• Home
  • Metadata

CIFOR’s Poverty and Environment Network (PEN) global dataset (Version 2.3)


PEN (Center for International Forestry Research [CIFOR])


PEN, 2016, “CIFOR’s Poverty and Environment Network (PEN) global dataset”, doi:10.17528/CIFOR/DATA.00021, Center for International Forestry Research (CIFOR), V2.

Dataset persistent ID



Email Us


The PEN network was launched in September 2004 by the Center for International Forestry Research (CIFOR) with the aim of collecting uniform socio-economic and environmental data at household and village levels in rural areas of developing countries.

The data presented here, produced by CIFOR, were collected by 33 PEN partners (mainly PhD students) and comprise 8,301 households in 334 villages located in 24 countries in Asia, Africa and Latin America.

Three types of quantitative surveys were conducted:
1. Village surveys (V1, V2)
2. Annual household surveys (A1, A2)
3. Quarterly household surveys (Q1, Q2, Q3, Q4)

The village surveys (V1, V2) collected data that were common to all or showed little variation among households. The first village survey, V1, was conducted at the beginning of the fieldwork to obtain background information on the villages. The second survey, V2 was conducted at the end of the fieldwork period to collect information for the 12-month period covered by the surveys.

The household surveys were grouped into two categories: quarterly surveys (Q1, Q2, Q3, Q4) to collect income information, and, household surveys (A1, A2) to collect all other household information.

A critical feature of the PEN research project was the collection of detailed, high-quality data on forest use. This was done through quarterly income household surveys, for two reasons: first, short recall periods increase accuracy and reliability and, second, quarterly data allows us to document seasonal variation in (forest) income and thus, inter alia, help us understand to what extent forests act as seasonal gap fillers.

There are three partners (10101, 10203 and 10301), who, because of various particular circumstances, only conducted three of the four income surveys. In addition, 598 of the households missed out on one of the quarterly surveys, due to, for example, temporal absence or sickness, or insecurity in the area. These are still included in the database, while households missing more than one quarter were excluded.

Two other household surveys were conducted. The first annual household survey (A1) collected basic household information (demographics, assets, forest-related information) and was done at the beginning of the survey period while the second (A2) collected information for the 12-month period covered by the surveys (e.g. on risk management) and was done at the end of the survey period. Note, however, that we did not collect any systematic data on the time allocation of households: while highly relevant for many analyses, we believed that it would be too time-consuming a component to add to our standard survey questions.

The project is further described and discussed in two edited volumes by Angelsen et al. (2011) (describes in particular the methods used) and Wunder et al. (2014) (which includes six articles based on the PEN project).

Back to top

Sign up to our monthly newsletter

Connect with us