Main sources of information for social security statistics are:

• administrative data from social security schemes/ programmes
• sample household surveys

Both can be very useful data sources for social security policy research and yet in both instances it is important to take into account the quality of the data and the extent to which the data adequately addresses the policy questions that are being considered as part of the research exercise.

ILO main experience in social protection statistics and indicators was, and still is, based mainly on administrative data. Administrative data may be collected at the central, regional and local levels, although in the latter case unified data standards are necessary to assure regional data comparability.

 

 

 

Overview of ILO social security databases based on administrative data

Quick links:

• The Social Security Expenditure database
• Microinsurance scheme database
• The social security inquiry

 

 

 

 

Advantages and drawbacks of administrative data

• Advantages of administrative data
  • Unintrusive to collect (though measures need to be taken to ensure that the researcher has the right to use the data)
  • Few additional costs in making the data available as the dataset is already in place (unlike a household survey which is very costly to implement)
  • Is comprehensive in that all recipients are contained within the dataset.
• Drawbacks of administrative data
  • Administrative data usually contain ample information on those groups of the population that are covered by social security, but not on those who are not covered
  • While administrative data can be used to estimate the extent of coverage, they usually do not provide any insights on the causes and effects of non-coverage.
  • Eligible non-recipients are not captured.
  • Need to guard against double-counting if it is not possible to identify individuals using a unique reference number or ID when more than one benefit is involved.

Sample household surveys are designed to understand the situation and behaviour of individuals and households, and are particularly useful for welfare analysis, and for impact analysis of a given social security programme [see micro simulations].

Quick links

• What can survey data be used for?
• Drawbacks of sample household surveys
• Types of surveys

What can survey data be used for? [up]
Household survey data are particularly useful for estimating the following measures that can serve for social security monitoring:

• gathering information on the distribution of specific risks among the population, correlates of specific risks, and participation in social security
• identifying existing gaps in coverage and analyzing factors related to these gaps (e.g. where social security is not reaching specific social and economic groups or regions). This is often best undertaken in conjunction with administrative data.
• calculating aggregates of individual or household income and expenditure that allow for the estimation of insurable earnings, average wages, average expenditure. In some cases, where information on taxes and contributions incurred is available and reliable, the aggregate level of taxes and contributions could be assessed. Listed aggregates are often used as denominators for the calculation of specified performance indicators (e.g. average replacement ratios of benefits in payment, administrative costs in relation to total insurable earnings)
• calculating aggregates of the total employed population, total insured population and total beneficiaries, that are used to estimate scheme indicators (e.g. scheme demographic ratio)
• evaluating the effectiveness of schemes in terms of outcomes achieved (e.g. increasing levels of income in relation to the poverty line, the impact of social protection on education and health outcomes)
• modelling the performance of social security schemes in the long term , particularly with respect to coverage and benefit levels as well as to the impact of external factors (e.g. increasing/decreasing unemployment, increasing consumption and levels of income).

Drawbacks of sample household surveys [up]

• the scope of information on social security varies between different countries. In many cases, even though information on benefits received may be available, information on contributions is not sufficient. Thus it is often not possible to properly adequately evaluate the number of contributors to different schemes.
• another drawback for using survey data to assess social security coverage is also the impossibility of distinguishing between new beneficiaries and the stock of beneficiaries.
• cost issue — household surveys can be very expensive
• response bias as low income people are less likely to participate in surveys
• the difficulty of designing sample frames from which to draw representative samples (particularly in developing countries which may not have electoral registers or geocoded address files etc.)
• respondent fatigue — some deprived areas feel 'over-researched';
• intrusiveness
• difficulties of recall, especially for detailed income/expenditure questions. There is also the risk of under-reporting of social assistance receipt.

Types of surveys include [up]

• Labour Force Surveys
• Household Budget Surveys
• Income and Expenditure Surveys
• Living Standards Measurement Surveys
• Priority Surveys
• Core Welfare Indicators Surveys
• Demographic and Health Surveys
  • Demographic Health Survey (DHS) resources
• Multiple Indicators Cluster Surveys
• Population and Agricultural Censuses