential growth. 6

8 H 13 : The greater the relative size of scheduled castes, the greater the risk of violent political conflict. H 14 : The greater the growth of scheduled castes, the greater the risk of violent political conflict. H 15 : The greater the relative size of scheduled tribes, the greater the risk of violent political conflict. H 16 : The greater the growth of scheduled tribes, the greater the risk of violent political conflict. 3. India: a case for population induced violence? India is mentioned by Thomas Homer-Dixon as a particularly pivotal state because of high population growth, serious water scarcity, cropland fragmentation, erosion, deforestation and desertification. He claims that these factors threaten to cause major internal violence or disintegrate the whole state (Homer-Dixon, 1999: 19 21). India is a highly fragmented country, with more than 24 languages spoken by more than one million people, and with two large religions, Hinduism and Islam. In addition, the caste system can function as a source to collective action, mainly between casteless Hindus (dalits) on one side and Hindus belonging to a caste on the other. Tribal systems also exist many places in India, particularly in rural areas. DeVotta (2002) shows how differential growth has been used in political agitation in India, especially by radical Hindu leaders who have argued that high Muslim growth rates will outnumber Hindus. 4. Research design and operationalizations The study covers all Indian regions states and union territories for the period from 1956 to 2002 (see Appendix A). 5 The modern state system of India came into effect in 1956 following the States Reorganisation Act. This new system differentiated between states that enjoy substantial autonomy, and union territories that are run by a Governor appointed by the federal government. While language has been the most central criterion for delineating states, the Indian federal government has been strongly opposed to the idea that states should encompass religious groups. Since 1956 there has been a number of changes in the state system, with larger states splitting and some union territories achieving statehood. By 2003 there are 28 states and 7 union territories in India. The total number of units (region-years) is This study uses internal conflicts for India from the PRIO/Uppsala dataset (Gleditsch et al., 2001; Eriksson et al. 2003) as a measure of violent political conflict. For a conflict to be listed in this dataset, it has to be between the Indian federal or local government and one or more organized opposition groups, and there has to be at least 25 annual battlerelated casualties. The incompatibility is defined as being over either governance or territory. A list of internal Indian conflicts between 1956 and 2002 is found in appendix B. During this period there were 11 conflicts taking place in nine different states. Two of the conflicts are defined as being over government, while the rest were over territorial issues. However, for this paper I do not distinguish 5 For Jammu & Kashmir, data is only available from the parts of the territory controlled by India. 7

9 between the two forms of conflict. I have been using the conflict location dataset developed by Halvard Buhaug to identify the states in which the conflict took place (Buhaug & Gates, 2002). The total number of state years in conflict is 119. The variable is coded as a dummy variable with the value 1 for years in conflict and 1 for years in peace. I am following de Soysa (2002) when analyzing all years in conflict rather than just the onset. 6 I am using logistic regression with a control for conflict previous year to account for dependence between years of continuing conflict, and clustered on states to account for dependence between observations over time within states. The demographic variables are all constructed from Census information. Censuses were held in independent India in 1951, 1961, 1971, 1981, 1991 and Most of the information used for the construction of demographic variables is census data published in annual editions of Statistical Abstract India (CSO, annual). This enables the assignment of values to territorial units that undergo changes between two censuses. All population growth rates are decadal rates. Growth of total, urban and rural populations are assigned to the years of the period where the growth actually happened (i.e. the decadal growth rates for the period are assigned to the years ). For growth of religious, caste and tribe populations, however, the rates are assigned to the years of the following period (i.e. growth rates for the period are assigned to the years ). The reason for lagging this variable is that it is assumed in this paper that it is the perception of imbalances in growth between groups that matters to conflict. 7 Data on age structure used to construct variables youth bulges and young male bulges are not available from the Statistical Abstract and have been collected from census publications directly (Census of India 1961, Census of India 1991). 8 Data on literacy and mineral production are collected from the Statistical Abstract. Generally, Indian census data are assumed to be quite reliable. Sub-national statistical case studies have some advantages over comparable cross-national studies because data are collected by highly similar procedures for different regions and local communities. Data are thus assumed to be highly comparable over geographical units and over time. 6 Predictors of conflict are thus assumed to influence aspects both of why conflicts break out, and why the continue. A valid concern when using such design is the potential problem of endogeneity. For most the demographic factors I study I assume that the potential problem of endogeneity is small (although major conflicts may trigger large-scale inter-regional migration this will be further addressed in later versions of this paper). I would assume that growth rates of religious groups would be among the most likely factors to be influenced by ongoing conflict. In this design, this potential problem is mitigated by the lagging of the religious growth rate variables. 7 Consequently, differential growth rates should be measured from the year data were released, and not from the year of the actual census as done here. This will be updated in future versions of this paper. 8 In Norway, census publications for India are only available for 1961 and The age structure data will be updated for the other periods in future versions of this paper. 8

10 5. Results Correlating demographic data for India with the PRIO/Uppsala conflict data does not provide much support for neo-malthusian concerns over population pressure on the base of natural renewable resources (see Table 1). Regions with high overall population growth are not more susceptible to violent conflict. In fact, overall population growth is negatively, although statistically insignificant, associated to conflict under all model specifications. Population density 9 is negatively associated with conflict, although the statistical significance of this relationship vanishes if union territories are excluded or the density measure is logtransformed. This indicates that a statistically significant negative relationship is largely driven by a small number of peaceful and highly urbanized union territories. 10 The interaction between growth and density is robustly unrelated to conflict. When differentiating between rural and urban growth rates, some interesting results appear. Regions with high urban growth rates are significantly less prone to conflict. This result is surprisingly robust, also when excluding union territories. This may indicate that although urban areas may experience severe environmental and social problems, high growth rates of urban populations do not necessarily produce more violent conflict. 11 The negative and significant relationship between urban growth and conflict is assumed to be spurious, probably reflecting that migrants tend to move towards economic prosperous urban areas. Whether a region is predominantly rural or urban is not a statistically significant predictor of conflict, although the sign of the relationship between a large rural population and conflict is robustly positive. The most surprising result is that the level of literacy seems to be unrelated to conflict. Cross-national studies have more or less unanimously concluded that level of economic and social development are strong predictors of conflict, but there is no indication that increasing levels of development in India measured in terms of literacy is associated with peace. 9 The conventional density measure is a low-validity indicator of population pressure as it tells us nothing about the population relative to the carrying capacity of the land. In a previous cross-national study I have applied a measure of population pressure on total productive land area (Urdal, 2003). A major future task for this project is to develop an improved index of population relative to productive land. 10 Most notably New Delhi, Chandigarh and Pondicherry. 11 I would expect that urban pressures are more likely to lead to rioting than organized violent conflict. I am eager to learn whether this conclusion holds for other measures of violent political conflict. 9

11 Table 1: Neo-malthusian factors in violent political conflicts in India Explanatory Variables Model 1 Model 2 Modell 3 Model 4 Model 5 Model 6 States only States only States and UTs States and UTs States Only States and UTs Population Growth ß st.e (0.033) (0.036) (0.021) Population Density (0.0014) (0.0013) ** (0.0008) * ( ) Population Density ln (0.256) (0.220) Pop. Growth * Pop. Density interaction ( ) ( ) ( ) Rural share (0.0032) (0.030) (0.027) Rural growth (0.011) * (0.010) (0.0098) Urban growth ** (0.004) *** (0.0046) *** (0.0035) Sex ratio (0.004) (0.0033) (0.0039) (0.004) Literacy (0,001) (0.0008) (0.0013) (0.0013) Conflict previous year 5.90 *** (0.43) 5.76 *** (0.42) 6.00 *** (0.41) 5.96 *** (0.45) 5.94 *** (0.46) 6.02 *** (0.43) Constant *** (0.58) (3.86) (3.13) (3.73) N ,277 1,249 Log Likelihood Pseudo R Significant at 0.05 level, Significant at 0.01 level, Significant at <0.005 level * (2.99) (3.36) 1, Turning to other major concerns in Table 2, there is limited support for the proposition that ethnic heterogeneity is associated with conflict. Nor is there strong support in favor of the hypothesis stating that regions with a non-hindu majority are more likely to experience conflict. The positive and significant relationship of Model 9 is not robust over changing model specifications. While religious heterogeneity in itself seems to be unrelated to conflict, some aspects of differential growth between religious groups appear to be associated with conflict. Great demographic disadvantage experienced by the Hindu population relative to the largest non-hindu religious group is associated with an elevated risk of conflict. Other measures of demographic disadvantage between the largest and second largest groups, or between Hindus and Muslims, do not come out statistically significant. Also, in Model 10, growth of scheduled castes is associated with conflict. Both these relationships are quite robust. I previously voiced some concern over the potential problem of endogeneity, in this case one could speculate whether conflictridden areas would produce out-migration of Hindus and high-castes. However, when running the same models on conflict onset only, the relationships are robust. 10

12 Table 2: All population factors in violent political conflicts in India Explanatory Variables Model 7 Model 8 Model 9 Model 10 Model 11 Model 12 States and UTs States only States and UTs States only States and UTs States and UTs Population Density ß st.e * (0.0009) (0.0015) * (0.0010) (0.0017) Population Density ln (0.262) (0.553) Pop. Growth * Pop. Desnity interaction ( ) (0.0001) ( ) Rural share (0.032) (0.037) (0.035) (0.049) * (0.035) (0.051) Rural growth (0.012) (0.030) (0.011) (0.038) (0.012) (0.033) Urban growth *** (0.003) ** (0.0075) *** (0.005) *** (0.010) ** (0.002) *** (0.010) Sex ratio (0.004) (0.006) (0.0044) (0.008) (0.005) (0.009) Religious heterogeneity (0.903) Non-hindu majority (0.397) (0.595) * (0.455) Hindu disadvantage *** (0.002) ** (0.0029) Scheduled caste proportion (0.052) Scheduled caste growth *** (0.167) Scheduled tribe proportion (0.015) Scheduled tribe growth (0.073) Literacy (0.001) (0.0019) (0.0019) (0.002) (0.002) (0.0012) Young male bulges (0.026) Young male bulges * Sex ratio (inverted) ( ) Mineral production (0.335) Conflict previous year 5.88 *** (0.44) 5.53 *** (0.49) 5.67 *** (0.46) 5.57 *** (0.65) 5.48 *** (0.53) 6.97 *** (0.99) Constant ** (3.58) (5.74) * (4.06) (6.71) N 1, , Log Likelihood Pseudo R Significant at 0.05 level, Significant at 0.01 level, Significant at <0.005 level * (3.99) (6.20) In Model 11, the proportion of young men to all adult men is negatively associated with conflict and borderline significant. 12 Under some model specifications, this negative relationship becomes statistically significant. However, variation on this 12 The measures for youth bulges and young men bulges are highly correlated (0.9991) and replacing one with the other does not alter the results. 11

13 variable is very limited both in time and space, and the number of observations is currently low. If the relationship should hold for the expanded dataset, one possible explanation could be that young men in particular are drawn towards prosperous areas that are peaceful for other reasons. The interaction effect between sex ratio (inverted) and proportion of young men is positive, and when only considering territorial conflicts, the effect is statistically significant. Finally, Model 12 does not provide any support for the proposition that regions with resource wealth are more likely to experience conflict. 6. Conclusions The present analysis of the effects of population pressure on political violent conflict in India using the PRIO/Uppsala conflict data has yielded little support for the hypotheses set out in the beginning of this paper. The amount and geographical distribution of this particular form of conflict may not provide sufficiently variance for a statistical research design. A primary aim of the project will be to collect other kinds of data on political violent conflict, such as riots, and of political instability more generally. One possible avenue will be to see whether population pressure indicators, in particular differential growth rates, may influence on regional electoral results. There are some relationships of interest found in this study. The general picture is that similarly to what is found in many cross-national studies, the support for the neo-malthusian conflict scenario is limited. Of the proposed indicators of neomalthusian population pressure, only urban growth comes out statistically significant. But contrary to the expectation, high urban growth is associated with peace. A possible explanation may be that high urban growth rates may act as a proxy of promising economic opportunities. A similar explanation may be used for the negative relationship between young male bulges and conflict. Again could it be that the relationship reflects that economically prosperous (and hence peaceful) areas are where the young men go. There may be some indication that the interaction of high young male proportion and male dominated population is associated with conflict. This relationship will have to be further investigated. The only relationships that provide fairly robust support for any of the hypotheses are the measures for Hindu disadvantage and for growth of scheduled castes. In regions where the largest non- Hindu religious group grows much faster than the Hindu population, and where scheduled castes increase the most, there is a statistically elevated risk of conflict. Parallel to the collection of other measures of political violence, I will attempt to expand the dataset with several more indicators of population pressure. These include measures of migration, the creation of an index of population relative to the carrying capacity of a region, availability of other forms of renewable resources and state capacity/governance. To broaden the study of the effect of religion, I will be constructing a measure of horizontal inequality inequality on group level between religious groups using individual data from the Demographic and Health Surveys (DHS) (see Østby, 2003). Finally I will be looking for measures of interreligious social capital or institutions (such as labor union membership numbers) to test Varshney s (2001) proposition that the existence of interethnic civil society may moderate ethnic conflict. 12

14 7. Literature Austin, Jay E. & Carl E. Bruch, eds, The Environmental Consequences of War: Legal, Economic and Scientific Perspectives. Cambridge: Cambridge University Press. Bookman, Milica Zarkovic, The Demographic Struggle for Power. London: Frank Cass. Boserup, Ester, Population and Technological Change: A Study of Long-Term Trends. Chicago, IL: University of Chicago Press. Boserup, Ester & T. Paul Schultz, eds, Economic and Demographic Relationships in Development. Baltimore, MD: Johns Hopkins University Press. Braungart, Richard G., Historical and Generational Patterns of Youth Movements: A Global Perspective, Comparative Social Research 7(1): Buhaug, Halvard & Scott Gates, The Geography of Civil War, Journal of Peace Research 39(4): Bächler, Günther, Violence Through Environmental Discrimination. Dordrecht: Kluwer. Choucri, Nazli, Population Dynamics and International Violence: Propositions, Insights and Evidence. Le xington, MA: Lexington. Census of India Vol I, Part II-C(i), Social and Cultural Tables. New Delhi: Registrar General & Census Commissioner. Census of India State Profile 1991 India. New Delhi: Registrar General & Census Commissioner. Collier, Paul, On the Economic Consequences of Civil War, Oxford Economic Papers-New Series 51(1): Collier, Paul, Doing Well Out of War: An Economic Perspective, in Mats Berdal & David M. Malone, eds, Greed & Grievance: Economic Agendas in Civil Wars. Boulder, CO & London: Lynne Rienner (91 111). CSO, annual. Statistical Abstract India. New Delhi: Central Statistical Organisation. de Soysa, Indra (2002). Paradis e is a Bazaar? Greed, Creed, and Governance in Civil War, Journal of Peace Research 39(4): DeVotta, Neil, Demography and Communalism in India, Journal of International Affairs 56(1): Ellingsen, Tanja, Colorful Commu nity or Ethnic Witches' Brew? Multiethnicity and Domestic Conflict During and After the Cold War, Journal of Conflict Resolution 44(2): Eriksson, Mikael, Petter Wallensteen & Margareta Sollenberg, Armed Conflict : A New Dataset, Journal of Peace Research 40(5). Esty, Daniel C. et al., State Failure Task Force Report: Phase II Findings. McLean, VA: Science Applications International, for State Failure Task Force. Copies may be requested from Fearon, James D. & David D. Laitin, Ethnicity, Insurgency, and Civil War. Paper presented at the American Political Science Association. San Francisco, CA, 30 August 2 September. Gleditsch, Nils Petter, Armed Conflict and the Environment, in Paul F. Diehl & Nils Petter Gleditsch, eds, Environmental Conflict. Boulder, CO: Westview ( ). Gleditsch, Nils Petter Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg & Håvard Strand, Armed Conflict : A New Dataset, Journal of Peace Research 39(5). Goldstone, Jack A., Demography, Environment, and Security, in Paul F. Diehl & Nils Petter Gleditsch, eds, Environmental Conflict. Boulder, CO: Westview (84 108). Hauge, Wenche & Tanja Ellingsen, Causal Pathways to Conflict, in Paul F. Diehl & Nils Petter Gleditsch, eds, Environmental Conflict. Boulder, CO: Westview (36 57). Henderson, Errol A. & J. David Singer, Civil War in the Post-Colonial World, , Journal of Peace Research 37(3): Homer-Dixon, Thomas F., On the Threshold: Environmental Changes as Causes of Acute Conflict, International Security 16(2): Homer-Dixon, Thomas F., Environmental Scarcities and Violent Conflict: Evidence From Cases, International Security 19(1): Homer-Dixon, Thomas F., Environment, Scarcity, and Violence. Princeton, NJ: Princeton University Press. Homer-Dixon, Thomas F. & Jessica Blitt, Ecoviolence: Links Among Environment, Population and Security. Lanham, MD: Rowman & Littlefield. Horowitz, Donald L., Ethnic Groups in Conflict. Berkeley, CA: University of California Press. JIA, Demography and State Stability, Journal of International Affairs 56(1): in press. Special issue. 13

15 Kaplan, Robert D., The Coming Anarchy, Atlantic Monthly 273(2): Krebs, Ronald R. & Jack S. Levy, Demographic Change and the Sources of International Conflict, in Myron Weiner & Sharon S. Russel, eds, Demography and National Security. New York/Oxford: Berghahn (62 105). Lomborg, Bjørn, The Skeptical Environmentalist: Measuring the Real State of the World. Ca mbridge: Ca mbridge University Press. Mesquida, Christian G. & Neil I. Wiener, Human Collective Aggression: A Behavioral Ecology Perspective, Ethology and Sociobiology 17(4): Myers, Norman, Ultimate Security: The Environmental Basis of Political Stability. New York: Norton. Østby, Gudrun, Horizontal Inequalities and Civil War: Do Ethnic Group Inequalities Influence the Risk of Domestic Armed Conflict? Cand. Polit. thesis, Department of Sociology and Political Science, Norwegian University of Science and Technology. Renner, Michael, The Anatomy of Resource Wars. Worldwatch Paper 162. Washington DC: Worldwatch Institute. Sambanis, Nicholas, Do Ethnic and Nonethnic Civil Wars Have the Same Causes? A Theoretical and Empirical Inquiry (part 1), Journal of Conflict Resolution 45(3): Sciolino, Elaine, Radicalism: Is the Devil in the Demographics?, New York Times, December 9. Simon, Julian L., Lebensraum: Paradoxically, Population Growth May Eventually End Wars, Journal of Conflict Resolution 33(1): Steinberger, Michael, So, Are Civilisations at War? Interview with Samuel P. Huntington., The Observer, October 21. Toft, Monica D., Differential Demographic Growth in Multinational States: Israel's Two-Front War, Journal of International Affairs 56(1): Urdal, Henrik, Nasjoner som kategorier. En analyse av folketellingene i Bosnia-Hercegovina [Nations as categories: An analysis of the population censuses of Bosnia-Herzegovina], Tidsskrift for Samfunnsforskning 42(4): Urdal, Henrik, The Devil in the Demographics: How Youth Bulges Influence the Risk of Domestic Armed Conflict Paper Presented at the 43rd Annual Convention of the International Studies Association (ISA), New Orleans, LA, March. Urdal, Henrik, People vs Malthus: Population Pressure, Environmental Degradation and Armed Conflict Revisited. Paper Presented at the 44th Annual Convention of the International Studies Association (ISA), Portland, OR, 25. February 1. March. Varshney, Ashutosh. Ethnic Conflict and Civil Society: India and Beyond. World Politics 53(3): Weiner, Myron & Sharon S. Russel, eds, Demography and National Security. New York/Oxford: Berghahn. Weiner, Myron & Michael S. Teitelbaum, Political Demography, Demographic Engineering. New York/Oxford: Berghahn. Zakaria, Fareed, The Roots of Rage, Newsweek 138(16):

16 8. Appendices Appendix A. Indian States and UTs My ID Name Start End 101 ANDAMAN AND NICOBAR ISLANDS ANDHRA PRADESH ARUNACHAL PRADESH ASSAM BIHAR BOMBAY CHANDIGARH CHHATTISGARH DADRA AND NAGAR HAVELI DAMAN AND DIU DELHI GOA GUJARAT HARYANA HIMACHAL PRADESH JAMMU AND KASHMIR JHARKHAND KARNATAKA (MYSORE) KERALA LAKSHADWEEP MADHYA PRADESH MAHARASHTRA MANIPUR MEGHALAYA MIZORAM NAGALAND ORISSA PONDICHERRY PUNJAB RAJASTHAN SIKKIM TAMIL NADU TRIPURA UTTAR PRADESH UTTARANCHAL WEST BENGAL

17 Appendix B. Conflicts in India Indian state Opposition Territory Begin End Orissa Naxalites/CPI (-Marxist) Andhra Pradesh Naxalites/PWG, MCC Andhra Pradesh Naxalites/PWG, MCC Assam NNC, NSCN Nagaland Assam/Nagaland NNC, NSCN Nagaland Nagaland NNC, NSCN Nagaland Assam MNF Mizoram Tripura TNV Tripura Tripura ATTF Tripura Tripura ATTF, NLFT Tripura Manipur PLA Manipur Manipur PLA Manipur Manipur UNLF, KNF, PLA Manipur Punjab Sikh insurgents Punjab/Khalistan Punjab Sikh insurgents Punjab/Khalistan Punjab Sikh insurgents Punjab/Khalistan Punjab Sikh insurgents Punjab/Khalistan Jammu and Kashmir Kashmir Insurgents Kashmir Jammu and Kashmir Kashmir Insurgents Kashmir Jammu and Kashmir Kashmir Insurgents Kashmir Jammu and Kashmir Kashmir Insurgents Kashmir Assam ABSU, BPAC Assam Assam ULFA Assam Assam BDSF, ULFA, BLTF, NDFB Assam Bihar Jharkand Mukti Morcha Jarkhand

18 Appendix C. Descriptive statistics Variable Obs. Mean Std. Dev. Min Max Total population Rural share Urban share Rural growth Urban growth Density Population growth Sex ratio Religious heterogen Hindu-Musl diff Religious disadv Hindu disadv Sched. Caste share Sched. Caste growth Sched. Tribe share Sched. Tribe growth Literacy Youth bulges Young men bulges Mineral production Density ln