Skip to content
Back to search results

Spatial Gini Concentration Index

Introduced by
Corrado Gini (coefficient); locational adaptation in regional science / economic geography
Year
1991
Type
Descriptive index of how unevenly an activity is distributed across space
Family
Regional concentration indices
Font size

What does it do?

The spatial (or locational) Gini concentration index adapts the classic Gini coefficient to geography, summarizing in a single number between zero and one how unevenly an activity — an industry, a population group, a resource — is distributed across spatial units relative to a benchmark such as total population or land area.

It is the workhorse measure for quantifying geographic concentration and agglomeration in economic geography.

The intuition behind the method

Imagine ranking regions from least to most intensely endowed with an activity and drawing the cumulative share of the activity against the cumulative share of the benchmark — a Lorenz curve for space. If the activity is spread exactly in proportion to the benchmark, the curve is the 45-degree line and the Gini is zero. The more the activity piles up in a few places, the more the curve bows away from the diagonal and the closer the Gini climbs toward one. The index is just twice the area between the curve and the diagonal, capturing 'how lumpy is this activity across the map' in one figure.

History

Corrado Gini introduced his coefficient of inequality in 1912 for income distributions, and demographers and geographers soon recognized that the same Lorenz-curve machinery could measure how unevenly any quantity is spread across categories. Otis Dudley and Beverly Duncan's 1955 methodological analysis of segregation indexes placed the Gini and dissimilarity measures on a rigorous footing for spatial and group distributions, and the index became a standard descriptor of geographic concentration. It returned to prominence in economic geography with Paul Krugman's 1991 work on agglomeration and increasing returns, where locational Gini coefficients are used to document the spatial concentration of industries.

When to use it

Use the spatial Gini concentration index when you need a single, comparable summary of how concentrated or dispersed an activity is across a set of regions, and you have the activity and a benchmark measured for each unit. It is ideal for ranking industries by agglomeration, tracking whether an activity is concentrating or dispersing over time, and comparing concentration across countries or sectors. It is descriptive and aspatial in its classic form — it ignores whether the high-share units are adjacent — so when the geographic arrangement (clustering versus scatter) matters, pair it with a spatial-autocorrelation statistic, and when a formal significance test of concentration is needed, use the Ellison–Glaeser index or a permutation test.

Limitations and considerations

  • Aspatial: the classic locational Gini does not use adjacency, so a checkerboard and a single cluster with the same shares score identically.
  • Sensitive to the spatial unit definition and the modifiable areal unit problem — coarser zones lower measured concentration.
  • Does not, on its own, provide a statistical test of whether observed concentration exceeds chance.
  • Like all single-index summaries it can mask very different distributional shapes that yield the same value.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Reduces a whole distribution of regional shares to one interpretable number on a fixed zero-to-one scale.
  • Benchmark-relative, so it isolates concentration beyond what the population or area distribution would produce.
  • Comparable across activities, regions, and time, making it ideal for ranking and trend analysis.
  • Backed by the well-understood Lorenz-curve apparatus, with clear graphical interpretation.

Applications

  • Measuring and ranking the geographic concentration of industries to study agglomeration.
  • Quantifying residential segregation and the uneven spatial distribution of population subgroups.
  • Tracking whether economic activity is concentrating or dispersing across regions over time.
  • Assessing the spatial inequality of access to resources, services, or amenities.
  • Comparing concentration of biodiversity, pollution, or land use across administrative units.

How it works

How it works

1

For each spatial unit u, compute its share of the activity, x_u = a_u divided by the total activity, and its share of the benchmark, y_u = b_u divided by the total benchmark (population, employment, or area). These paired shares describe how the activity is allocated across units relative to the reference distribution.

Formula
x_u = \frac{a_u}{\sum_v a_v},\qquad y_u = \frac{b_u}{\sum_v b_v}
2

Sort the units by the ratio x_u / y_u — equivalently their location quotient — from lowest to highest, then form the running cumulative sums X_k of activity shares and Y_k of benchmark shares. Plotting X against Y traces the spatial Lorenz curve.

Formula
\text{order units by } x_u/y_u;\quad X_k = \sum_{u\le k} x_u,\; Y_k = \sum_{u\le k} y_u
3

The Gini coefficient is twice the area between the Lorenz curve and the line of perfect proportionality; the trapezoidal formula shown computes it directly from the ordered cumulative shares. Values run from zero, when every unit holds the activity in exact proportion to the benchmark, up toward one as the activity becomes confined to fewer and fewer units.

Formula
G = 1 - \sum_{k} (Y_k - Y_{k-1})\,(X_k + X_{k-1})
4

A Gini near zero means the activity is geographically even relative to the benchmark; a Gini near one means it is highly concentrated in a small number of places. Because it compares against a benchmark rather than against equal counts per unit, the index measures concentration over and above what the underlying distribution of population or area would imply.

Formula
G \in [0,1]:\; 0 \Rightarrow \text{even (proportional) distribution},\; 1 \Rightarrow \text{maximal concentration}

Common pitfalls

  • Interpreting a high Gini as spatial clustering: it measures unevenness across units, not geographic contiguity of the concentration.
  • Comparing Ginis computed at different spatial resolutions, which conflates real change with the modifiable areal unit problem.
  • Using equal weights (counts per region) instead of a benchmark share, which confounds concentration with the underlying size distribution.
  • Ignoring that the index is a point estimate with sampling variability when counts per unit are small.

Frequently asked questions

How does the spatial Gini differ from the ordinary income Gini?

The mechanics are identical — both are twice the area between a Lorenz curve and the diagonal — but the units and the benchmark differ. The income Gini ranks individuals by income against an equal-population diagonal; the spatial Gini ranks regions by their activity-to-benchmark ratio against a proportional-distribution diagonal, so a Gini of zero means the activity is spread in proportion to population or area rather than equally per region.

Why is the locational Gini called 'aspatial'?

Because it depends only on the set of regional shares, not on where those regions sit relative to one another. Re-shuffling the high-share regions to opposite ends of the map leaves the Gini unchanged. To capture whether high values are geographically clustered you need a spatial-autocorrelation statistic such as Moran's I or Getis-Ord, which the Gini does not provide.

How is it related to the location quotient?

The location quotient is the per-region ratio of local activity share to benchmark share — exactly the quantity by which units are ordered to build the spatial Lorenz curve. The spatial Gini then aggregates the whole set of location quotients into one concentration number, so LQs are the regional detail and the Gini is the system-wide summary.

Related methods

Sources

  1. Duncan, O. D., & Duncan, B. (1955). A methodological analysis of segregation indexes. American Sociological Review, 20(2), 210–217.

    Open DOI reference
    Copied.
  2. Krugman, P. (1991). Increasing returns and economic geography. Journal of Political Economy, 99(3), 483–499.

    Open DOI reference
    Copied.