Glossary

Clustering

A method of sample design wherein the population is first divided into units such as villages or street blocks. Households to be surveyed are then selected from within a cluster. Clustering reduces survey costs, but it leads to higher variances.

Consumer price index

The weighted averages of individual commodity prices regularly observed in the market. The weights come from the budget shares in the reference period, typically as reported in a household expenditure survey.

Engel’s Law ¹⁷

The share of budget devoted to food tends to decrease as total real consumption expenditure increases. Problems with this assumption:

• The relationship between the food budget share and consumption will generally differ across households due to differences in relative prices, demographic differences, or differences in preferences.
• The income elasticity of demand for food can be very close to unity for poor households.

Headcount

The proportion of people deemed to be poor. ¹⁸

Panel datasets

Datasets compiled from a number of observations over time on a number of cross-sectional units like individuals, households, firms, or governments. Each “panel” consists of cross-sectional data from a specific observation period.

Poverty

A level of material well-being below that deemed to constitute a reasonable minimum by the standards of a given society.¹⁹ Due to the subjective nature of this concept, analysts will often construct an artificial level of material well-being called the consumption aggregate.

Recall period

The period of time to be taken into account. For example, in the following question, the previous thirty days is the recall period. “On average, how often have you eaten lentils in the past thirty days? One time or less per week, two to four times per week,….”

Stochastic Dominance

The random variable X first-order stochastically dominates the random variable Y if Pr[X > a] ≥ Pr[Y > a] for all a. If the distribution of X is F and the distribution of Y is G, then X first-order stochastically dominates Y if F(a) ≥ G(a) for all a. ²⁰

Stratification

A method of sample design wherein the population is first divided into different groups (such as urban vs. rural or rich vs. poor) and households to be surveyed are selected from within each strata. This method can provide a more representative sample, and so reduces variance. Stratification may also reduce costs by allowing lower sampling rates for more homogenous sub-groups or more populous areas.

¹⁷  Ravallion, Martin. 1992. “Poverty Comparisons: A Guide to Concepts and Methods.” Living Standards Measurement Study Working Paper No. 88. Pgs 21-22.

¹⁸  Ravallion, Martin. 1992. “Poverty Comparisons: A Guide to Concepts and Methods.” Living Standards Measurement Study Working Paper No. 88. Pgs 31.

¹⁹  Ravallion, Martin. 1992. “Poverty Comparisons: A Guide to Concepts and Methods.” Living Standards Measurement Study Working Paper No. 88. Pg 4.

²⁰  http://www.econ.hku.hk/~hrneswc/uncertainty/fsd.pdf

Back to Comparing Poverty Over Time