1 Situation Up to 2008

The period up to 2008 saw a number of efforts at defining the parameters of social protection in Pakistan. The most important among these was a review undertaken by a task force on social protection working with the Planning Commission, which was mandated to draft a social protection strategy for the country. The task force completed its work in 2006, and its report was published by the Planning Commission in June 2007. The N ational Social Protection Strategy (NSPS) of 2007 was formally adopted by the government, but there was little or no progress towards its implementation.

The NSPS 2007 can be taken as a useful starting point as it is the first comprehensive official statement with regard to social protection which was based on a detailed review of existing p rogrammes and interventions. The strategy defined social p rotection as:

...a set of policies and programme interventions that address poverty and vulnerability by contributing to raising the incomes of poor households, controlling the variance of income of all households, and ensuring equitable access to basic services. Social safety nets, social

Economic & Political Weekly

EPW

insurance (including pensions), community programmes (social

funds) and labour market interventions form part of social protection.

(Government of Pakistan 2007a: 14)

Moreover, the document lays out in broad terms the framework for the goals of social protection policies. The first of these is close to the original provenance of social protection as mitigation for risk and uncertainty: the government needs to address market failures in risk pooling and insurance. The second relates to a direct role of government in reducing social inequity through income transfers, asset build-ups and other redistributive interventions. Finally, the NSPS 2007 argues that the government might be interested in countering social exclusion and marginalisation through the promotion of social mobilisation of the poor. These three rationales read together provide a very broad and comprehensive approach to social protection which encompasses conventional social insurance as well as the more ambitious goal of social transformation (Devereux and Sabates-Wheeler 2004).

References to redistribution and social exclusion notwithstanding, the NSPS 2007 for the most part remained firmly embedded in the “risk mitigation” paradigm. The review of social protection programmes, for example, is focused mainly on social assistance and social insurance interventions, respectively. The main programmes under social assistance were Zakat, or cash transfer funded from a religious levy, Pakistan Bait-ul-Maal (pbm), tax funded cash transfer, Tawana Pakistan, or healthy Pakistan, a school-feeding programme for girl students, and an untargeted wheat price subsidy. Under social insurance, the main programmes were the Employees Old-age Benefit Institution (EOBI), the Workers ’ Welfare Fund (WWF), and the Employees’ Social Security Institution (ESSI), all of which were funded using payroll levies on employers. In addition to these, the review of schemes in NSPS 2007 i ncluded microfinance and public works programmes.

As the NSPS 2007 (Government of Pakistan 2008b) was a landmark document for defining the parameters of social protection in Pakistan for the first time, it is useful to identify what was missing. Most of the programmes included in the NSPS 2007 r eview were federal government programmes. The document made approving mention of informal social safety nets and private philanthropy, but these could not clearly be included under the rubric of social protection. What was missing from the list were provincial government interventions, asset transfer schemes such as those for the allotment of state land, land redistribution, or regularisation of housing rights, and the considerable social protection measures enjoyed by segments of state employees, particularly those with service in the defence sector (Sayeed 2004; Siddiqa 2003).

The NSPS 2007 selection of social protection interventions also contrasted with the claims of the Poverty Reduction Strategy P aper (PRSP) which counted a range of budgetary expenditures as “poverty-reducing” (Government of Pakistan 2009b). The PRSP assigned a wide range of sectors as representing povertyreducing expenditures without any analysis of the poverty impact of expenditures on these diverse sectors such as infrastructure construction, education including higher education, and law and order. The PRSP met its targets of taking poverty-focused e xpenditures to above 4% of gross domestic product (GDP) by 2005-06, much of this increase occurred in sectors where the poor or the vulnerable were not direct beneficiaries.1

The budgetary calculations of the NSPS 2007 took the more realistic route of counting only direct social assistance – cash transfers, school feeding, social care services and workfare – as part of the social protection outlay. It was found that the federal government spent around Rs 11 billion for social assistance in 2006-07. The bulk of this spending was in the form of Zakat (Rs 5.9 billion) and the FSP which was in fact a cash transfer under pbm (Rs 4 billion). Expenditure, specifically ear-marked for social security, welfare and social safety nets was projected to decline from 0.3% of GDP, or around 7% of PRSP spending in 2002-03 to 0.15% of GDP and 3% of PRSP spending in 2007-08 (Government of Pakistan 2008a).

The NSPS 2007 also reported on studies of existing cash transfer programmes which found serious gaps in targeting mechanisms and flaws in design and implementation. It was recommended that cash transfers programmes ought to be consolidated into a unique social protection database from which beneficiaries should be selected. The NSPS 2007 further recommended a transition to conditional cash transfers and the use of methods such as proxy means test-based surveys, geographical and community-based targeting for beneficiary identification. It failed to specify in detail the organisational mechanisms that were necessary to affect these changes, and left its recommendations at a generic level. It was strongly recommended that provincial governments be involved as stakeholders.

Reforms

The NSPS called for a gradual implementation of reforms spanning five years during which it was envisaged that the expenditure on social assistance will go up from Rs 11 billion to Rs 36 billion. The bulk of the recommended change was to come in the form of an increase in conditional cash transfers, and a relatively large allocation (of Rs 15 billion) for employment schemes under the public works. It was recommended that cash transfers needed to be consolidated into one streamlined scheme under a unified ministry for social protection. For public works some existing allocations could be used, but with organisational changes which would allow more direct stress on and accountability of actual person days of employment made available and utilised by the poor. The total number of beneficiaries of social assistance programmes was to rise from 2.6 million households to over six million.

The NSPS 2007 noted the limited reach of existing social insurance schemes in an economy dominated by informal employment. It called for streamlining of existing social insurance programmes, a switchover to employee contribution-based programmes, and the expansion of coverage.

Summing up, the NSPS 2007 represented a major advance in Pakistan in policy thinking about social protection. The strategy proposed an encompassing vision for social protection for the first time. By framing a number of apparently disparate schemes and programmes under the rubric of social protection, the NSPS 2007 was able to offer post hoc coherence to existing outlays and interventions. Although its conceptual framework included

july 9, 2011 vol xlvi no 28

2 The Change and Its Antecedents

Although the NSPS process had been completed by around 2006 there was a delay on the part of the government in endorsing the strategy and officially publishing it until mid-2007. The cabinet endorsement of the NSPS 2007 was accompanied by an instruction to the Planning Commission to “catalyse the implementation process”. This was one way of shelving the recommendations, since it was the cabinet itself which would have to make the relevant decisions regarding resource allocations and programme streamlining as virtually all of these decisions required cross- departmental coordination. The NSPS 2007 recommendation of creating a new integrated ministry for social protection too was not a task which could conceivably be initiated by the Planning Commission.

The political and economic turmoil in Pakistan from around 2007 onwards, which culminated in the election of a new government and the exit of direct military rule, however, proved to be a powerful catalyst for change. Soon after assuming office, the federal government announced the launch of the BISP, and in its very first budget which was presented to parliament barely 10 weeks after government formation, an amount of Rs 34 billion was allocated for this programme (BISP 2010). At around the same time the Punjab provincial government announced its FSP and a subsidised bread scheme with a total outlay of Rs 22 billion. There was an element of political competition between the federal and the Punjab provincial governments in the launching of these programmes. Although nominally in alliance, the parties leading the two governments Pakistan Peoples Party (PPP) at the federal level, and the Pakistan Muslim League-Nawaz (PML-N) in the Punjab province also eyed each other as rivals.

Whatever the immediate political motivations, it was clear that something quite dramatic was underway. The NSPS 2007 recommendations of raising fiscal allocations to social assistance programmes to Rs 36 billion over the course of five years seemed meek compared with the bold increase in the first year amounting to a total of Rs 51 billion. This promised fiscal allocation was in addition to the existing programmes, since there was no discussion as yet about reducing or streamlining existing programmes. In nominal terms the total outlay for social assistance was going to rise from Rs 11 billion to around Rs 62 billion – or a nearly sixfold rise. Moreover, the BISP went beyond any other previous social protection intervention in Pakistan and identified women as its primary beneficiaries.

Economic & Political Weekly

EPW

In the event the increased fiscal allocation was not as large as promised. Out of the Rs 34 billion set aside for the BISP the programme was able to utilise around Rs 14 billion by the end of the financial year (Government of Pakistan 2009b)2. This was mostly as a result of delays in setting up acceptable implementation mechanisms. The Punjab FSP – not to be confused with the cash transfer programme with the same name run by the Punjab Baitul Maal – utilised around Rs 14 billion. The pre-existing cash transfer programmes, in the meanwhile had remained steady at Rs 9 billion. The total cash transfer outlay, therefore, was Rs 39 billion compared with historical amounts of Rs 9 billion. Taking the high rate of inflation (around 25% annually) into account the fiscal year 2008-09 represented not a sixfold, but a threefold increase in cash transfer outlays (Figure 1). While not quite as ambitious as claimed in the first instance, the change was nevertheless quite dramatic.

Figure 1: Federal and Punjab Provincial Cash Transfer Disbursements

(billion rupees, current prices)

45 40 35 30 25 20 15 10 5 0 Zakat BISP Punjab-FSP PBM 2001-02 2002-03 2003-04 2004-05 2005-06 2006-07 2007-08 2008-09 2009-10

Source: Authors’ compilation based on Government of Pakistan (2003, 2004, 2005b, 2007d, 2009b, 2011a, 2011b).

What is remarkable is that unlike the period leading up to the drafting of the NSPS 2007, economic conditions considerably worsened from the fiscal year 2007-08 onwards. There was a macroeconomic crisis culminating in rapid price increases, falling exchange rates and increasing unemployment. The roots of the economic problems faced by Pakistan from 2007-08 onwards were broadly similar to those encountered by many countries suffering from the financial crisis. In Pakistan too, the financial sector had absorbed and then created high levels of liquidity, leading to inflationary bubbles in asset prices. There were domestic price rises some of which corresponded with international price movements in basic commodities and fuels, while others were caused by local shortages and regulatory failures. In August-September 2008 the Pakistani currency came under a great deal of pressure and depreciated by a third against the dollar. Soon after negotiations were concluded with the International Monetary Fund (IMF) and Pakistan entered a stabilisation agreement.

The political parties that came into office in 2008 had already been critical of economic management under the previous government, and particularly of the perception of increases in income inequality. There was strong political will at that moment, therefore, to combine stabilisation with measures for protecting the consumption levels of the poor and the vulnerable. The argument was made that painful price rises had to be ameliorated for the poorest.

By default, then, Pakistan began its first foray into what was effectively a counter-cyclical social protection policy. Direct transfers to the poorest had to increase at a time of economic downturn and stabilisation. The position taken by the Pakistan government in its negotiations with the IMF was based on recommendations of various working groups, all of whom had argued for enhanced social protection spending as an essential compensatory feature of cutting subsidies elsewhere in the economy. A panel of economists convened by the Planning Commission, for example, argued for not only a continuation but an expansion of the BISP (Government of Pakistan 2010b). The IMF for its part not only accepted these positions of the Pakistan government, but made its first tranche conditional on agreement between the government and the World Bank over the enhancement and streamlining of the cash transfer programme.

Two factors were important in the unprecedented policy interes t and political will for social protection in Pakistan in 2008. The political mobilisation for change of government, and election campaigns of political parties raised the salience of economic inequality under conditions of growth and then vulnerability under conditions of crisis. Political leaders were already familiar with the arguments for social protection, and some of them had understood and accepted the fiscal implications of any policies or programmes that might have had an impac t on significant numbers of people. This prior policy thinking, even if it was sketchy and fragmented in the minds of some political leaders and more developed in the engagements of others, alread y reflected a paradigm shift. Until now social protection had been seen as a minor and relatively unimportant “giveaway”. The acceptance of the idea that anything worthwhile needed to be at a large scale had the potential for far-reaching consequences.

The second factor was the acuteness with which the economic crisis affected the Pakistani economy, and the widely held view that inflationary spirals in essential commodities had badly hit the poorest. The need to “do something” was difficult to argue against. These two factors propelled the policy shift with respect to large-scale direct social assistance. The prior work done by NSPS 2007 allowed these political impulses to be retrospectively framed as part of a social protection strategy. The fact that the NSPS 2007 had already laid down some of the broad parameters of international current practice, meant that the ideas of Pakistani political leaders and their advisers found a ready-made frame of reference which could be used to convert political impulse into concrete policy actions.

3.1 Benazir Income Support Programme

Although BISP was allocated funds in the fiscal year starting in July 2008, the first disbursements did not begin until October. The main cause of delay was the absence of effective implementation mechanisms on the ground. The core of the BISP was a monthly cash transfer of Rs 1,000 ($12) to 3.4 million beneficiaries. Ever-married women were identified as the primary beneficiaries,

62 and ever-married men could apply for inclusion in cases where no ever-married woman was a family member. Families were defined as nuclear units consisting of parents and children. The decision to identify women as primary beneficiaries represented a break from past practice and institutional habit of constructing the relationship between the state and individuals through the male heads of family. The fact that other interventions such as a land grant programme were also targeted specifically at women for the first time signalled that the initiative had been taken or supported at the highest level in the top political leadership.

Table 1: Comparison of BISP with Income Support Components of Zakat and Baitul Maal

Zakat Baitul Maal BISP Total

2007-08 Beneficiaries (000)	607	1,457	0	2,064
Disbursement (million rupees)	1,141	4,371	0	5,512
Mean payout (rupees)	1,880	3,000		2,671
2008-09 Beneficiaries (000)	241	754	1,756	2,751
Disbursement (million rupees)	882	2,263	14,003	17,148
Mean payout (rupees)	3,660	3,001	7,974	6,233

Source: Author’s compilation from Government of Pakistan (2009b).

Political attention to BISP coincided with the decline in existing federal government cash transfers (Table 1). In fact, while BISP led to a major increase in federal spending on cash transfers the total number of beneficiaries of federal cash transfer programmes increased by only a third in the first year. The main quantitative change due to the BISP in its first year was in the size of the cash benefit paid out per beneficiary, which more than doubled. As the programme consolidated, however, its fiscal impact and outreach clearly began to dominate previous programmes (Tables 2 and 3). An analysis of federal government spending on untargeted food subsidies (Table 2) shows that even in 2009-10 the latter continued to command fiscal resources which were comparable to the BISP.

Table 2: Federal Government Spending on Cash Transfers and Untargeted Food Price Subsidies as Proportion of Total Government Spending (%)

2006-07	2007-08	2008-09	2009-10
Zakat	0.3	0.2	0.2	na
Baitul Maal	0.2	0.2	0.1	na
Targeted cash transfer (BISP)			0.9	1.8
Untargeted food price subsidies	0.6	2.2	1.2	1.4

Source: Author’s calculations based on Federal Budgets and Economic Survey (various years).

By 2009-10, fiscal allocations to BISP reached Rs 50 billion ($590 million).3 The programme was reported to have reached some 1.8 million women beneficiaries in 2008-09, and estimated to have included another million women in its second year of

o peration. If the average beneficiary had four dependents, the reach of the programme would have extended to over a tenth of the national population (Table 3).

Table 3: BISP Allocations and Beneficiaries

Disbursement (billion rupees)	Beneficiary Families (millions)	Proportion of Population (%)
2008-09	14	1.76	5.2
2009-10	28	2.75	6.9

Source: Author’s calculations based on Government of Pakistan (2010c), BISP (2010).

The initiation of the BISP posed a major challenge to the government implementation machinery. While until 2008 the main constraint to the expansion of social protection had been the absence of political will and policy commitment to effect

july 9, 2011 vol xlvi no 28

SPECIAL ARTICLE

significant resource allocations, once resources were made available organisational gaps became conspicuous. The BISP was initially set up as a special project under the annual development plan. This was, clearly, a provisional measure. Within the course of a year the BISP had been brought under its own law and set up as a programme with funding from the cabinet division. This new l egal and institutional cover for the BISP signalled not only political ownership and policy commitment, but also an acknowledgement that the existing organisational arrangements for delivering cash transfers, Zakat and Baitul Maal, were not to be utilised.

Zakat is not tax financed, but funded through the collection of a religious levy on bank accounts held by Muslims. Beneficiaries too must be Muslim. While the religious levy was mandatory at the outset, account holders can now opt out. This has resulted in a secular decline in Zakat collections and disbursements. Zakat beneficiaries are selected by local mosque-based committees. Their coverage across the country is neither systematic nor uniform. Baitul Maal is a tax-financed transfer which is administered through district level government-nominated committees. Committees decide on applicants’ eligibility against loosely defined criteria. Independent assessments of Zakat and Baitul Maal, statistical as well as qualitative, have raised serious concerns about inclusiveness, efficiency and fairness (Kabeer et al 2010). Besides these technical and organisational concerns, there were practical political considerations at work. The committees for Zakat and Baitul Maal beneficiary selections were thought to have been stacked by nominees of the previous regime.

Initially it was thought that BISP could be implemented quite largely using organisational structures already present elsewhere in government. It was considered, for instance, if it might be possible to use records on adult citizens available with the National Database Registration Agency (NADRA), which is an agency of the federal interior ministry charged with registering citizens for national identity cards. Some of the information held in NADRA, such as education level, could conceivably be used as a proxy. A list of filters was developed with the cooperation of NADRA to identify potential beneficiaries through proxy means such as education, age and reported occupation. It was further presumed that NADRA will be able to obtain information on bank accounts. These filters were mostly of a “negative” type – that people with a certain level of education, or bank accounts in foreign banks, were to be excluded from the beneficiaries. In some cases it might have been possible to make possible identification based on information provided by individuals to NADRA at the time of registration. It was considered, for example, that people who had reported their occupation as “labour” should be included. The reporting of occupations in NADRA was, nevertheless, considered to be less robust than schooling or education.

In the event the exclusive reliance on NADRA had to be abandoned when it was pointed out that as many as a quarter of eligible adults did not possess NADRA identity cards, and therefore, were entirely excluded from the database. Any system of beneficiary identification that required NADRA registration as a precondition was likely to exclude the poorest since there were p recisely the population segments where NADRA registration was relatively weak. An alternative measure which was eventually

Economic & Political Weekly

EPW

july 9, 2011 vol xlvi no 28

used was to refer to elected representatives to carry out the initial identification of potential beneficiaries. These beneficiaries were to be identified by parliamentarians in their respective constituencies using the same criteria which could be then verified through NADRA records. The big difference from the earlier proposal was that now a person could be recommended for BISP by a parliamentarian even if he or she did not have NADRA i dentity cards.

Nuanced Design

The shift away from an exclusive focus on NADRA also meant a more nuanced design of the programme which corresponded with policy objectives rather than merely data availability. One problem with the exclusive use of NADRA was the fact that women are disproportionately represented among the non-registered population. The political leadership was eager to promote BISP as a women-focused programme. Now women could be primary beneficiaries even if they did not have NADRA identity cards to begin with. Their registration for BISP, in fact, could coincide with their registration with NADRA. It was reported that there was a big rise in NADRA applications as a result of the BISP, particularly among poor women in relatively underdeveloped rural areas.

The parliamentarian-based targeting allocated equal numbers of application forms for BISP to each Member of the National A ssembly (MNA) and senator. Every parliamentarian (whose total number was 442) was given 8,000 applications and supplied with selection criteria. The parliamentarians were then supposed to identify beneficiaries in their constituencies whose eligibility would then be assessed by NADRA using the prescribed criteria. Although the selection criteria included reference to income – that the family’s monthly income must be less than Rs 6,000 – it was assumed that income could not be measured reliably. The proxy means used as filters were also not easily verifiable since the information supplied by an applicant in the BISP form was only being matched against information supplied by the same person to NADRA.

Parliamentarians also found it difficult to manage the selection process, given the difficulty actually establishing physical contact with 8,000 constituents, and verifying their personal information. In most cases the parliamentarians used their existing party workers to identify beneficiaries. While a statistical audit and impact evaluation of BISP beneficiaries is pending, rapid a ssessments in five villages in Punjab and Sindh suggest that the programme does have a visible presence on the ground (Gazdar and Mallah 2010). This is in contrast with previous cash transfers which, due to their smaller scale, were relatively less well known in poor communities.

BISP was present in four out of five villages where the rapid a ssessment was conducted. In the one village where no BISP beneficiaries were found it was reported that the residents were political factional rivals of the local parliamentarian who had distributed BISP application forms. Neighbouring villages received the benefit, and there was a strong demand from potential women beneficiaries of this omitted village for inclusion. Some of the women had even registered for the NADRA identity card in anticipation of a BISP application. In the other four villages there were beneficiaries ranged from a quarter to a third of all households. These ratios cannot be read as representative, but illustrative of the villages in question. Patterns of inclusion and exclusion showed that while nearly all of the beneficiaries belonged to poorer households within the village, some of the poorest were excluded due to their weak political connections.

The problems with the parliamentarian-based identification of beneficiaries were acknowledged by the government from the very outset. In fact, it was announced that the parliamentarianbased system was only a stop-gap until a more institutionalised arrangement was in place. Besides the somewhat subjective n ature of beneficiary selection – even if it did not entail serious inclusion errors – there was the obvious issue of political bias. Since most parliamentarians have stable support bases in their constituency, it was expected that the beneficiaries will be mostly s elected from among their own supporters. The government rightly claimed that this did not amount to party-political bias at the n ational level – in constituencies where opposition candidates had been elected the bias was in favour of opposition supporters. The systematic exclusion of people on grounds of political affiliation even if this was happening only at the local level, was not a sustainable strategy from a social protection point of view.

The government reached an agreement with the World Bank to change the basis for beneficiary identification from the parliamentarians to a poverty census in which basic information would be collected from all households in order to generate a “poverty score”. The proxy variables and the scorecard designed for this purpose was based on household survey data from 2005-06. In May 2009 a pilot poverty census was initiated in 15 out of the 132 districts of the country. The process of designing the poverty census revealed even more acutely the organisational gaps in social p olicy implementation in Pakistan. The poverty census was subcontracted to three different organisations – the Population C ensus Organisation (PCO), the Rural Support Programme Network (RSPN) and the Pakistan Poverty Alleviation Fund (PPAF) – as it was clear from the outset that no single organisation within or outside government had the requisite capacity. A fourth o rganisation – viz, NADRA – was entrusted with the task of data entry and analysis. The nascent BISP organisation, as yet with a relatively thin presence on the ground, was expected to play a coordination role.

The poverty census was based on a complete enumeration of all households and individuals within the pilot districts. The forms were returned to NADRA which then generated poverty scores for all households. Beneficiaries were to be selected using these scores. The poverty census has been completed in the 15 districts, and its results have started to be analysed. There is a commitment on the part of the BISP that once these data are compiled and analysed, there will be a switchover to new beneficiary lists which will replace the lists generated through the parliamentarians. A national roll-out of the poverty census has been a nnounced. The switchover to new lists will further test the political resolve of the government to the institutionalisation of the BISP.

There are clear signs that the federal government is interested in using the BISP as a platform for the introduction of other dimensions of social protection. The law establishing BISP which was e nacted with cross-party support in parliament sees the program me as an “autonomous social safety net authority to coordinate the design and implementation of targeted programmes for the poor”. The BISP was used as a vehicle for the provision of emergency assistance to internally displaced people in the conflictaffected regions in the north-west of the country. A one-off lump sum transfer – framed as start-up capital for poor families – has also been introduced as part of the BISP. Finally, the government has a nnounced its interest in linking health insurance with the BISP.