Foster–Greer–Thorbecke Poverty Measures¶
A parameterized family of poverty measures averaging powered, normalized welfare shortfalls below a line, with additive subgroup decomposition.
Core Idea¶
The Foster–Greer–Thorbecke (FGT) family measures poverty by averaging normalized shortfalls below a positive line \(z\) across a whole population of \(N\) comparable welfare observations. For \(\alpha\ge0\), \(P_\alpha=N^{-1}\sum_{y_i<z}((z-y_i)/z)^\alpha\). \(P_0\) is the headcount share; \(P_1\) is the population-average poverty gap; \(P_2\) averages squared gaps and gives more weight to deeper shortfalls. For disjoint exhaustive subgroups using compatible conventions, the total equals the sum of subgroup indices weighted by their population shares.[ref-612a561aef11][ref-27c780a268ca]
This draft counts observations strictly below \(z\); the original paper permits equality, which can change \(P_0\) when an observation falls exactly on the line.[^ref-612a561aef11]
Scope of Application¶
The original 1984 paper applied \(P_2\) to 1970 Nairobi Household Survey subgroups to illustrate decomposition. A later World Bank Georgia analysis reports urban, rural and total \(P_0/P_1/P_2\) in 2003 and 2006 under two different monthly poverty lines. The formula transfers, but its welfare measure, poverty line, price basis and \(\alpha\) must be stated for each comparison. It does not automatically become the Alkire–Foster multidimensional measure, which uses a different deprivation-identification architecture.[ref-612a561aef11][ref-1c9d0e4c0c5d]
Clarity¶
The family separates incidence, average shortfall and squared severity. \(P_0\) cannot change when people already below the line become poorer without crossing the line; \(P_1\) records their normalized gaps, while \(P_2\) weights deeper gaps more. The denominator is \(N\), not the number poor. The live Poverty Gap Index corresponds to \(\alpha=1\) and is not a synonym for the whole parameterized family.[ref-612a561aef11][ref-27c780a268ca]
Manages Complexity¶
One formula packages several poverty summaries while leaving their sensitivity choice visible. Additive decomposition makes a total index equal to \(\sum_g(N_g/N)P_{\alpha,g}\) for a coherent partition, so analysts can attribute measured poverty to groups without confusing an unweighted average of group rates with the population aggregate. This is an accounting relation, not a causal explanation or a replacement for the underlying welfare distribution.[^ref-612a561aef11]
Abstract Reasoning¶
Choose comparable welfare observations, a positive line and \(\alpha\). Assign zero to observations at or above the line, calculate normalized gaps for the rest, power them, and average over all \(N\) units. To compare places or dates, first harmonize welfare and line conventions. To decompose, use disjoint groups and their population shares. If depth-sensitive changes matter, compare \(P_0\), \(P_1\) and \(P_2\) rather than reporting headcount alone.[ref-612a561aef11][ref-1c9d0e4c0c5d]
Knowledge Transfer¶
Nairobi household-income subgroups and Georgia household-expenditure urban/rural comparisons instantiate the same line–gap–exponent–population-mean–share-weighting relation, despite different surveys and purposes. What transfers is the method, not a historical threshold or an empirical conclusion. Live prime Aggregation is the proposed broad parent; Measurement is related but not established as a strict parent. A more general threshold-shortfall aggregator would be a separate future-prime question. The family remains a domain-specific poverty-measure identity.[ref-612a561aef11][ref-1c9d0e4c0c5d]
[^ref-612a561aef11]: James Foster, Joel Greer and Erik Thorbecke, “A Class of Decomposable Poverty Measures”, Econometrica 52(3) (1984), pp. 761–766; author-uploaded accessible text. [^ref-27c780a268ca]: World Bank, Handbook on Poverty and Inequality, Chapter 4, printed pp. 67–72. [^ref-1c9d0e4c0c5d]: World Bank, A Unified Approach to Measuring Poverty and Inequality, Chapter 3, Table 3.4, printed pp. 164–165.
Relationships to Other Abstractions¶
Current abstraction Foster–Greer–Thorbecke Poverty Measures Domain-specific
Parents (1) — more general patterns this builds on
-
Foster–Greer–Thorbecke Poverty Measures is a kind of Aggregation Prime
Each FGT member deliberately collapses many individual welfare observations into one poverty summary.
Hierarchy path (1) — routes to 1 parentless root
- Foster–Greer–Thorbecke Poverty Measures → Aggregation → Micro Macro Linkage
Neighborhood in Abstraction Space¶
Foster–Greer–Thorbecke Poverty Measures sits in a sparse region of the domain-specific corpus (80th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Financial & Economic Ratios (22 abstractions)
Nearest neighbors
- Theil Index — 0.85
- Poverty gap index — 0.84
- Gini Coefficient — 0.82
- Lorenz Curve — 0.82
- Identifiable Victim Effect — 0.81
Computed from structural-signature embeddings · 2026-10-08