Hunger Safety Net Programme Survey 2016, Phase 2

Kenya, 2016

Get Microdata

Reference ID

KEN_2016_HSN-P2_v01_M

DOI

https://doi.org/10.48529/xmcg-sm67

Producer(s)

Oxford Policy Management Limited

Collection(s)

Impact Evaluation Surveys

Metadata

Documentation in PDF DDI/XML JSON

Created on

Apr 30, 2020

Last modified

Apr 30, 2020

Page views

28360

Downloads

1475

Data file: v2_1_HQ_main

This file contains data at the household level and corresponds to the HSNP2 2016 Household Survey Questionnaire.

Variables C_7 to C_19 from the questionnaire have been excluded from this dataset for confidentiality. Interview results and details variable are also excluded. The file contains completed interviews only.

This dataset includes variables for section 5 - Assets; 6 - Land and crops; 9 - Household transfers; 10 - Food security; 11 - Subjective poverty; and 12 - Savings and borrowing.

This dataset also contains constructed indicators for monthly food and non-food consumption expenditure, both in nominal terms (n_ variables) and adjusted for differences in prices across sub-locations (r_ variables). Values are expressed in Kenyan Shillings (KES). Aggregated food consumption and non-food consumption are constructed at household level, as well as per capita and per adult equivalent. The Anzagi-Bernard scale is used, similarly to what has been used in the Kenya Integrated Household Budget Survey 2005.

In addition to food and non-food consumption aggregates, this dataset contains monthly household health expenditure, education expenditure, and rent expenditure. Variable 'nfoodexp_imputed' is a dummy equals 1 if non-food expenditure has been imputed. Variable 'normpind' is the normalised Paasche index constructed to adjust nominal value of consumption for differences in prices across sub-locations. The Paasche index can be understood as an 'intra-survey' inflation rate. This is the 'price' in a sublocation divided by the average price in our survey.

Based on monthly consumption per adult equivalent adjusted for differences in prices, The Foster-Greer-Thorbecke (FGT) indices have been constructed:
· int_poor. This indicator measures FGT0, the proportion of households whose monthly consumption expenditure per adult equivalent, adjusted for regional price differences, is lower than a poverty line set at KES 2,317.6
· int_hh_povgap. This indicator measures FGT1, the average distance between households monthly consumption expenditure, adjusted for regional price differences, and poverty line set at KES 2,317.6.
· int_hh_sevpov. This indicator measures FGT2, the average squared distance between households monthly consumption expenditure, adjusted for regional price differences, and poverty line set at KES 2,317.6.

In addition to the FGT indices, int_foodpoor measures the proportion of households whose monthly food expenditure per adult equivalent, adjusted for regional price differences, is lower than a poverty line set at KES 1,779.3

Variable normreg in the household Proxy Means Test (PMT) Score. The score has been normalised to take into account county PMT cut-offs to be eligible for HSNP2 cash transfers. Cut offs are as follows: Turkana KES 320.39; Mandera KES 600.17; Wajir KES 708.15; Marsabit KES 576.28.

Variable sample_cut_off categorises households based on their normalised PMT scores, for the purpose of the household impact evaluation and the Local Economy-Wide Impact Evaluation (LEWIE). The bandwidth chosen for the household impact evaluation relying on the Regression discontinuity approach was a distance of 400 above and below the normalised eligibility cut-off. The treatment group are those households with PMT scores just below the eligibility threshold and the control group are those households with PMT scores just above the eligibility threshold. The sample of households for the LEWIE analysis includes households outside the bandwidth.

Lastly, this dataset include sampling weights.For analysis at the household level, the household weights included in this dataset should be used, (weight weight2 weight3), while for analysis at population level, the population weights included in this dataset should be used (popweight_multiple, popweight). See weighting sub-section in the Sampling section for a description of the weighting procedure and the survey settings that should be used.

This dataset can be linked to all other Household survey datasets using the unique household identifier (H_ID).