The limits of prescription: courts and social policy in India and South Africa

ABSTRACT This paper explores the social policy-making role of supreme courts in India and South Africa. It argues that both significantly shaped social policy. But neither imposed its will on elected government – both recognised that judicial power is limited and sought negotiation with the government and other interests to ensure compliance with rulings. Despite the difference between them, both courts promote and support collective action by the poor or their allies in civil society. The paper traces the institutional roots of the relative strength of the two courts and their relations with their governments and links their rulings to the political environment.

did they seek an adversarial relation with the governmentcommonalities between the courts and the executive were at least as important as their differences. Governments and ruling parties are not monolithic and court rulings can strengthen the hand of sections of government who support a greater social role, while judges often share broad perspectives with government office holders.
Central to our approach is the assumption that the role of the courts is essentially political. This view is often resisted by jurists and legal scholars who either insist that courts are guided by a purely legal logic or, at the very least, that their political interventions are constrained by their need to convince legal practitioners that their judgements are credible (Roux, 2009). But this does not gainsay the reality that court judgements are political interventions, in their effect and intent. On the first score, the rulings affect the distribution of (scarce) resources. They also help to shape the social policy environment by obstructing or assisting those within and outside governments who fight for the adoption of particular social policies. On the second, the view that courts in constitutional political orders simply hand down legal rulings which have no political intent is contradicted by the courts' history (e.g. Leuchtenberg, 1963). It is widely accepted in constitutional democracies that judges are influenced by their values and political positions, even if they are required to fashion judgements which will convince other lawyers: this is why appointments to constitutional courts usually require a political process.
This means that courts are one of the many actors who shape social policy in a political process. And, like all such actors, they are obliged to take into account the power of the other interests who also play a role in policymaking. We therefore argue that the courts in India and in South Africa have been able to effect social policy-making not because they tried to impose courses of action but because, in different ways, they avoided that the Indian court by engaging with the government, civil society and other actors in the policy community and issuing orders and judgments that they knew were realistic and feasible, its South African counterpart by concentrating on procedural remedies rather than stipulating how the process should end. This suggests that the effectiveness of courts depends not on their ability to issue instructions to governments and to society, but on recognising the limits of their power by taking into account that of other actors.
The paper will also point to differences between the two cases, of which the most important is the Indian court's willingness to prescribe policy outcomes in contrast to its South African counterpart's preference for introducing procedures rather than laying down the specific policies. Despite this difference, the success of both courts at influencing social policies relies on one common denominator: through different modalities, both have promoted and supported collective action by the poor themselves or their allies in civil society. The South African Court has mandated the government to engage in a dialogue with the poor and their representatives to find a shared solution to their grievances. The Indian Court has based its rulings on a prolonged dialogue with civil society organisations and the government of India. Hence, while in South Africa the Court has mandated engagement with the poor, the Indian Court has facilitated this dialogue before issuing an order. We argue that this is, at least in principle, more likely to produce sustainable social policy and to support democratic politics than dictation by the courts to the governments. Collective action by the poor themselves or their allies in civil society are, in our view, not only more conducive to democracy because citizens play a greater role in decisions but also far more likely to produce and sustain social policy which is sympathetic to the poor: courts which enable this action rather than dictating specific outcomes are thus likely not only to strengthen democratic decision-making but also to enhance citizens' ability to press governments to introduce social policies aimed at addressing poverty and to ensure that those policies endure.
The purpose of this paper is not to add to legal scholarship by offering a comprehensive account of the two courts' role in social and economic rights jurisprudencenor does it claim to canvass fully scholarly debates on the socio-economic judgements of the two courts. Because its purpose is to explore the political and policy impact of the courts' roles, it discusses only those cases and scholarly opinions which are relevant to this themea more detailed exposition and analysis of the South African case is contained in a published article by one of the authors (Friedman, 2016). To understand the specific role that the two supreme courts have played in shaping social policies, it is important to begin by pointing out some institutional and political factors that have constrained and shaped the courts' role.

Socio-economic rights in India and South Africa
In principle, socio-economic rights (SER) in India and South Africa's constitutions have very different statuses. In South Africa, a wide range of SERs is constitutionally protected, alongside civil and political rights (CPR). In principle, there is no difference between the two categories of rights, as, according to Section 7 (2) of the Constitution, the state 'must respect, protect, promote and fulfil the rights in the Bill of Rights'including SERs. However, the Constitution does differentiate between CPRs and SERs by attaching to the latter an escape clause that mandates the state to 'take reasonable legislative and other measurers, within its available resources' to fulfil these rights (emphasis added). This obviously makes the enforcement of SERs more dependent on government capacity and funding than CPRs. But it does not alter the reality that SERs are fully justiciableit is the court, not the government, which decides whether the recognition of SERs is reasonable in the circumstances. In India, the constitution clearly differentiates CPRs from SERs. The former are included in the 'Fundamental Rights' section of the Constitution; the latter were downgraded to 'Directive Principles', which are supposed to inspire the state's policies. Hence, CPRs were fully justiciable, whereas SERs were not.
In practice, however, despite the difference in constitutional provisions, the justiciability of SERs is roughly comparable in South Africa and India. The Indian Supreme Court, through successive rulings since the late 1970s, has expanded the scope of the (justiciable) Article 21 of the Constitution (which protects the right to life), to include, among others, the right to health, livelihood, education, shelter, drinking water and food (Deva, 2009, p. 25). What the constitution explicitly gives the South African court, its Indian equivalent has fashioned for itself through its jurisprudence. Both courts therefore enjoy the power to shape social policy. Their willingness to do so is, however, shaped by the context in which they operate, in particular the relation between the court and the government and the relative strength of the two parties.

The courts and the government
What enables and constrains constitutional courts when they instruct governments to fashion their social policies? Literature on the South African case offers a framework for addressing this question, which is also applicable to the Indian case (and, indeed, all others). It offers two competing explanations for the court's roleoddly, both are written by the same author, Theunis Roux. Both seek to explain why the court has been able to hand down rulings which may have inconvenienced the government (Roux, 2009, p. 106). First, he adopts an explanation which, although he does not say this, relies on an approach akin to game theory. He rejects 'political science accounts' which see the court's rulings purely as political decisions rather than interpretations of the lawbecause, as noted earlier, judges' legitimacy in the legal community depends on adopting 'forms of reasoning' acceptable to other lawyers (Roux, 2009, p. 108). This, he implies, implicitly sets up a confrontation between a court concerned with applying the constitutional norms of the legal community and 'the political branches' which may be offended by its rulings.
The court's goal is thus to find strategies which will enable it to play its role despite the likelihood that its ruling will force the government to do what it does not wish to do. It employs a combination of principle and pragmatism in its attempt to do what legal reasoning requires it to do while recognising the power of the government to bend the court to its will. It must tread carefully and pick its battles prudently (Roux, 2003, pp. 97-98). He uses this model to explain why the court has avoided, in the main, handing down rulings which tell the government what its social policy should be and has instead shown a 'preference for procedural remedies that promote political solutions when addressing social and economic rights claims' (Ray, 2011, p. 108). In the only case in which the court has instructed the government to adopt a policy when it ordered the provision of medication to prevent mother to child transmission of HIVthe government was 'politically isolated' and 'sliding toward an embarrassing political defeat'. The court thus had no reason to fear it and 'may even be said to have rescued it by providing an "objective" legal basis for the reversal of its policies' (Roux, 2009, p. 124/125).
The second view rejects the notion that courts and governments have fundamentally different goalsloyalty to the legal community's view of the constitution on the one hand and the desire to wield power on the other. While judges do have to take legal reasoning seriously, this leaves much latitude for personal political perspectives to shape judgements, which is why judges who win the respect of their colleagues are clearly recognised as 'liberal' or 'conservative'. If, then, the political positions of most constitutional court judgements were fundamentally at variance with that of governing politicians, the government would constrain the court (Leuchtenberg, 1963). So in this view the court's ability to continue 'correcting' government actions must be based not on the difference between its goals and that of the government but on their similarity.
This does not mean that judges are government lackeys. But it does mean that they must share enough of the government's values to make it likely that it will tolerate the court. In this view, constitutional jurisprudenceon SERs as well as all other issuesis not a strategic game between actors with different goals but a conversation between jurists and politicians who share goals, but differ on how to achieve them. Roux acknowledges this: 'Few constitutional courts anywhere in the world are independent in the strict sensecomposed of people with political views opposed to those of the governing political elite' (Roux, 2003, p. 94). While this interpretation of the judiciary's independence might appear too drasticindependence does not necessarily mean opposition -Roux is right when he points out that the South African court has comprised jurists who, while talented and independent, share a common 'social transformation project' with the government. Disagreement is about means, not ends. Indeed, Roux goes on to argue that, even when the court appears to be challenging the government, it is, in reality, seeking to assist it by ensuring that its agenda for change is procedurally fair and thus safeguarding it from political attack (since government opponents are unlikely to challenge a ruling which seems to constrain it by telling it how to implement its agenda) (Roux, 2003, p. 107).
At first glance this seems to be an elaborate conspiracy theory in which judges pretend to impose courses of action on the government, so hiding their desire to support it. But this is not required for this view to be accurate. All that is required is that judges who broadly support the government's goals should feel that they are best implemented if different methods were used. It is also worth noting that the approach which stresses competition between government and the courts tends to assume that governments share a single mind and approach, which they rarely do: even rulings which impose on some in government are, provided they remain within the shared value system, likely to be supported by others within government. The South African case supports this view. The vision shared by the courts and government can best be described as a common commitment to work to end the effects of apartheid, the system of racial domination which prevailed until 1994 and against which many post-1994 judges fought. This appears to explain both the government's willingness to accept the court's rulings and the court's reluctance to instruct it to take action it might refuse to take. The result has been a court which has ordered the government to revise policy but has, with only one exception, stopped short of describing what it wants put in its place.
In India, there was no such clear-cut consensus between the judiciary and the government. While there was a shared vision of a new India built after the colonial regime, there was no agreement on what that meant. Thus, while Nehru's post-independence government proclaimed its intention to build a 'socialist pattern of society', the courts protected the rights of elites, particularly the landed gentry. For the first decades after independence, India's main redistributive strategy was land reform. By and large, the project failed, if only because the governing party (the Indian National Congress) relied on the support of relatively rich landowners to distribute patronage and extract votes (Frankel, 2005). But another important reason for the failure of the Indian state to redistribute land to the poor was the Supreme Court's strenuous defence of the right to propertya Fundamental Rightover the pursuit of social justice and land redistribution (Austin, 1999). In practice, in contrast to South Africa, whose constitution protects property rights but also (in Section 36) enables the government to override them if doing so is 'reasonable and justifiable in an open and democratic society based on human dignity, equality and freedom' there was no shared commitment to the constitution as a 'transformative' document (Sunstein, 2000(Sunstein, /2001. This ensured a different relationship between the court and the government to that in South Africa. This relationship can be divided into two phases. The first was characterised by a strong confrontation between the two institutions: the main issue at stake was the socio-economic transformation that the Indian government wanted to promote, which was opposed by the constitutional court. The conflict became more intense in the late 1960s when Prime Minister Indira Gandhi moved to the left. A vicious cycle was established, in which 'the parliament [could] pass legislation, the courts [could] determine its unconstitutionality, the parliament [could] try to circumvent the courts by amending the constitution, the courts [could] pronounce that parliament ha[d] limited powers of amendment, parliament … and so on and on' (Mehta, 2005, p. 187). In the process, Gandhi tried to limit the independence of the judiciary (Maiorano, 2015, chap. 4); this was strenuously resisted by the court until its final capitulation during the emergency regime (1975)(1976)(1977), when it endorsed and legitimised the suspension of democracy imposed by the Prime Minister. This was a major blow to the Court's reputation.
However, the defeat of Gandhi's Congress party at the 1977 election was a turning point for the Supreme Court. In an attempt to regain the confidence of public opinion, the court sought to assume the role of the defender of the public interest, broadly understood. This attempt included the defence of the interests of the disadvantaged sections of the society (Mehta, 2007). It is in this context that it made SERs justiciable and introduced an instrument, public interest litigation (PIL), which was intended to give access to the court to the poor. It should be noted that defending the interests of the poor was just one aspect of the Court's strategy to regain public confidence. Its increasing interventionism in areas such as pollution, corruption and accountability contributed in a very significant way to increasing the Court's reputation among the middle class too. This is important for at least two reasons. First, this is a section of India's society that is extremely influential (Fernandes, 2006); and, second, the middle class is generally wary of increasing spending on the welfare of the poor, when this is seen to reduce the projects designed to benefit them. However, the Supreme Court enjoys respect among the middle class, which is more likely to stomach increased spending for the poor if the government is directed to do this by the court. In fact, middle class opposition to spending on the poor is also linked to a perception that this is invariably a way to distribute patronage and feed a corrupt system (see Rothstein, Samanni, & Teorell, 2012 on OECD countries). The court is much less likely to be seen as a dispenser of patronage than politicians. In South Africa, middle class resistance to anti-poverty measures exists but takes subtler forms, largely as a result of the salience of race in the society's politics. To oppose in principle measures to address poverty is to risk appearing to seek to perpetuate racial privilege, which has lost legitimacy in the public debate: claims that the government wastes money which could be spent on the poor is one weapon in the arsenal of those in the middle class who seek to discredit a majority government.
The two courts have thus been able to influence policy for different reasons. In India, the government and the court fought a war that the court first appeared to have lost when it submitted to the state of emergency but eventually won, when the electorate rejected Indira Gandhi after the proclamation of the emergency regime. After that, no Indian governmentand definitely not a Congress-led governmentcould afford to be suspected of undermining the independence of the Court. The war with the government was instrumental in putting the Court into a position of powerwhat does not kill you, makes you stronger. It has also been suggested that the Indian court has been strengthened by a political context in which no party was able to form a government on its own between 1989 and 2014. This severely eroded the central government's ability to centralise power and to interfere with the functioning of the country's institutions (Manor, 2016). The Court has been able to fill 'a governance vacuum' (Deva, 2009, p. 30) created by a series of weak central governments. In the South African case, as noted earlier, it was the commonality between the court and the political elite which has allowed the court to play a role in social policy. This makes the position of the Indian court far stronger than that of its South African counterpart since it has proven public support which constrains political attacks on it. South Africa's court has not tested its support and so it is possible that it would be unable to withstand a similar attack to that which its counterpart faced in India.
This may well have important implications for the respective styles of the two courts. We will argue that mandating negotiation rather than imposing outcomes shows a clearer understanding of the appropriate boundaries for a court in a democracy, partly because it respects the choices of citizens (in particular those of the poor). But that does not necessarily mean that this is why the South African court has chosen the former path and the Indian court the latter. The nature of its relationship with government may make the South African court more wary of intruding too much into policymaking, while the Indian court may have been emboldened by its experience. This shows that political context is a far more reliable determinant of the influence of courts than constitutional texts: South Africa's court has the formal power to impose outcomes but largely chooses not to. India's does not (on SER cases) but has fashioned out of political circumstances the latitude to do what its counterpart is legally entitled to do. In both cases, politics, not what the constitution says, is decisive.
In India, since the early 1980s, the government has abandoned land reform as its main poverty-reduction strategy (Maiorano, 2015, chap. 3), which had been a major issue of confrontation between the two institutions. Since then, the government has focused on specific policies and programmes to tackle poverty and the Supreme Court has been very active in trying to ensure their effective implementation, despite the fact that, as noted earlier, it did not have a strong history of intervening on behalf of the poor (Rajagopal, 2007). However, both the government and the Court share a 'minimalist' vision for India's development according to which at least something must be done for the poor. On the other hand, as we have already noted, the more cordial relation between the Court and the government is also due to their respective power positions. It has been argued that the Supreme Court was able to assert itself mainly because of the increasing fragmentation of the party system since the late 1980s (Rudolph & Rudolph, 2001;Shankar, 2009). However, the Court's activism started at least one decade earlier as a consequence, of its strategy of regaining the confidence of public opinion (Mehta, 2007); and, as a response to the growing assertiveness of a number of 'non-party political movements' during the 1970s which were empowered by the new ways to access the Court and tackle public interest issues in the presence of an unresponsive government (Ruparelia, 2013a). In this situation, the government simply did not have the strength to confront an increasingly popular and increasingly active Supreme Court. This was true during the 1980swhen major crises in Punjab, Assam and Kashmir kept the government busyand in the 1990s, when the fragmentation of the party system further solidified the Court's position. This helped to shape a relationship between the two institutions which was less adversarial and more collaborative. As in South Africa, the court's role is less about getting the government to do what it does not wish to do and more about trying to get it to pursue its own stated goals more vigorously. The Indian court is also arguably as sensitive of the limits of its powerand thus of the need to seek support for its rulingsas its counterpart, since it does negotiate its judgements with government and civil society organisations. In both countries the government and the constitutional court have a relation that is far less adversarial than analyses which stress the divergent goals of the two institutions suggest. This too helps explain why the two courts are able to play a role in shaping social policy.
Even in India, where the court has fought a battle with the government over its power to interpret the constitution, the 'game theory' view of the relationship between them, in which their competing goals requires the court to develop a strategy to maximise its influence, does not describe the dynamics adequately. While the Indian court may have achieved greater freedom to act by taking on the government, it, too, appears to recognise that its influence rests primarily on seeking to work with it. Throughout the world, courts are limited by their need to garner support in other institutions (Gauri & Brinks, 2008) and both cases confirm that courts are likely to influence policy only if there is a significant degree of consensus between them and the government.
The approach of the two courts However, despite these broad commonalities, the two judicial institutions have adopted differing approaches to the protection and extension of SERs which require further discussion. While the Indian court has, by and large, limited itself to asking the government to implement policies as they were meant to be implemented, it has sometimes assumed a direct role in policy-making and has even taken on some executive functions (Sathe, 2002). In South Africa, direct intrusions into the legislative domain occurred only twice. The court has sought to enforce the SER provisions of the constitution in two ways. First, it has applied the 'reasonableness standard' which, according to Roux, requires it 'to assess whether a social programme unreasonably excludes the segment of society to which the plaintiff belongs'. It relies on tests of fair process, not on determining policy outcomes (Roux, 2003, p. 97). Second, it has repeatedly ordered the state authorities to negotiate outcomes with the poor, in order to find a mutually acceptable solution to the dispute between them.
While both courts have been widely praised for their innovative and effective approach to the protection of SERs (Epp, 1998;Gauri & Brinks, 2008), neither has (despite much prompting in South Africa from activists and scholars) tried to define SERs by setting a 'minimum core content', by specifying the minimum which the government is expected to provide to comply with the SER provisions of the constitution.

South Africa
Since 1994, South Africa's court has intervened in a variety of cases with social policy implications. But its reputation for 'pro-poor' intervention is based primarily on two cases dealing with housing and health, respectively.
Given the amount of legal academic ink spilt on discussing the court's role in SER jurisprudence, it would be easy to conclude that it had devoted much of its time to it. Some of its more enthusiastic supporters tend not to mention that its earliest social and economic rights ruling, the 1997 Soobramoney case, was a refusal to come to the aid of a critically ill patient who asked it to order a public hospital to offer him dialysis treatment. In reality, it has handed down judgments which deal with housing, health, education, water and the financing of municipal servicesbut these have been relatively sparse. The view that the court has played a substantial role is, therefore, based on an assumption of quality rather than quantityit assumes that the limited number of rulings have wielded significant influence on policy and practice. This is probably true of all or most judicial systems in which SERs are justiciablewhile the impression is sometimes created that the courts can be a frequent remedy for people seeking to enforce their social and economic rights, their use is usually sparing, its effect judged by the ripples it sends through the legal system and society.
The first judgment on which the court's reputation is built is that in the 2001 Grootboom case. It was brought on behalf of homeless people living on a field from which the provincial authorities wished to evict them, and the court partly upheld a lower court judgement ordering the government to devise a housing policy which would provide for the needs of the plaintiffs and people in similar circumstances. The ruling had little or no impact on Irene Grootboom, the homeless person in whose name the application was brought, or her community. While the applicants were not evicted, they did not benefit directly from a government housing programme and, as critics of the court never tire of pointing out (much to the irritation of its defenders), Grootboom died without receiving a house (Tolsi, 2012). But the case has subsequently been used to secure court orders compelling municipalities to devise housing policies which would provide for the needs of the poor rather than to evict them (Legal Resources Centre, 2004). It has also been argued that the judgement fundamentally moved property law in a direction favourable to the vulnerable because it 'dislodged the normality assumption that an owner is entitled to exclusive possession of his property … ' (Albertyn, 2011, p. 597).
The second was the 2002 Treatment Action Campaign (TAC) case mentioned earlier, one of only two occasions on which it directly ordered the government to adopt a particular policy (rather than to initiate a process such as framing a policy or initiating negotiations). The context was TAC's sustained campaign to press the government to introduce a comprehensive treatment regime for people living with HIV and in particular to supply them with antiretroviral medication (Friedman & Mottiar, 2005). The ruling attracted attention both because of the extensive publicity which the battle received and because it ruled against the government on an issue on which it felt very stronglythe Minister of Health initially threatened to ignore the judgement if it went against the government (but was almost immediately forced to retract) (Roux, 2009, p. 124): The intrusive nature of the remedy sought, together with a climate of public distrust over the ANC government's policies on AIDS, made Treatment Action Campaign one of the most politically controversial cases to come before the (Court) in the first ten years of its existence … (Roux, 2009, p. 134) The other case in which the government was ordered by the court to adopt a specific policy was the 2004 Khosa case in which it ordered the government to extend social security benefits, until then available only to South Africans, to permanent residents from other countries (the plaintiffs were Mozambicans). While this could well have become a controversial ruling in a political climate resistant to extending any entitlements to foreigners, the number of people affected was relatively small and so the court's ruling went almost unnoticed.
Since those halcyon days, the court has, in the view of legal scholars and activists, failed to realise the constitution's potential for intervention on behalf of the poor and vulnerable. None of its subsequent rulings have attracted the enthusiasm which Grootboom and TAC evokeda decision in the Mazibuko case in which the court also rejected the argument that it should force the state to provide the poor with a 'minimum core' of services (in this case a significantly increased entitlement to free water) has also been attacked for ignoring the interests and needs of the poor (Dugard, 2008). Legal scholars have criticised the court's insistence on judging government actions by whether they are 'reasonable' rather than by whether they conform to minimum core contenta minimum floor of entitlements for the poor (Ray, 2011;Roux, 2003). In this view, the court has, in its desire to avoid antagonising government or private economic power-holders, failed to realise the promise of the earlier judgements.
This view ignores the fact that the court has handed down some significant rulings, of which perhaps the best-known is the 2009 Abahlali basemjondolo decision to strike down a provincial law permitting the eviction of shackdwellers on the grounds that the authorities had a duty to 'ensure that [residents'] housing rights are not violated without proper notice and consideration of other alternatives'. The case was brought by shack-dwellers' movement, Abahlali basemjondolo, which has been engaged in a bitter conflict with the KwaZulu Natal provincial government: the authorities have been accused of trying to drive the movement out of shack settlements, both because it challenged the local political elite and, later because it had successfully brought the court action (South African Civil Society Information Service, 2009). The court has also handed down rulings instructing local governments to engage with residents threatened with eviction (Wilson, 2011). In 2013, it overruled a decision by the governing body of a suburban school to exclude a pupil on the grounds that it had reached its capacity. This seemingly technical issue goes to the heart of some of South Africa's race and class divisionsthe power of (mostly white) suburban school governing bodies to exclude poor black learners. It has also intervened to strike down discrimination against women in customary marriage and inheritance.
Some of these rulings have constrained the government, others have done the same to private power-holders. The image of a court unwilling to intervene on behalf of the poor if this means trampling on the toes of the powerful is inconsistent with its record in the decade since Grootboom and TAC.
Why the court should have sought to play this roleand why the government has allowed it do thishas been discussed above. In contrast to India's court (see below), South Africa's has not overtly sought to negotiate its rulings with the partiesthe prevailing legal culture would find that unacceptable. But nor has it sought to tackle the government in an adversarial manner, as its critics would like. Whether this is calculated or a sign of its commonality with the government, it has been concerned at all times not to step outside the consensus that the goal of policy is to counter apartheid's effect on society. As a consequence, it has ruled on whether particular measures are reasonable given the government's stated goals and has sought to substitute negotiation for laying down a minimum core of entitlements. But this has not diminished its influence. On the contrary, it may be precisely why it has been able to come to the aid of people threatened with eviction or seeking greater access to public health care. And, as we will argue below, its insistence on negotiation, while it may be shaped more by prudence than innovation, is a greater potential advance for the poor than the prescriptive role which its critics urge it to pursue.

India
India's court's role in social policy-making has expanded since the introduction of the PIL. In Common Law systems, only those whose rights are directly affected by a dispute can approach the court. The introduction of the PIL in the 1980s by the Supreme Court allowed any person or organisation to approach the court through a letter or a petition in cases involving the public interest. Since then, the court has heard PIL on a wide variety of subjects, ranging from environmental and poverty issues, women and prisoners' rights, to Richard Gere's 'obscene' kiss of an Indian actress, Shilpa Shetty. 1 While the objective of PIL was to grant access to the Court to the disadvantaged, among the tens of thousands of letters received by it, only a handful has made it to trial (Gauri, 2009, p. 10). Moreover, NGOs and civil society organisations seldom choose to approach the court in their campaigns, as they find the process extremely costly and time-consuming (Shankar & Mehta, 2009). Only the richest NGOs can afford litigation as a constituent part of their campaigns: according to Krishnan (2003, p. 24), 56 per cent of the NGOs in the richest quintile use litigation, but only 7 per cent in the bottom quintile do. In fact, many PILs are introduced by individualslawyers in many cases. This does not mean, however, that NGOs and civil society organisation do not use Court's rulings in their campaign. Quite the contrary, as we shall see below, these are a very powerful tool in the hands of the supporters of an expanding welfare state. Only rich NGOs may be able to afford to go to court, but there are enough of them in India to ensure a significant number of SER-related cases. It is also possible for a single case to impact on a range of policies. Despite the constraints mentioned here, PIL is important as a 'judge-made human rights mechanism' (Birchfield & Corsi, 2010, p. 715). The most significant SER case on which the court has ruled was the product of a PIL.
In April 2001, the People's Union for Civil Liberties (PUCL) filed a 'writ petition (196 of 2001)' on the right to food in the Supreme Court. In 2001, vast areas of India were suffering from a severe drought for the third consecutive year. Starvation deaths and acute hunger were reported throughout the country. The petitioners argued that this was a violation of the right to life protected by article 21 of the Constitution, which, according to several Supreme Court rulings (e.g. Francis Coralie Mullin vs. The Administration, 1981), included the right to food. They also presented compelling evidence that food stocks had reached unprecedented levels and that millions of tonnes of food grains were left rotting instead of being distributed to droughtaffected households. The petitioners asked the court to order the government of India and six state governments to distribute food through the public distribution system and to provide drought relief provided for by the famine codes. In effect, the petitioners asked the government to respect its own laws and to implement its own schemes in the way they were meant to be implemented in order to respect people's fundamental rights.
An important factor often forgotten in the literature is the role of individual judges. One of the PUCL's lawyers (Colin Gonsalves) recalls that at the first hearing of the case he did not even have to argue. The presiding judge, Justice Kirpal, opened the hearing by saying 'this cannot be. We cannot allow the state of affairs to continue' (Gonsalves, 2011, p. 8). When the Attorney General, Soli J. Sorajbee, tried to argue that the Indian state did not have the necessary resources to end hunger, Justice Kirpal told him 'to cut the flab somewhere else' or 'we will tell you how to do it' (Gonsalves, 2011, p. 9). But only a few years before, a very similar petition (Kishan Pattanaik vs. State of Orissa 1989) had been dismissed because the judges felt that there was no reason not to trust that the government would take the necessary steps to avoid hunger. In the Right to Food case, however, the Court not only accepted the petition, but, in an order dated 23 July 2001, invited the petitioners to implead all other state governments, making the case a nation-wide litigation. Differences in approaches between judges partially explain the inconsistent approach of the Court to the protection of SERs (Rajagopal, 2007) and its selective intrusion into policy-making. In some cases, it felt that it did not have the authority to intrude, especially if this had important budgetary implications; at other times, it did not exercise this restraint (Deva, 2009, p. 37).
The right to food case also exemplifies India's Supreme Court's approach to the protection and promotion of SERs, which has been termed conditional upon state action (Khosla, 2010). Khosla rightly argues that the Court has seldom conceptualised SERs as absolute rights of every individual. Rather, its rulings have repeatedly emphasised that a violation of a given SER 'can only occur when the state undertakes an obligation, but does not fulfil it' (Khosla, 2010, p. 751). While this approach is arguably more modest than the South African court's much criticised 'reasonableness' standard, in some cases, the Court went well beyond that, assuming a role 'strikingly similar to law-making' (Birchfield & Corsi, 2010, p. 700) in two ways.
First, in the right to food case, in an order dated 28 November 2001, it converted eight food-related government schemes into legal entitlements. This was a major intrusion into policy-making. Government schemes in India are designed and implemented by line ministries, but are seldom backed by an act of the national Parliament. They can, at least technically, be discontinued or amended through a simple government order. Moreover, if beneficiaries of a government scheme feel that they are denied their due, they cannot go to court. A Supreme Court order in November 2001 radically changed this. Not only are the provisions of the schemes mentioned in the order fully justiciable; but, following another court order dated 27 April 2004, these schemes cannot be 'discontinued or restricted in any way without the prior approval of [the] Court'.
Second, the Court has ordered the government to amend some schemes, even when this had major financial implications. Perhaps the most striking example is the universalisation, in an order dated 7 October 2004, of the Integrated Child Development Services (ICDS), India's main policy to tackle child malnutrition. The Court ordered the government to open an ICDS disbursement centre in every single habitation in the country and not to restrict access to the programme in any way. This meant opening 800,000 centres in addition to the existing 600,000 and had major financial implications.
The Supreme Court has also expanded its ambit into the executive domain. In particular, it has entrusted the responsibility for implementing its orders (and therefore, of welfare schemes) to the Chief Secretary (the highest ranking officer) of each state government, creating, in an order dated 8 May 2002, an accountability structure that overlaps and sometimes collides with that designed by the state governments for the implementation of those schemes. It said, in an order dated 29 October 2002, that, if people die of starvation, 'the Court may be justified in presuming that its orders have not been implemented and the chief secretaries/administrators of the states/Union Territories may be held responsible for the same'. Second, in an order on the same date, it established its own monitoring system through the appointment of two Commissioners to whom it granted extensive powers, in particular to investigate any violation of the orders and to demand corrective actions from the state governments. The powers of the Commissioners go further, as the Court has empowered them to monitor any food-related 'measures and schemes', even if they are not specifically mentioned in the orders. The Commissioners' reports significantly influenced the shaping of the court's orders: when the Court suggested giving priority to the appointment of Dalits (former untouchable castes) as cooks in the Midday Meal Scheme, an important change since traditionally, upper caste people are not supposed to eat food cooked by Dalits, a programme which provides one cooked meal to every child enrolled in public schools, it did so on the suggestion of the Commissioners (Hassan, 2011, p. 5).
This intrusion into policy-making is, however, the exception rather than the rule. The Indian court's judgements often seem more intrusive than they really are because they often entail ordering the government to do what it is committed to doing but has not done. For example, in an order dated 28 November 2001, it ordered all state governments to implement the Midday Meal Scheme. Before the order, only Tamil Nadu and Gujarat were implementing it. While this seems like a major intrusion into policy-making, the Court was simply ordering that the scheme be implemented in the manner the government intended: all states were supposed to provide a cooked meal to all children enrolled in public primary schools. Similarly, it has issued extremely specific guidelines for the scheme, laying down minimum calorific requirements of the meals or the financial arrangements between the central and the state governments. But again these guidelines did not lay down a minimum core content of the right to food; they simply reiterated the government's own policy guidelines. The Court has, in most cases, asked the government to do what it had committed to do. This applies to most schemes covered by Supreme Court orders.
A further qualification stems from repeated statements by the judiciary and the government's representatives that the right to food case was not adversarial, but a collaborative project (Deva, 2009, p. 26). In reality, most Supreme Court orders are the result of a dialogue between the Court, the network of civil society organisations led by the PUCL, and the government. For example, at the beginning of the case, the Court had proposed to the government that it distribute excess food stock free of charge. The government replied that it could not afford that and proposed to impose a nominal price, which is what the court directed in its orders (Interview, N. C. Saxena, January 13, 2016). In another instance, the Court asked the government if it agreed with the number of ICDS centre to be opened as proposed by the petitioners. The government argued that the number was slightly too high, but in this case the Court, in an order dated 7 October 2004, argued that the evidence provided by the government was not convincing and it endorsed the view of the petitioners. That the orders are not impositions on a reluctant government can also be inferred from the occasional willingness of the government to go beyond the guidelines issued by the Court. For example, the government autonomously revised the minimum calorific requirements of the Midday meals (from 300 to 450 calories) without any input from the Court.
Another sign that this was not an adversarial process is that the government showed no interest in introducing legislation to nullify the orders. Many of the court's rulings are based not on an interpretation of the constitution, but on its reading of existing laws. In these cases, it can override the court: the most notable was the Shah Bano case in the mid-1980s, when Parliament voted to overcome a Supreme Court judgment granting the right to alimony to a divorced Muslim woman. Not only has Parliament not found the intrusion of the Court into policy-making in the right to food case inconvenient enough to amend the lawsince 2004, it has passed laws which incorporate many provisions in the Court's orders, a sign that there is a relative consensus between it and the court on right to food issues.
Nor is the right to food case an isolated example of apparent consensus. According to N. C. Saxena (the Commissioner to the Supreme Court), in most cases the government is quite happy about the court orders. (Interview, Saxena) Committing resources to the poor has been very difficult for Indian governments, as it prompts strong reaction from a section of the middle class which sees welfare schemes as 'handouts' at best, patronage politics at worst. 2 However, if the Supreme Court orders a welfare measure it is easier for the government to commit the resources and, more generally, to take what would otherwise be politically difficult (perhaps even unfeasible) decisions. Even when the Court goes beyond the government policy guidelines, this is the result of the triangular dialogue between civil society organisations, the state and court (including the Commissioners).
Much of the dialogue between the Court and the state governments is mediated by the Commissioners. They are immensely respected figures who, through formal and informal contacts with key implementers and policy-makers at the state level, are able to negotiate important policy correctives that often prevent further intrusions into policy-making by the Court. In fact, much of the 'effectiveness in achieving implementation often stems from [the Commissioners'] diplomatic relations with the state governments' (Birchfield & Corsi, 2010, p. 729).
To sum up, India's Supreme Court's conceptualises SERs as conditional upon state action. Rather than adopting a minimum core content approach to define what a given SER means, the recurring feature of the Court's rulings is to push for the proper implementation of those policies that the government had committed to implement on its own initiative or to amend those policies on the basis of an evidence-based dialogue with civil society and the government.
Less is more: the importance of citizen agency In theory, a court which tells the government what to do would seem more likely to ensure egalitarian social policy. But both theory and practice suggest that it is less likely to do this.
At the normative level, the training and expertise of judges does not equip them to choose between competing social policy options. Democratic principle requires that decisions be taken by citizens, either directly or through elected representatives. The 'proceduralisation' of social and economic rights (Ray, 2011, p. 107) is thus a recognition that policy should be made by elected representatives, provided that they follow democratic procedures. A court which orders a public authority to engage with plaintiffs allows citizens to decide rather than imposing a 'minimum core' for social services on them. It also enables the agency of the poor by ensuring that their organisations enjoy a say in the outcome. It understands 'transformative constitutionalism' not as 'achievement of certain tangible results or outcomes' but as 'the radical change of the institutions and systems that produce results themselves' (Solange, 2011, p. 456).
On the first score, respect for elected representatives, the American legal scholar Cass Sunstein argues that the South African court's stress on procedure solves the problem of how courts can intervene on the side of the poor without removing the prerogatives of elected governments. It was respectful of democratic prerogatives and of the limited nature of public resources, while also requiring special deliberative attention to those whose minimal needs are not being met. It therefore did not pre-empt democratic deliberation, but ensured 'democratic attention to important interests that might otherwise be neglected in ordinary debate' (Sunstein, 2000(Sunstein, /2001. On the second, it has been argued that a stress on negotiation is an 'emphasis on participatory democracy and the ability of procedural remedies to democratise the rights-enforcement process' (Ray, 2011, p. 107).
But the stress on negotiated outcomes is also more likely to ensure more enduring and sustainable social policy. Policy is rarely sustainable unless those who benefit from it can act collectively not only to achieve policy change, but to ensure that it is implemented. This is illustrated by the fact that, almost a decade after the South African court's celebrated Grootboom ruling which held that people may not be evicted unless housing is available, residents of the Johannesburg inner city were still fighting evictions (Wilson, 2011). The judgement's effect was limited because it was not accompanied by collective action by the original plaintiffs or by others in the years after the judgement. By contrast, the TAC judgement, which was accompanied by collective action, was followed by substantial increases in public provision of the therapy sought by campaigners (Thom, 2013). And a successful campaign to prevent evictions in inner cities (Wilson, 2011) shows that even a modest infusion of collective action into housing disputes can produce very different results from those in Grootboom.
Stuart Wilson, one of the lawyers who fought the legal battle on behalf of residents, argues that recourse to the law was a product of tenants' weakness the 'organisational resources' needed to sustain a successful campaign 'were simply not present' (Wilson, 2011, pp. 137, 138). But there were 'grassroots organisations' operating in the inner city who attempted to link residents to legal assistance (Wilson, 2011, p. 140). After the court mandated engagement between the authorities and the plaintiffs, the parties negotiated and reached a settlement which 'represented an almost comprehensive surrender on the City's part' (Wilson, 2011, p. 148). The court's ruling mandating engagement is credited with ending the evictions and enabling residents to exert at least some influence on their future. In another case in which negotiation was mandated, a non-governmental organisation assisting people who were evicted reported that: 'It was only after the Court ordered engagement over the details of the eviction process itself … that the government finally took seriously … a key demand of the residents' (cited in Ray, 2011, p. 113).
While courts cannot substitute for citizen organisation, where even weak organisation exists, rulings have assisted collective action which seeks social policy more attuned to the needs of the weak. The result is likely to be more effective policy gains than those which might be accompanied by simply mandating policy outcomes.
Despite the fact that the Indian court has occasionally intruded into the legislative and executive domains, its approach is closer to that in South Africa than might appear from a reading of its judgements. Its rulings have been based on a dialogue with both the government of India and civil society organisations (led by the PUCL) that negotiated on their behalf. Given PUCL's and the Right to Food Campaign's (RTF, a nation-wide network of activists, academics and concerned citizens that has played a prominent role in supporting the right to food case) long experience and involvement with the poor at the grassroots, this has ensured that what the judges ordered was not only reflective of the views of an important section of the poor. It also ensured its legitimacy in the eyes of the central government which, because it was party to the judicial process, did not see the Court's orders as the policy choices of a few unelected judges.
Of course, simply mandating negotiation does not ensure that citizens in need of social policy interventions will be able to negotiate on equal terms with the authoritiesthe balance of power is clearly stacked against them. But courts do not need to restrict themselves to ordering negotiationthey can also impose on the authorities obligations which make it more likely that they will take seriously the concerns of their bargaining partner (Ray, 2011, pp. 125, 126) This suggests that effective social policy can best be protected by insisting on a 'minimum core'but of engagement, not of substantive outcomes. This is unlikely to substitute for severe power imbalancesbut it can act as a catalyst for effective collection action to achieve and maintain social policy gains.

Conclusion: the effectiveness of modesty
The evidence presented here suggests that courts can contribute to a more expansive social policy regime capable of addressing not only poverty and inequality, but also power imbalances of which they are a symptom. But it suggests too that their role is modest. Not only have they been most effective when they have supported the collective action and organisation of activists and the poor, but a key to their influence is a recognition of the limits to their power.
Despite the different approach of the two courts to mandating outcomes, both have influenced the shaping of their countries' welfare regimes by recognising the limits within which they are forced to function. Even when courts seem to enjoy great power on paper, they have no power to impose their rulings on governments or society. They rely on the tacit or express consent of those who their rulings affect and so, as Roux (2009) points out in relation to governments, they are forced to find ways of ensuring that their rulings are accepted and implemented which recognise the power of those who they judge. Not only is simply imposing rulings likely to be resistedit could delegitimise the authority of the court if its orders are regularly disregarded. A purist view of rights is likely to make it more difficult to entrench them: 'creating rights which cannot be enforced devalues the very notion of rights as trump' (Deva, 2009, p. 36). Once courts have ruled, their rulings are only likely to be implemented by a reluctant government if the organisations and activists which sought the judgement act to ensure that what the court orders is done (Friedman & Mottiar, 2005).
Whether courts tell governments what to do and whether governments listen is also shaped by the political context. This is illustrated by the South African TAC case discussed above and by the response to the Indian Court's order to the government to universalise the ICDS in November 2001. At the time, the order to universalise ICDS was completely ignored with no consequence. Two subsequent orders issued in 2004 were substantially ignored too. It was only with the order of December 2006 that the government started to act. This was not due to a sudden realisation that it was supposed to respect the court's orders but the changed political landscape. At least four factors contributed to open up a window of opportunity. First, the Congress-led United Progressive Alliance (UPA) had included the universalisation of the ICDS in its electoral manifesto. Second, in late 2006 two reports (FOCUS 2006;NFHS-3 2006) were published that showed that almost half of the country's children were undernourished. They received national (e.g. The Hindu, 2006) and international media attention (e.g. New York Times, 2007). The extent of child malnutrition revealed by the reports shocked the country, especially the urban middle class (Interview, Neerja Chowdhury, New Delhi, 9/8/2013). Third, the two reports galvanised civil society groups to campaign for the respect of the Supreme Court's orders and to increase efforts to tackle child malnutrition. Fourth, civil society activists had penetrated policy-making agencies since the victory of the UPA at the 2004 elections. In particular, the key actors were grouped around the RTF Campaign. Many people associated with the campaign were members of the National Advisory Council, some others were advisors to the Planning Commission or ministries, while still others worked with the Special Commissioners to the Supreme Court. The Commissioners themselves have strong personal and professional links with activists of the RTF. In other words, the ideas generated within the RTF Campaign penetrated to a significant extent into policy circles, and their pressures were effective in bringing the attention of the government to the issue of child malnutrition and the respect of the court's orders.
Eventually, the ICDS was nearly universalised and the budgetary allocations were increased by more than 450 per cent between 2004 and 2005 and 2013 and 2014. A similar story explains the universalisation of the Midday Meal Scheme. The Court's orders were ignored for a long time, until the RTF Campaign and other organisations started pushing for respect of the orders and a relatively progressive new governmentor at least influential members within it or influential upon itwas open to be pushed in that direction.
Even then, however, the Supreme Court's orders are far from being fully implemented. Its Commissioners have repeatedly shown in their reports (www.sccommissioners.org) how the orders are regularly violated and how implementation varies widely across the country. In particular, the Court can do very little to address administrative incapacity and/or apathy, especially in a federal context like India's, where the implementation of government schemes is in the hands of 29 different state level administrations (see Tillin & Pereira, this special issue). In South Africa, the government has always accepted the court's orders but implementation has often been tardy (Berger, 2008) and activists have been required to fight almost as hard for the implementation of the ruling as they did to win the change it introduced.
While public opinion enables or restricts the role of the courts, they are able at times to shape that opinionthe Indian court in particular has influenced social policy by stimulating debate on the rights of the most disadvantaged sections of the society. The most visible outcome of this debate has been the promulgation of right-based laws during the UPA governments (2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014). These included the Mahatma Gandhi National Rural Employment Guarantee Act (2005) that established the right to work; the Right to Education Act (2009), which made education free and compulsory for children aged 6-14 years; the National Food Security Act (2013) which significantly expanded food security programmes and gave legal backing to the Midday Meal Scheme and the Public Distribution System and to a few provisions of the ICDS. Supporters of these laws within and outside government used the court's rulings to pressure policy-makers, informally (Interviews, activists and officials, New Delhi between 1/2013 and 8/2013) and in Parliament: 'virtually all of India's new rights-based acts credit prior landmark judgments by the Supreme Court' (Ruparelia, 2013b, p. 36). Its rulings have also forced the government to justify its policy priorities to public opinion. Thus, the government has only rarely argued that financial constraints are a good reason to limit its commitment to the welfare of the poor, as doing this would have called for an official explanation of why, for example, it spent about 5 per cent of the GDP in tax concessions for the better off (Drèze & Sen, 2013).
In South Africa, the court has operated mostly within an implied consensus between it and the government (as well as important sections of public opinion). For this reason, its attempt to influence the public debate has been less obvious. But, besides its interventions in the housing, health and education debates, activist organisations have sought to use it not only in the hope of securing favourable rulings but also as a means of publicising unpopular causes such as protecting the poor from evictions.
This context explains both the limits and possibilities of courts' role in social policy. It shows, as this article has repeatedly stressed, why courts can only influence policy if they recognise the limits of their power and respect both the mandate of elected governments and the agency of citizens. But it suggests too that, if they share a perspective with the governing elite, and are able to rely onor createsignificant public legitimacy, they can play a key role in nudging society towards social policies which are more likely to recognise the needs of the poorand, if South African experience is a guide, to enable citizens who are able to act collectively to do so more effectively and so to ensure that policy is not only about the poor, but is a product of their choices and actions.