90. Analysis of social protection expenditure is hindered by serious data and definitional problems. Four factors make analysis of spending on social protection particularly difficult. First, there is no functional classification of expenditure in Madagascar, so the government itself does not produce data on social protection as a function. This makes it necessary to make estimates based on the programme or administrative classifications of expenditure, which is complicated by lack of clarity about the nature or purpose of much of the expenditure recorded in these classification systems and the lack of consistency from year to year in the definition of programme budgets (see Ralaivelo, 2011, regarding the interpretation of programme budgets). Second, the programme budgets exclude general administration costs, including personnel, which cannot easily be attributed to specific programmes, including those with a social protection character. Third, some expenditure financed by donors is not included in the budget or the government accounts. Fortunately, there is a comprehensive data-base on aid flows managed by the Aid Coordination Unit in the Prime Minister’s Office (see Primature, 2010, and World Bank, 2011), which has survived the political crisis and the related withdrawal of donors from the use of government systems. It is in theory possible to add expenditure from this source to data on government financed expenditure, but in practice there are risks of double-counting. Fourth and perhaps most challenging of all, there are ambiguities about the definition of social protection and thus deciding what expenditure should or should not be included. One option would be to apply the definition of the social protection function in the internationally adopted ‘classification of the functions of government’ (COFOG) used for the purposes of national accounts and government financial statistics by the UN, the IMF and the World Bank. This notably includes all humanitarian action as part of social protection, but excludes expenditure classified in other standard functions such as education, health or agriculture. Other approaches exclude pure humanitarian relief and/or include programmes with social protection characteristics in a wide range of sectors.
91. Estimates of government expenditure on social protection vary widely. Ralaivelo (2011a) has made estimates of government expenditure on social protection based on the programme classification, by interpreting the declared objectives of programmes, and using a broad definition of social protection. The latter included all programmes for the creation of temporary employment (through HIMO programmes), environmental protection and humanitarian relief, social assistance to specific vulnerable groups, nutrition and reduction of health risks, access to water and sanitation, improvement of access by vulnerable groups to education, and social security for state employees. On this broad basis, Ralaivelo found that ‘social protection’ expenditure, on a commitment basis, declined from A303 bn in 2008 to A246 bn in 2009 and A150 bn in 2010. As a share of total expenditure, there was a decline from 16.4% in 2008 to 12.1% in 2009. However, Ralaivelo’s approach includes many categories of expenditure that would not normally be considered ‘social protection’, including for example all expenditure on water and sanitation, vaccination services and other supply-side health services and part of the supply of education services. With this approach social protection becomes almost synonymous with the social sectors generally. Furthermore, a large part of this expenditure, equivalent to roughly half of total ‘social protection’ expenditure in 2009, consists of social security payments by the State for its employees. This is really part of their remuneration package and should be spread across all functions of government, not classified specifically as ‘social protection’. A recalculation of Ralaivelo’s data on a more restrictive basis, excluding supply-side provision of health, education, water and sanitation services as well as social security payments for government employees, found that social protection expenditure declined from
57 A138.6 billion in 2008 to A62.3 billion in 2009 and only A15.3 billion in 2010 (see table 9). The share in total government spending declined from 7.5% in 2008 to 3.1% in 2009. On the other hand, both sets of figures are also incomplete because they miss three crucial categories of expenditure: (1) expenditure on salaries, which are not part of the programme budgets from which Ralaivelo’s data are drawn; (2) some off-budget donor funded expenditure; and (3) some government social protection expenditure that is not readily ‘visible’ in the government accounts, notably the Tsena Mora programme in the Presidency.
Table 9: Government expenditure on social protection, commitment basis, 2008-10 (mn Ariary)
2008 2009 2010 % change
2008-2010 In million Ariary
Social protection (broad definition, current prices) 303,027.4 246,243.8 149,690.2 -50.6
of which, social security for government employees 109,654.7 124,031.6 100,284.5 -8.5 Social protection (narrow definition, current prices) 138,577.6 62,310.6 15,267.1 -89.0
Social protection (narrow definition, constant 2008 prices) 138,577.6 57,165.7 12,850.0 -90.7 As % of total government expenditure
Social protection (broad definition) 16.4 12.1 ...
of which, social security for government employees 5.9 6.1 ...
Social protection (narrow definition) 7.5 3.1 ...
Source: Ralaivelo, 2011a; recalculation of data by Ralaivelo on more restrictive definition of social protection to exclude supply-side service provision in education, health, water and sanitation as well as payment of social security contributions for government employees.
92. The bottom line, however, is that government social protection expenditure has fallen dramatically since 2008. At current prices, social protection expenditure on the broad definition used by Ralaivelo declined by 51% between 2008 and 2010. On the narrow definition, the decline was steeper:
89% at current prices and 91% at constant (2008) prices. In part, this has simply reflected the general decline in public expenditure by a government that has been intent on preventing major budget disequilibria arising from the sharp decline in revenue and aid since the crisis in early 2009. The decline in the relative share of social protection in total government expenditure suggests also that social protection has had less priority than other spending areas as budgets are cut, although this finding may be distorted by the incompleteness of the data, notably with respect to the financing of the Tsena Mora programme.
i. Government spending on ‘population and social affairs’ has declined sharply.
Focusing only on the ‘core’ government agency responsible for social protection, known since 2009 as the Ministry of Population and Social Affairs, budget execution data show that expenditure commitments (engagements) fell by 66% from 2009 to 2010 (see table 10).
Expenditure on salaries increased, by 21%, but investment fell by 88% and non-salary recurrent expenditure by 29%. As a proportion of total government expenditure, on a commitment basis, population and social affairs declined from 1.34% in 2009 to only 0.41% in 2010.
ii. Spending on education social protection measures has been slashed. The September 2010 budget revision law brought large cuts in spending on social protection measures by the Ministry of National Education. The ministry’s overall budget was slashed by 56% from the amount allocated in the original budget for 2010. Among the consequences were the cancellation of the distribution of free school kits at the start of the 2010/11 school-year and a large reduction
58 in the transfer of funds to schools (caisses écoles), which are intended to reduce the burden of school levies and charges on parents.
iii. Government HIMO spending has almost ended. Even before the 2009 political crisis, government expenditure on HIMO type programmes, mainly through ONN, had virtually halted due to the closure of the main funding source – the HIPC debt relief fund. ONN expenditure on its social protection programme fell from A3.2 billion in 2008 to only A25 million in 2009 and A10 million in 2010, according to data supplied by ONN.
iv. Tsena Mora is now the main government social protection programme in terms of funding. Although no data on actual expenditure by Tsena Mora are available, this programme has reportedly received an allocation of A25 billion (Ralaivelo, 2011b), i.e. more than the total for government social protection expenditure in 2010 calculated above. If correct, this would imply a recovery of overall government social protection expenditure since the programme was launched in October 2010, but with expenditure heavily concentrated on this one programme, which is exclusively focused on the large cities.
Table 10 : Expenditure on population and social affairs, commitment basis, 2009-10 (mn Ariary)
2009 2010 % change
Salaries 1,989.70 2,403.30 20.79
Non-salary recurrent expenditure 4,372.00 3,119.20 -28.66
Investment 21,004.50 2,594.10 -87.65
Total 29,375.20 10,126.60 -65.53
As % of total budget expenditure commitments 1.34 0.41
Source: Suivi annuel de l’exécution budgétaire, Ministère des Finances.
93. Health service users fund most social protection in the health sector. In the specific case of the health sector, it is noteworthy that the government finances very little of the limited social protection measures to facilitate access to health services. Under the procedures for FANOME, the equity funds in the CSB are financed entirely from a fixed 3% deduction from the 35% mark-up on the sale of drugs. As Poncin and Le Mentec (2009) have noted, this means that ‘while the [equity fund] financing mechanism introduces the principle of solidarity between the non-poor who finance the system and the identified poor beneficiaries, it does not introduce the principle of solidarity between the sick and the non-sick’. In effect, neither the non-sick (strictly speaking those who do not use the CSB) nor the government fund this mechanism, which helps to explain why its coverage is so limited. The equity funds in the hospitals are different in so far as the government contributes partly to their financing (44% and 51% in the case of the Mahajanga and Fianarantsoa CHUs in 2008, for example), but the balance is provided by fixed deductions on the mark-up of drugs sales and revenue from consultations, again falling on the sick using government hospitals (Poncin and Le Mentec, 2009). In the case of the regionally limited hospital funds for free access to emergency obstetrical and paediatric care, these are funded entirely by aid (the World Bank and UNFPA). MHOs, where they exist, are funded entirely on a contributory basis by their members, although donors assist administrative and start-up costs, as in the case of AFAFI.
94. The decline in government spending has been offset partially by increased donor aid.
Using the data-base of the Aid Coordination Unit in the Prime Minister’s Office, the World Bank has estimated that aid disbursements for social protection rose from USD 26.0 million in 2008 to USD 40.7 million in 2009, before declining slightly to USD 36.8 million in 2010 (World Bank, 2011). Social protection measures in the education sector are now largely being financed by donors, notably through the Catalytic Fund of the Education for All Fast Track Initiative, managed by UNICEF, and through WFP’s
59 support for school feeding. Increased aid for HIMO type projects, notably from the World Bank through the FID, has more than offset the sharp fall in government financing (through ONN), making it possible to expand HIMO coverage. However, a large part of the aid included here is for ‘pure’ humanitarian assistance, rather than social protection in a more long-term developmental sense. Moreover, some of this aid is in the government accounts and so is already taken into account in the figures above on the decline in government social protection expenditure. Some additional resources are provided by NGOs: A10.8 billion in 2010, according to a study by Kone (2011), but it is unclear whether or to what extent these resources are really additional to the flows recorded in the donor aid data-base, as most NGOs act primarily as contractors for official aid agencies.
60 CHAPTER6:STRATEGIC CHALLENGES FOR SOCIAL PROTECTION
95. This chapter highlights the main challenges that should be addressed in a future national social protection strategy. As the Introduction explained, the purpose of this report is not to produce a national strategy for social protection, which is premature in the existing political context, but rather to contribute to the future development of such a strategy by assembling and analysing the available data and thereby highlighting some of the critical issues that a future strategy will need to address. The aim of this chapter is therefore to build on the analysis in the preceding chapters by highlighting a number of strategic challenges for the improvement and expansion of social protection provision in Madagascar.
This takes into account both the analysis of the existing situation in Madagascar (the risk/vulnerability profile, the strengths and weaknesses of existing social protection programmes, and the policy, institutional and financial framework) and experience in other low-income countries, especially in Africa, which can stimulate reflection on potentially appropriate social protection policy options in Madagascar.
The analysis below focuses in particular on the priorities for expansion of social protection, the challenges of improving the efficiency of social protection programmes, the question of political ‘ownership’ or leadership, the need for harmonization and improved coordination among the multiple actors engaged in social protection initiatives, the prospects for increasing ‘fiscal space’ for social protection financing, and the requirements for institutional development and capacity building.