SAE motivation

Following the Colombian National Development Plan for 2018-2022, it was mandatory that the Department of Social Prosperity (DPS) redesigned monetary transfer programs Jóvenes en Acción and Familias en Acción, the most critical conditional monetary transfers in Colombia. According to Law 1948 of 2019, access to these programs was restricted to poverty and extreme poverty populations. Hence, DPS programs must meet the criteria of attention to these populations. Therefore, there was a need to know the distribution of poverty rates at the municipal level, a deeper disaggregation than the one provided by the Large Integrated Household Survey (state level).

In census years, it is possible to estimate multidimensional poverty at the municipality level accurately. However, given the complexity of monetary poverty, it is impossible to include this topic in the census questionnaire.

In addition, the primary socio-economic survey in Colombia (GEIH) only allows disaggregations for some departments and main metropolitan areas of the country. More disaggregated data on poverty and extreme poverty are important for the Department of Prosperidad Social to provide its social protection services efficiently and to monitor its social programmes in the country. More disaggregated poverty mapping also helps raise awareness of the poverty issues in the country.

Input data

For monetary poverty estimates, input data include:

  • Integrated household survey – Gran Encuesta Integrada de Hogares (GEIH). GEIH is the largest household survey in Colombia and is carried out monthly by DANE (National Administrative Department of Statistics). The survey collects data on labour market, monetary poverty, extreme poverty and inequality. The survey covers about 230,000 households and 438 municipalities (out of 1,122 in the country). Data are available for the national level, by urban and rural residence and for 13 metropolitan areas and 24 states (Departments).
  • Population census, carried out by DANE every 10 years, covers the entire population. The latest census was carried out in 2018. Census is the official source for multidimensional poverty. Census does not collect data on monetary poverty.

Building the SAE model

The SAE methodology used a household-level (unit-level) model that took into account the National Population and Housing Census and the GEIH by applying the best empirical predictor under a model with nested errors; as the one proposed by Molina and Rao (2010). This model was used to estimate households' average income in Colombia to estimate indicators of interest such as the incidence of monetary poverty, the incidence of extreme poverty, and some measures of inequality.

Figure: Stepwise approach taken for SAE

Source: Presentation made by Natalia Arteaga Gutiérrez, Departamento de Prosperidad Social de Colombia during the ECLAC-ISWGHS joint webinar on poverty mapping, 1 July 2021

Benchmarking/data validation

Poverty estimates using the small area estimation method were compared with direct estimates produced by DANE, at the national, urban/rural, principal cities and states level. Poverty estimates were also compared with data on roads, energy and electricity coverage and a positive correlation was found. That is, more roads, energy and electricity consumption is associated with areas with lower poverty level.

Coefficient of variation for SAE estimates to be published is set at 30%.

Update of SAE methods

In 2019 DANE updated the value of the poverty lines; however, GEIH did not change anything in the questionnaire or the collection mode. This way, the EBP methodology remains pretty much the same. According to these new definitions, households are now classified as poor or not poor in the prediction stage. i.e., the methodology is the same, but the poverty lines changed compared to those from 2018.

This means that comparability does not hold between 2018 and 2019. However, this is not a specific issue of the SAE methodology but a general issue on poverty comparability in Colombia.

Use of SAE estimates

SAE estimates on monetary poverty are used for more targeted strategies to combat poverty for certain areas such as areas with armed conflict, and the Caribbean and Pacific regions that have been priority areas in the national development plan. The disaggregated poverty data are also used as an input for the definition of territorial coverage of Cash Transfers programs in the country.

Capacity building

Aware of this challenge, the Colombian Department of Social Prosperity and UN-ECLAC established a technical cooperation agreement for capacity-building in Colombia for implementing SAE methodologies to estimate poverty at the disaggregated level. This way, maps of monetary poverty and extreme monetary poverty were created and published for the entire national territory.

UN-ECLAC Division of Statistics gave a set of lectures on fundamentals of household surveys analysis to the DPS team. After that, basic concepts on area-level and unit-level models were also introduced. The regional advisor gave guidelines on the fit of EBP models using the data from GEIH and the Colombian census. ECLAC also provided the DPS team with computing codes based on the statistical computing R software. Both ECLAC and DPS teams meet weekly during 2020 to discuss the processes involved in the EBP methodology. Finally, poverty and extreme poverty maps were created based on ECLAC technical assistance.

The SAE methodology allowed estimating poverty indicators for domains or areas that were not considered within the objectives of the GEIH sampling design. Thus, building information systems at the geographical level with a substantial-quality level for the entire national territory is possible.

Source: Andres Gutierrez Rojas, Regional advisor, UN-ECLAC Statistics Division

  • No labels