Power Sharing in the World’s Largest Democracy: Informal Consociationalism in India (and Its Decline?)

India is one of the most diverse countries of the world but operates with a majoritarian Westminster constitution and simple plurality electoral system, albeit also with a federal structure. It was eventually coded as consociational by Arend Lijphart (1996) but this coding was questioned by authors such as Wilkinson (2000) and Adeney (2002). This article assesses the nature of both de jure and de facto power sharing in India over its 70 years of independence and tracks the evolution of de jure and de facto power sharing in relation to four dimensions of diversity: religion, caste, territory and language. It questions whether the electoral success of Hindu nationalism and the increasing acceptance of ethnic majoritarianism has reduced the degree of power-sharing in India.


Introduction: 'The Puzzle of Indian Democracy' revisited
Lijphart's theory of consociationalism was developed to explain the stabilisation of democracy in societies with entrenched cleavages. These were defined as 'fragmented but stable democracies ' (1969, 211). Initially focused on explaining the cases of the Netherlands, Belgium, Austria and Switzerland, he (and others) developed his argument and applied it to cases such as Cyprus (prior to the island's partition), Lebanon (pre and post-civil war), Malaysia, South Africa (interim constitution) and Northern Ireland (McCulloch 2014). Consociationalist theory posits that deeply divided societies can only survive as democracies if they adopt four key features: (1) executive power-sharing through a grand coalition; (2) segmental or group autonomy (in the form of cultural, educational or territorial selfrule); (3) proportionality and (4) minority veto. Lijphart later identified powersharing, especially at the executive level, and group autonomy as the two prime features of consociational democracy (2002,39).
Until 1996, Lijphart excluded India from his examples of consociationalism. Yet in the 1990s Lijphart admitted that the exclusion of India was a mistake, echoing Paul Brass's (1991, 342) observation that 'consociationalists ignore the experience of India, the largest, most culturally diverse society in the world that has… functioned with a highly competitive and distinctly adversarial system of politics.' Brass argued that the omission of India had the potential to undermine the premises on which consociational theory rests.
At first sight, India's political institutions do not fit the key features of consociationalism very well. Despite its deep social divisions based on caste, religion, tribe, language and territory, the Indian constitution adopted something much closer to the adversarial Westminster model. India's simple plurality electoral system mainly generated single party majorities until 1989. India lacks a formal requirement to incorporate different groups based on religion, caste or territorial origins in its national executive, let alone entrust any of these groups with a veto. Although India adopted federalism (its constitution refers to India as a 'Union' of states and territories), it has many non-federal features, including the right of the central parliament to legislate in state matters, and, should the need arise, impose direct rule (Article 356 or so-called 'President's Rule').
In spite of this, Lijphart questioned whether 'Indian democracy is an undiluted application of the British [majoritarian] model ' (1977, 180). While India is largely lacking in formal consociational features, he noted that there are many informal features of the Indian system that are more consociational than majoritarian (1977,181). In 1996 Lijphart even went further and argued that India 'was an impressive confirming case" of consociationalism (1996,259). However, many authors have criticised Lijphart's re-coding of India as consociational (Lustick 1997, Wilkinson 2000, Adeney 2002, for reasons that will be detailed below. Indian politics has substantially changed since Lijphart's 1996 article and the rebuttals of Lustick, Wilkinson and Adeney. This makes the case for a re-evaluation of the Indian case pressing. Furthermore, although the earlier analyses provided a compelling critique of Lijphart's characterization of India as consociational, they did not map systematically the representation of different segments of society in the Indian executive or a range of political institutions. In this article we seek to advance an original and rigorous assessment of both de jure and de facto levels of consociationalism in the Indian case, across time and across different sources of diversity, i.e. religion, caste and territory. By adopting a longitudinal and multidimensional approach to the issue, we seek to answer three related questions: (1) was Lijphart correct in belatedly classifying India as a confirming case of consociationalism? (2) was Lijphart correct in his suggestion that India reached the peak of consociationalism during the Nehruvian (1952)(1953)(1954)(1955)(1956)(1957)(1958)(1959)(1960)(1961)(1962)(1963)(1964) period and that consociationalism declined thereafter? (3) what has been the effect of the rising electoral success of Hindu majoritarianism (with a single party majority in the 2014 Lok Sabha elections) on the power-sharing features of Indian democracy?
In the remainder of this article, we first summarize Liiphart's case for considering India as a consociational democracy and its critique. Next we identify the methods by which we will assess India's consociational features over time and across a range of group dimensions before presenting our analysis. In the conclusion we relate our findings to the three questions listed above.
with ministers belonging to all the main religious, linguistic and regional groups' (1996,260). Cultural autonomy was met by linguistic reorganization, the educational autonomy afforded to religious and linguistic groups (as well as 'full government financial support' for these (1996,260)) and the existence of separate personal laws for different religious groups. In spite of Simple Plurality, proportionality was met through the fact that 'plurality does not disfavour geographically concentrated minorities' such as India's linguistic groups, that the Congress Party is the 'party of consensus,' and has accorded 'a share of ministerships remarkably close to proportional' for religious groups (1996,261). He also cited the reserved seats for Scheduled Tribes (STs) and Scheduled Castes (SCs) and the reservations for Other Backward Classes (OBCs). He argued that his final element, the Minority Veto, was met through citing examples such as the retention of English as the additional official language (a veto of the non-Hindi speaking states) and the reversal of the Shah Bano judgement (a veto of the Muslim community) (1996, 261-2).
Lijphart's categorisation of India as a consociation has been criticised by Lustick (1997), Wilkinson (2000) and Adeney (2002) on three grounds. First, the diversity of representatives in the Cabinet that Lijphart cites as evidence of a grand coalition were 'Congressmen and women, very different from an ethnically defined elite cartel with authority over and autonomy from the community they represent (Adeney 2002, 28). They were 'token, not representative' individuals (Lustick 1997, 115). The second criticism was that 'although every Indian cabinet had its minority representatives, these men and women were kept well away from positions of real power' (Wilkinson 2000, 773). Finally, rather than this Grand Coalition being a 'deliberate or joint stabilizing effort by elites' it was dependent on one man: Nehru (Lustick 1997, 116).
Although Lijphart argues that India met the proportionality criteria with regards to caste representation, as Wilkinson analyses, there was a 'substantial gap between the government's promises on minority rights and its performance'; under Nehru the 'actual representation in senior positions was only a tenth of [the SC and ST] … allotted quota ' (2000, 773, 771). The situation in some states was even worse. In addition to this, reserved seats for religious minorities in the legislature were removed after independence (Wilkinson 2000(Wilkinson , 780, 2002. Finally, despite linguistic reorganization giving linguistically defined states representation in the upper house, representation was roughly determined on the basis of population, ensuring 'the domination of the larger populated Hindi-speaking states of the north in both the lower and upper chambers of the central legislature' (Adeney 2002, 23).
Lijphart's argument that India meets the criteria for Minority Veto was criticized for ignoring the evidence pointing in the opposite direction. This not only includes the removal of reserved seats for religious minorities but includes the abolition of protection for Urdu speakers as well as the ban on cow slaughter in many states (Wilkinson 2000, 774-5). As Lustick observes, for Lijphart, '[n]o constitutional establishment of minority veto is said to be necessary because of crosscutting cleavages … that divide the Hindu majority.' He argues, as have others (Singh 2000) that the Indian state adopted many policies that marginalised non-Hindu groups (Lustick 1997, 116).
Finally, although there were areas where the Indian state allowed cultural autonomy, most notably in terms of the linguistic reorganisation of states (Wilkinson 2000, 777), reorganisation was fiercely opposed by Nehru who only conceded the demand under 'the electoral threat that linguistic mobilization posed to the Congress's cohesion' (Adeney 2007, 96). As Adeney argues, Nehru 'deliberately ignored ethnicity as a means of legitimising the state, basing affiliation on a civic notion of territoriality ' (2002, 25). This means that although there may have been segmental autonomy in practice, the rationale behind its adoption was a non-consociational one. Furthermore, the provisions that provided cultural autonomy for the linguistic or religious minorities within the newly created linguistic states were routinely ignored (Wilkinson 2000, 778). Despite the retention of personal laws for Muslims and Christians, the absence of a minority veto resulted in 'state attacks on minority cultural autonomy ' (2000, 778).

The Decline or Rise of Consociationalism?
Lijphart pointed to a decline in consociationalism after the Nehruvian era. This was on the basis that 'the Congress Party was transformed from an internally democratic, federal, and consensual organization to a centralized and hierarchical party.' Lijphart also cites the increasing centralization of the federal system through the partisan use of President's Rule and the increasing support for parties such as the BJP who protest at the 'government's alleged pandering to minorities ' (1996, 262).
Lijphart's characterisation has been contested vigorously by Wilkinson who argued that, rather than Indian democracy becoming less consociational after Nehru, 'India ha [d] in fact become more consociational' (2000,770). The fragmentation of the party system from one party dominance played an important role in strengthening the consociational traits of India's federal system. By 2004 Indian democracy had cemented itself around two political fronts, the National Democratic Alliance and the United Progressive Alliance, entrenching coalition politics and power sharing in both pre-electoral and post-electoral pacts (Arora and Kailash 2012). Parties with an exclusive regional following entered national coalition politics and were expected to use their powers to extract regional benefits or protect regional autonomy (Guha Thakurta and Raghuraman 2007). In this climate of coalition politics, the Supreme Court also developed a more activist stance in the policing of President's Rule, preventing its use for purely party political purposes (Sathe 2007, Swenden 2016a andb). Furthermore, the transition from a state-led to a more market-led economy strengthened the economic autonomy of the states. State governments became less bound by central economic regulations and even competed against each other for inward economic investment (Rudolph and Rudolph 2001). In parallel, the approval of Mandal I and Mandal II extended reservations in the civil service (Mandal I) and higher education (Mandal II) to OBCs, seeking to reduce the disproportionate influence of Forward Castes in politics, administration and higher education (Jaffrelot and Kumar 2009).
However, these overtures towards accommodation on the basis of territory and caste were accompanied by the rise of Hindu nationalism as a political and electoral force. Until 2014, Hindu nationalist forces could only operate at the level of the Indian central state in coalition with secular parties. In the 2014 general elections, the Hindu nationalist BJP secured a single party majority of seats in the Lok Sabha. Although it is important to stress that it would be more appropriate to talk of Hindu communities rather than a single Hindu community, there is a growing acceptance of the political ideals associated with the Hindu-right and with it a growing willingness to view India as a democracy by and for the Hindus (Jaffrelot 2017). The proportion of people agreeing that 'democracy means following the will of the majority community' has increased (to over 50% of the population) since 2004, and it is not just BJP voters who hold this view (Palshikar 2015, 731-2). This is significant, because (as we will discuss) as many of the consociational elements of India's political system are de facto rather than de jure ones, they are subject to revision.
It follows that for the period between 1996 (the entrenchment of coalition government at the centre) and 2014 we find forces that support more accommodation based on territory (self-rule) and caste; but we also see the weakening of support for accommodation on the basis of religion and (possibly given the association between Hindu nationalism and Hindi), language (Adeney 2015, 12). The alignment of political power with the Hindu right after the 2014 general elections has the potential to undermine the de facto consociationalism that had developed in the era of coalition politics (Sharma and Swenden 2018)).
Based on the literature review we can therefore suggest three propositions which will be critically examined in light of the earlier raised questions: (1) Although not entirely lacking (de facto) elements of power-sharing, India did not function as a power-sharing or consociational democracy between 1952 and 1989 (confirming Wilkinson and Adeney). (2) Following the demise of the one-party system (1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996) Indian democracy acquired more de facto and de jure consociational properties in relation to the accommodation of caste and territory compared with the oneparty dominant phase. (3) The rise of Hindu nationalism since the 1980s and the restoration of single party government (2014) has negatively impacted the accommodation of religious minorities, especially the Muslim community.

Methods: Measuring and Interpreting Power-Sharing
To gain a nuanced understanding of consociationalism in India we need to rigorously and systematically analyse the different elements of consociationalism over time in India, looking at both formal and informal elements of the political system. We reject the dichotomous categorisation of India as either consociational or non-consociational as it is possible that a political system may have elements of consociationalism while lacking others. Instead, we focus on identifying what elements of the system are consociational, building on Adeney's (2002) method of scoring elements of different parts of consociationalism. Unlike Adeney, we also consider the more informal elements that may be present within a system to more accurately reflect the de facto operation of the system. These informal elements may relate to the 'conventional' practice of dispersing office in the executive, highest courts of the land or military post-holders across different ethno-linguistic or territorial groups. They may also be reflected in the internal organisation of political parties (Bogaards 2014). In addition to assessing the absolute representation of different groups in many of these key institutions, we calculate the deviation from proportionality from the representation of a particular group based on their share of the population in decennial population censuses since 1951. Relaying representational data to different census points reflects that the share of the population who belong to a certain caste group, reside in a certain state or union territory, or practice a particular religion varies across time and space; reflecting migration patterns or variable population growth figures of different groups. For example, the Muslim population has increased from 9.8% in 1951, to 14.2% in 2011, and the Southern states' population has decreased vis-a-vis the North. We only use fixed population shares for caste (based on the Mandal Commission report, Jayal 2006) since until 2011 Indian population censuses did not ask respondents to identify with caste affiliation (and the relevant findings of the 2011 census have not yet been published). For regional affiliation we use regional groups which correspond with the regional zones and zonal councils in India, i.e. North, Central, East, South and West (Inter-State Council 2018) plus the North-East. We accept that tabulating groups according to their religion runs the risk of assuming that these groups are more homogeneous than they are. This applies both for majorities (such as 'Hindus') as well as minorities (such as 'Muslims'). However, we also disaggregate the Hindu community according to region and caste to overcome this problem.

Grand Coalition
India does not have a formal grand coalition. The Cabinet and government are formed by the party that can command a majority in the Lok Sabha and the Prime Minster is (normally) the leader of the largest party in the parliament. The President -the Head of State -is elected by an electoral college including national and state assemblies, and cohabits with the government after an election -something that has become more important as the role of the President has become more activist, particularly in relation to the use of Emergency powers of the constitution. As there is no formal requirement that the (coalition) government represents different segments of India's population, we concentrate on three elements to score the informal existence (or otherwise) of a grand coalition: (1) the existence of members of groups within the Cabinet; (2) the ministerial portfolios held by different groups and (3) the existence of groups within the two largest polity-wide parties: the INC and the BJP, following Bogaards' recommendation to focus on intra-party representation as an additional measure of consociationalism in a political system (2014,110). We consider each of these in turn.

Group representation in Cabinet
Using an original data set compiled by [insert reference here] our calculations include those members of the Council of Ministers who hold Cabinet Rank, excluding Deputy Ministers or Ministers of State. We have tabulated the members of the Cabinet by religious, caste/tribal and regional identity. Although authors have argued that the composition of the Cabinet reflects India's diversity (Lijphart 1996, Wilkinson 2000, Bogaards 2014, there have been few systematic evaluations of how this has changed over time and varies across different sources of diversity e.g. region compared to religion (Panandiker andMehra 1996, Jayal 2006). Source: 1947Kohli (2000, 2000-2018 compiled from https://archive.india.gov.in, https://www.india.gov.in/my-government/whos-who/council-ministers Given their very small numbers, Parsis and Buddhists have been subsumed under Others or N/A. For a Grand Coalition to materialize, groups that are small within the polity should be over-represented so that their voice is adequately heard. As Figure 1 reveals, despite making up 80% of the population, Hindus comprised more than 90% of recent Cabinets, especially those led by the Hindu nationalist BJP. In contrast, Muslims are the most under-represented, at least in Cabinets that have been formed since the 1990s. However, as noted above, Hindus are not a united community. As Table 1 demonstrates, variations in the geographical strength of the lead party in central government have affected the representation of different regions. For instance, factional infighting within Congress after the 1967 elections and the rise of the backward castes eroding Congress support in the Hindi belt during the 1990s generated a considerable under-representation of MPs from the Central states (which include the Hindi belt states) in the Cabinets of Indira Gandhi (1967Gandhi ( -1975 and Manmohan Singh (2004.   The change from a one-party dominant system to a multiparty system has increased regional representation, particularly for the states in the South and the West of the country. However, very few Indian Cabinets have compensated the small size and geographically isolated position of the North-Eastern states by offering them Cabinet berths above their demographic share in the population. And the Central and Eastern regions have not been proportionally represented either, interestingly given that these contain many of the Hindi heartland states.

Ministerial Portfolio Allocation in Cabinet
Turning to the position held by ministers in the Cabinet, a better way of assessing influence, we concentrate on religious minorities for reasons of space. Wilkinson argued that 'even when minorities were given ministerial appointments, they were kept well away from the most important and sensitive posts ' (2000, 779). In Table  2 we have tabulated their religious identity according to appointment. Although some may dispute Wilkinson's contention that the education portfolio is a 'less important' ministry, it is widely accepted that the portfolios of Finance, Home and Foreign Affairs are the most coveted.  (Hazarika 1994)) and later by another Sikh, in Rajiv Gandhi's administration during the continuing violence in the Punjab (Chawla 1988) The Defence brief, one of those identified by Wilkinson as being too 'sensitive' to be allocated to a Muslim (2000, 779) has never been allocated to a Muslim. Two Christians have occupied this position -one in the second and third Vajpayee cabinets and one in UPA cabinets headed by Manmohan Singh. Two Sikhs have also held the Defence brief; one in Nehru's first cabinet of 1947 and 1953, and another in two of Indira Gandhi's administrations. Interestingly, despite their preponderance in the army (see further), no Sikh has held either Defence or External Affairs since 1974.
A few other patterns emerge. In the Nehruvian era, Muslims held the majority of the Education minister portfolios: an important element for segmental autonomy. Since its creation in 2006, the Ministry for Minority Affairs has always been allocated to a Muslim member of the Cabinet. The other notable pattern has been the fact that Sikhs have held a large number of positions related to Food and Agriculture. This is potentially relevant, although beyond the purview of this paper, because Punjab is the 'agricultural basket' of India and its allocation to Sikhs in several different cabinets, all of whom have their base in, or were directly elected from, Punjab is also potentially indicative of segmental autonomy.

Group representation in the two main polity-wide parties.
For the final assessment of the existence of an informal grand coalition, we discuss group representation within the party executives of the INC and BJP. Both politywide parties have established organizational divisions or cells that look after the representation of certain group interests such as the All-India-Adivasi Congress cell, (Indian National Congress, 2018) or 'morchas' aimed at channelling the interests of STs, SCs, OBCs, Kisans or peasants and religious minorities within the BJP (BJP 2018). However, these bodies do not necessarily wield influence in terms of party policy and candidate selection. To assess this, one needs to look more systematically at the composition of the party national executives, i.e. the Congress Working Committee (CWC) and the BJP national executive (and parliamentary board). Drawing from Jayal (2006, 176) we reproduce data that correspond with key periods in the history of Congress and the BJP. We focus on the evolution in the representation of religious minorities and expect to find a significant difference between the profile of both parties based on Congress' secular outlook and the BJP's profile as a Hindu nationalist party.

Figure 3: Over/Under representation of religious groups in Congress Working Committee (select years 1952-2018)
Source: Kochanek (1968, 365) for 1946-52, Jayal (2006 for 1972-2002, Congress Party website for data 2016 and 2018: https://www.inc.in/en/congress-working-committee/members Figure 3 reveals that Christians and Sikhs have generally been well represented within Congress, but the representation of the more sizable Muslim minority within Congress' key decision-making body has been patchier. This is so notwithstanding the importance of the party's Muslim 'vote-bank' (Palshikar and Suri 2014). Whilst Muslims were significantly over-represented in the party's 2002 national executive, this is no longer the case for the most recently formed executives, including the first executive with Rahul Gandhi as party president in 2017.
In comparison, and as expected, Figure 4 shows that the underrepresentation of religious minorities within the national executive of the BJP is more systematic, affecting Christian and Muslims in particular. The level of Muslim underrepresentation has also deteriorated across time.

Minority Veto
The stronger the participation of minority groups in key decision-making bodies of the state such as the Cabinet, the more likely policies can be expected to accommodate group sensitivities (i.e. generate 'substantive' representation).
Although an assessment of 'outcome' indicators falls beyond the scope of this article, for Lijphart minority groups should be able to 'veto' proposals that undermine their rights and autonomy in a consociational democracy (2017, 3).
In India, certain group rights are constitutionally entrenched. Some of these rights are not only protected against majoritarian revision (special bicameral majorities are required to amend the constitution, protecting linguistic, religious and lower caste groups), but occasionally also against any forms of political revision. Indeed, in 1973 the Supreme Court ruled that the 'basic structure' of the constitution could not be amended. In time, this ruling has been understood to rule out constitutional amendments which undermine among others (the secular and federal character of the constitution; and a set of individual rights and freedoms (as specified in Part III of the constitution) (Chinappa 2008, 60).
Secularism has to be understood as India's pledge to apply an equidistant approach to the various religious communities, rather than as a means to separate state and religion. In colonial times, personal law practices for each religion, adjudicated by separate courts, governed issues such as marriage, divorce, maintenance, inheritance, custody and adoption (Lerner 2011, 135-47 The ability of religious minorities to defend their customs has also come under threat in other areas. Most notably, the Indian constitution acknowledged strong pressures to protect the cow (a sacred animal in Hinduism) by urging states to prohibit the slaughter of cows or draught animals which could pull carts or ploughs. This provision, like the UCC, was only inserted as a directive principle, putting it beyond the purview of judicial enforcement (Kazmin 2017, 25). Although many states banned the slaughter of productive cattle in the 1950s, this ban did not apply to aged cattle. With the electoral success of Hindu nationalism, beef bans have been more rigorously enforced (often by cow vigilante groups or gau rakshaks rather than the state police (Jaffrelot 2017)). In May 2017 the central government passed a decree (which ensured that it did not have to be debated in parliament) that 'no person shall bring a cattle (sic) to an animal market unless upon arrival he has furnished a written declaration … stating that the cattle has not been brought to market for sale for slaughter' ( The principle of a minority veto applies more firmly to policies which pertain to divisions that cross-cut the Hindu majority community, namely, caste and region (and the linked issue of language). The implementation of the Mandal Commission, introducing reservations for OBCs in administrative and educational positions, faced strong resistance from Forward Caste Hindus when it was first announced. However, its recommendations were implemented when political parties representing Backward Castes came to political power. Now 'the political arithmetic of reservations is so overwhelming that no major politician will come out openly against them' (Wilkinson 2010, 266), rather, reservations continue to be extended (The Economic Times 2019).
Similarly, the linguistic diversity within the Hindu community as a whole required a carefully crafted constitutional and institutional compromise, which has governed India's language policy since the 1960s. For more than half of Hindus, mainly in the South of the country, Hindi is not their mother language. They successfully prevented the removal of English as a joint official language (Adeney 2007, 94). The linguistic reorganization of India in the 1950s also ensured that the states, controlled by India's multiple linguistic groups, have a say in constitutional amendments affecting the federal balance of power e.g. over the distribution of powers between the Union and the states and the representation of states in Parliament (although, interestingly not over the official language of India). The strength of regional identity in the Dravidian South, especially in Tamil Nadu, has sensitized South Indian communities to attempts to marginalize their political influence. They have secured a certain degree of over-representation in the Indian Parliament. Despite Article 81 of the constitution which recommends that 'seat allocation in such manner that the ratio between the number and the population of the state is, so far as practical, the same for all states' parliament in 1976 amended the Constitution to freeze the strength of the Lok Sabha and state legislatures until the 2001 census. This article was inserted in order not to penalise the more effective family planning control policies of the Southern states compared with the more populous states of the Hindi belt (especially the Central regions). The provision was extended in 2001 until 2026. A de facto minority veto therefore operates with regard to smaller linguistic groups, although, as noted above, constitutionally, the official language of the Indian Union can be changed by a two thirds majority in both houses of Parliament.

Proportionality
Lijphart argued that proportionality 'is a method of allocating civil service appointments and scare financial resources in the form of government subsidies among the different segments ' (1977, 38). Furthermore, 'not only should all significant segments be represented in decision-making organs, but they should also be represented proportionally ' (1977, 39). India has formal proportionality of the Lijphartean type only in regards to SC and ST representation in the Lok Sabha, bureaucracy and higher education institutions. In 1994 administrative reservations were extended to OBCs for up to 27% of places, following the implementation of the Mandal Report (Jayal 2006, 162). These were further extended to higher education in the 2000s (Chauhan 2008). Despite the provision of formal reservations in many institutions, these reserved positions are not always filled. Furthermore, we need to consider informal measures of proportionality in relation to non-caste groups. These include the representation of different religious and linguistic groups in the Lok Sabha and key institutions of the state over time.

Proportionality in the Lok Sabha
Reserved seats and administrative and educational reservations have certainly helped to improve the representation of SCs and STs. In the Lok Sabha, 84 seats (15.46%) are reserved for SCs and 47 (8.66%) for STs. Consequently, their level of representation stands in line with their share of the population as a whole. 2 seats are also reserved for members of the Anglo-Indian community. However, SCs in particular obtain very few general seats, typically only around five (1.2% of general category seats). Parties either do not field SC candidates in these seats (even though they are not barred from standing) or, as is the case for the Bahujan Samaj Party (BSP), a party that seeks to give voice to the Dalits, Dalit candidates are only fielded where their chances for success are marginal, i.e. outside of Uttar Pradesh where the BSP has a limited following (The Hindu 2016). In the case of STs however, the number of seats from non-reserved constituencies is generally higher, usually around 10 (Jayal 2006, 145). This may reflect the more concentrated character of ST populations in some seats which are not reserved, meaning that they are able to secure representation in simple plurality electoral contests.
Unlike SCs and STs, OBCs do not benefit from reserved seats. This has had major implications for their level of underrepresentation as Figure 5 demonstrates. Although Hindu OBCs saw a marginal improvement in the Lok Sabha from the 1980s, reflecting a strengthening of their representation in the states of the Northern and Central regions of the so-called Hindi-belt (Jaffrelot and Kumar 2009), since 2009 the level of OBC representation in the Lok Sabha has declined (Jaffrelot and Verniers 2015). It continues to fall well short both of the 27% reservation quota for OBCs in the bureaucracy and higher educational institutions, and their share of the population.

Figure 5: Deviation from Proportionality for Lok Sabha 1952-2014
Source: Calculations authors' own. (Jayal 2006, Shankar and Rodrigues 2011, Trivedi unpublished dataset Lok Sabha 1952. Numbers do not round to 100% due to the small number of candidates not able to be classified by caste or religion.
Reserved seats were removed for religious minorities after independence. Many have argued that religious minorities have become under-represented as a result (Khalidi 1993, Sachar 2006. Of the minority religious communities, the Muslims are the most under-represented Their level of under-representation has progressively increased since the late 1980s. The tally of 20 MPs in the 16 th Lok Sabha was the lowest figure ever; the governing BJP did not have a single Muslim among its 282 MPs. The degree of under-representation is generally lower for the Sikh, Christian and Buddhist minorities.

Proportionality in Administrative Positions (Central Government Services)
SCs and STs benefit from affirmative action policies in India. Places are reserved for these groups in civil service and educational posts. These reserved places were extended to the OBCs in the civil service when the Mandal Report was implemented in 1994, and to educational places in 2006. Figure 7 reproduces the representation of SCs (Dalits) in central government jobs by employment category. Category A jobs reflect the highest status and level of responsibility (typically associated with senior management and including the All-India Services), Category D jobs comprise the  1952 1957 1962 1967 1971 1977 1980 1984 1989 1991 1996 1998 1999 2004 2009 2014 Forward Hindu Backward Hindu Scheduled Caste

Scheduled Tribe Muslim
Other religious minority lowest level of responsibility (and can include menial jobs such as sweeping). Figure  6 reveals that SCs remain slightly under-represented in the prestigious category A jobs (despite an increase in the 1980s) although their position is improving. In contrast, SCs are overrepresented in the lower, menial category D job bracket.
Although not reproduced below, government employment for STs demonstrate a similar pattern. Although the introduction of OBC reservations has increased the representation of OBC in government jobs, OBCs remain significantly under-represented across all categories. Figure 7 demonstrates that OBC representation falls short of meeting even the 27% OBC quota, let alone their share of the Hindu population (44%), across all four categories. More often than not jobs that are earmarked for OBCs are not filled (e.g. in the case of category A jobs because not enough candidates from disadvantaged background present themselves or meet the relevant entry requirements). However, reserved positions were only created in 1994, and, there is evidence of some improvement in the numbers of OBCs being appointed to senior positions over the last 15 years. This is in contrast to the Muslim communities. The absence of group specific quotas has hampered the representation of religious minorities in central government services, especially Muslims. Table 3 lists their representation during two data points for which information could be found, 1980 and 2006. IAS, IPS (Indian Police Services) and IFS (Indian Forest Services) posts are the most prestigious among the category A jobs. The representation of Muslims in these posts is 10% or more below their share of the population. Put differently, the share of Muslims is between a sixth or a third (depending on the type of service) of what it should be. The Narendra Modi government claimed that it was trying to redress this imbalance: its 2014-2015 intake comprised close to 9% of 'minorities' (Annual Report of Minority Affairs 2016, 54). However, this also includes Sikhs, Buddhists, Christians, Jains and Parsis.   (Saxena 1983, Sachar 2006

Key Political, Military and Judicial Offices
Finally, in consociational systems, key political, judicial or military offices are often held by different communities to strengthen their political and symbolic representation within the state. Figures 8 and 9 list the distribution of these offices in the period between 1947 and 2018 (March) based on religion and region.  been SCs. Although we have not listed it in Figure 8, India's most important political office, that of the Prime Minister, has always been occupied by Forward Caste Hindus, except for Narendra Modi (the first OBC Prime Minister) and his predecessor Manmohan Singh, a Sikh. Both these Prime Ministers were of the two main national parties, but the fact that they took office in the era of coalition politics is further evidence that Indian politics has become more inclusive in the coalition politics era.
Of the offices listed above, the Supreme Court Chief Justice renders decisions in a collective manner since judgements require majority support in benches of three or more judges. Yet, as for other political institutions, the representation of Muslims is weak and has been declining.  (Chandrachud 2014, 254-58). As such, we do not observe a considerable rise in lower caste representation in the Supreme Court.
Finally, the allocation of key offices to regions provides a broad picture of the extent to which the polity is geographically inclusive. Figure 9 confirms a few patterns already observed above. The small North-Eastern region is extremely underrepresented. The regions of the North and West have produced more army chiefs than the regions of the East and Central area, again, reflecting path-dependent legacies of perceived 'martial races' during colonial times (Sikhs, Jats, Rajputs) from North and West India (Wilkinson 2015, 154-191) The populous, but relatively poor Central zone (including the states of Uttar Pradesh and Madhya Pradesh) has been under-represented in all four offices identified. In the case of the Speaker's office this may be due to the weakness of the Congress in the Hindi belt from the late 1980s.

Figure 9: Deviation from Proportionality for office holders based on region
Source: As Figure 9 Segmental Autonomy The final area we assess relates to segmental autonomy. Segmental autonomy entails 'the delegation of rule-making and rule-application powers to the segments'; important for providing groups with the right to make decisions over issues which affect only their group, as opposed to those over which there is 'common interest' (Lijphart 1977, 41). This is an area of India where there are more formal safeguards at least for linguistic, religious and territorially concentrated minorities.
In relation to formal territorial and linguistic segmental autonomy, the provisions of the Indian constitution that states can choose the language(s) in which they operate and communicate with the Indian Union was a form of segmental autonomy that was strengthened after the linguistic reorganization of states in the 1950s and the Official Language Act of 1967. This decision contrasted with the policy of neighbouring Pakistan that not only initially imposed one official language -Urduon the entire population but also refused the right of provinces to adopt provincial languages for use within their territory (Adeney 2007, 80). Linguistic reorganization in India thus secured the right of linguistic majorities at a regional level to implement their own cultural and educational decisions. The STs of India also possess territorial segmental autonomy through the provisions of the Sixth Schedule to the constitution regarding cultural protection and exceptions to national laws. As Tillin has discussed, this constitutes a form of asymmetrical federalism (2007). There is no provision for segmental autonomy along territorial lines. Although Kashmir possesses a special status under Article 370 of the constitution and non-Kashmiris are prohibited from purchasing property in the state, this is a result of the terms of the accession to the Indian Union rather than explicit religious segmental autonomy.
Formally speaking state autonomy is not secure. Article 266 of the constitution enables the central parliament to legislate on exclusive state subjects if they become a matter of national concern. More importantly, by invoking emergency powers (socalled President's Rule as per Article 356 of the constitution), the central government can suspend state autonomy. However, although frequently abused for party political purposes under one party dominant rule, the Supreme Court in its Bommai judgement (1994) adopted a much more restrictive reading of the conditions under which President's Rule can be legitimately invoked. Consequently, the incidence of President's Rule has reduced considerably (Swenden 2016b, 500-501). In 2016 the Supreme Court not only struck down the application of President's Rule in the states of Uttarakhand and Arunachal Pradesh, but reinstated the two disposed Congress-ruled governments, an action it had never undertaken before (Saxena and Swenden 2018). At the time of writing the state of Kashmir remains under a controversially imposed President's (Governor's) Rule. The de facto autonomy of the states has also been enhanced as a result of economic liberalization (Rudolph and Rudolph 2001), and the decline of the one-party dominant system, requiring national parties to make alliances with regionally based political parties.
In relation to non-territorial segmental autonomy, the retention of Muslim and Christian personal laws, already discussed under the mutual veto section of this paper, fits the criteria. It is notable that this does not extend to Sikh, Buddhist or Jain personal laws, constitutionally described as being 'Hindu'. However, as above, personal laws are coming under pressure in India. It is too soon to predict their abolition, but the changing political discourse surrounding the legitimacy of retaining laws is a threat to their continued existence. Constitutionally linguistic groups have the right to preserve their language even outside states where it is a recognised language (or, indeed, in the case of those languages such as Sindhi, where it is not a state language). However, as the Commissioner for Linguistic Minorities Report makes clear (2014) these constitutional provisions are upheld more in the breach than in the observance.
In relation to caste, no formal segmental autonomy provisions exist. The Indian constitution prohibits the practice of Untouchability and the 1955 Untouchability (Offences) Act specifies sanctions, including imprisonment, for those that practice it. The provisions for uplifting backward castes (SCs and OBC) as discussed in the proportionality section of this article are designed to undermine the occupational basis of the caste system. None of this however can detract from the fact that the divisions of society on the basis of the caste system are still a lived experience for the majority of the Indian population, especially in, but not confined to, rural areas. Lijphart (1996) claimed that India should be considered a consociational polity in practice. Yet his analysis was not based on a longitudinal and systematic empirical analysis of the extent to which the institutions and policies of the Indian state displayed the four key features of consociationalism: grand coalition, minority veto, proportionality and segmental autonomy. In this article, we provided an original and longitudinal systematic analysis. We did so by relating key institutions such as the Union Cabinet, senior political offices in the state, the composition of the Lok Sabha, the administrative services and the internal structure of polity-wide parties to the consociational yardsticks of grand coalition or proportional representation.

Conclusion
Similarly, we looked at the extent to which caste groups, minority religions or territories possessed the ability to exercise a minority veto or acquired segmental autonomy. Our research generates new findings which make an original contribution to the understanding of consociationalism in three ways.
Firstly, we extend and further deepen early criticisms of Lijphart by Lustick (1997), Wilkinson (2000) and Adeney (2002) who argued that it is a stretch too far to consider India as a consociation on all elements. Our data clearly bears out that minority groups (other than dominant minority groups such as Forward Hindu castes) are not systematically co-governing the centre as part of a grand coalition. Some representation has been tokenist -allocating key executive portfolios to representatives of different groups. Key groups (in particular lower castes and Muslims) remain under-represented in relation to their share of the population, be it in senior political, administrative or judicial offices. The gap between constitutional or legal provision and actual practice is apparent when observing the share of unfulfilled positions attributed to lower castes and tribes in the central administrative services, although the position continues to improve for these groups. The fact that Narendra Modi was an OBC is a sign of how India's de facto power sharing has opened up the political system to some groups. However, others (particularly Muslims) remain excluded despite the presence of other consociational elements such as the existence of personal religious laws.
Secondly, our data confirms the importance of assessing consociationalism for different segments of a society, in India as well as elsewhere. Such a disaggregated analysis gives mixed evidence for Wilkinson's assertion that India has become more consociational in the post-Nehruvian period. We observed a slight increase in the representation of OBCs at the expense of Forward Castes in the Union Cabinet, Lok Sabha and senior political offices. Linguistic reorganization and the enactment of a set of language laws between 1952 and 1967 also appeared to have 'locked in' linguistic autonomy and transformed the states into territorial political communities (Swenden 2017). Coalition government and liberalization post-1989 strengthened territorial autonomy, helped by the Supreme Court's resolve to strike down political abuses of President's Rule. However, the (limited) rise of the backward castes and the deepening of territorial autonomy is undermined by the gradual weakening in the representation of religious minorities, especially Muslims. This can be observed across nearly all pivotal institutions of the state: Cabinet, Lok Sabha, polity-wide parties, key political offices. Although Muslims have been underrepresented throughout India's independent history, their level of representation has deteriorated since the mid to late 1990s, mirroring the rise of the BJP.
Thirdly, the Indian case demonstrates the dangers of political systems leaving power-sharing to 'informal rules and practices' alone. If they are not sustained by enforceable constitutional and legal provisions, they run the risk of being overturned should the elite consensus change. Following on from the point above, to the extent that the polity -leaving aside society and the economyappears to shut out a minority group which constitutes more than 14% of the population, the credentials of India, not just as a consociational democracy (which as we argued it never was) but even as a 'liberal' democracy appear to be at stake. The declining level of support for minority rights demonstrated in the National Election Surveys suggests that this trend may continue.
Finally, our article identifies areas of further research. More research needs to be conducted on levels of consociationalism at the state level in India. The level of representation for religious minorities and lower caste groups varies considerably across state assemblies and especially in Southern states these groups are better represented (Jaffrelot and Kumar 2009). Sometimes this is the result of more generous reservation policies, including for Muslims who are more often subsumed under OBC quota. This demonstrates that formal measures of minority protection help secure the position of non-dominant groups in divided societies, a lesson for other countries. The Indian case also raises wider questions on the impact of the electoral system on minority representation and politics. Reservations have helped to lift the representation of lower castes and tribes in the legislatures, civil service and higher education. However, territorially dispersed minorities (religious minorities, Dalits) may have stood a better chance of political representation had they been elected under a form of Proportional Representation.