Frequently asked questions

Why a social assistance dataset?

Social assistance consists of budget-financed and rules-based programmes supporting individuals and households vulnerable to poverty with the aim of facilitating their exit from poverty.

Social assistance provision has expanded in low and middle income countries in the 21st century.

The dataset collects information on the design, reach, implementation, institutions and budgets of social assistance programmes in all low and middle income countries, annually between 2000 and 2015.

In practice, the data collection focused on the transfer components of social assistance. Social assistance programmes include service provision components but information on services is scarce and hard to measure consistently across countries.

Is social assistance different from safety nets?

Bretton Woods institutions focus on safety nets. In recent publications – e.g. the World Bank’s State of Safety Nets annual publications – the terms safety nets and social assistance are used interchangeably. But they are distinct. Safety nets combine social assistance and emergency or humanitarian assistance. Emergency or humanitarian assistance provides support to groups affected by hazards.

In low and lower middle income countries, emergency and humanitarian assistance is an important component of development assistance and the work of international organisations.

From an analytical perspective, social assistance is distinct from emergency or humanitarian assistance. Emergency or humanitarian assistance figures in all forms of social organisation as a response to misfortune. Social assistance, on the other hand addresses poverty, vulnerability, and disadvantage as outcomes of economic structures.

The following table summarises institutional features of social assistance vis-a-vis humanitarian and emergency assistance.

Emergency assistance Social assistance
addresses misfortune addresses poverty and vulnerability (generated by economic system)
short term regular and reliable transfers
targets people affected by hazards, it does not select by socio-economic status targets citizens with incomes below a minimum income threshold
financed by budget reserves or international assistance budget/tax financed
Samaritan principle – help those affected by disasters/hazards Citizenship principle – commitment to ensuring minimum living standards
no link to domestic politics requires domestic political support
transfers are discretionary transfers are entitlements; they are rules-based
Provided by NGOs; charities; UN agencies Ministries of Social Development
limited legal framework except human rights if applicable Legislation/regulations
dominant instrument: transfers in-kind although increasingly in cash dominant instrument: old age transfers(budget)/ conditional income transfers (participants)
consumption and recovery consumption and social investment

Applying this distinction can be hard to do in some contexts. In sub-Saharan Africa for example, it implied excluding public works schemes and school feeding programmes. Public works schemes are short-term and do not generate entitlements as employment guaranteed schemes do. School feeding programmes are education interventions improving children’s capacity to learn. They are very useful but they are not social assistance.

How programmes were selected for inclusion?

The data collection included all countries defined as low and middle income in the 2016 version of the World Bank Country Classification.

An inventory of potential social assistance programmes was developed for each country. The definition described above was then applied to identify social assistance programmes. For some countries with a large number of small or localised programmes, the data collection focused on nationwide, large-scale, and/or leading programmes. For example, some states in India have localised programmes. These were excluded from the data collection. In sub-Saharan Africa some programmes are very small in scale but they are significant in leading the expansion of social assistance. They were included.

Where programmes consolidate pre-existing programmes, for example Brazil’s Bolsa Família, the dataset includes Bolsa Família as well as its component programmes.

How was the programme data collected?

Data were collected from a variety of sources: global and regional datasets (ASPIRE, ODI, CEPAL, ADB’s SPI, IPC-PG); national government websites; programme agency reports; research papers; evaluation reports; policy documents; IFIs project documentation and reports; personal communication with programme agencies.

The collection of the data was organised around a codebook, describing each of the variables and the specific coding of the information. The codebook was constructed after extensive consultation with specialist researchers. The codebook is available from the data webpage in the website.

Specialist consultants supported data collection in had-to-reach areas. The data collected were checked against alternative sources of information where available.

Programme Types

The policy literature often classifies social assistance programmes on the basis of their function: conditional transfer, unconditional transfer, old age pension, disability pension, etc. The variable profunc classifies programmes in line with a functional classification widely employed by international organisations.

From an analytical perspective the functional classification might not be helpful, especially as the categories overlap. The variable protype classifies social assistance programmes on the basis of their underlying conceptualisation of poverty. Programmes are classified into four ‘ideal types’ described in the table below.

 Understanding of poverty  ‘Ideal type’ of social assistance programme
Poverty as consumption deficit Pure income transfers
Poverty as consumption and productivity deficit Income transfers combined with asset accumulation

Employment guarantees (community assets)

Conditional income transfers (human capital)

Poverty as consumption, productivity and inclusion deficit Integrated antipoverty transfer programmes
Source: Barrientos [2013] Social assistance in developing countries, Cambridge: Cambridge University Pess.

Reach data 

Data on programme participants released by the respective agencies report the numbers of direct individual recipient or the numbers of participant households. For example, social pension programmes report the number of pensioners while conditional income transfers report the number of participant households. The dataset reports on the reach of programmes in terms of individuals and households. Where the raw data reported the number of individual recipients, it was multiplied by the average household size for the specific country to calculate the full reach of the programme.  Where the raw data reported the number of households, it was multiplied by the average household size for the specific country to calculate the full reach of the programme.

Information on average household size is from: UNDESA. (2017). Household Size and composition around the world 2017 – Data Booklet (Data Booklet No. ST/ESA/SER.A/405). New York: United National Department of Economic and Social Affairs, Population Division.

Financial data

All financial data – transfers values and programme budgets for example – are reported in domestic currency and in purchasing power parity. Purchasing poverty parity exchange rates enable consistent comparison across countries.

PPP exchange rates were taken from World Development Indicators (; accessed 21/11/2017). Purchasing power parity conversion factor is the number of units of a country’s currency required to buy the same amounts of goods and services in the domestic market as U.S. dollar would buy in the United States. For most economies PPP figures are extrapolated from the 2011 International Comparison Program (ICP) benchmark estimates or imputed using a statistical model based on the 2011 ICP.

Reporting on the financial resources associated with social assistance programmes shows significant variation across countries. The codebook included three data formats: (i) budgeted resources (bugt); (ii) programme expenditure (cost); and (iii) donor/government financing expenditure (dfinex/govfinex). One or more of these was reported for the majority of programmes. Where the amount reported referred to more than one year, it was allocated proportionally to each year.

A derived variable finres consolidates the information for each country/year by reporting only one data format where more than one was available. This was done by ordering the data (i), (ii), (iii) and selecting the first one available.  The variable finresour indicates the data format selected.