Quantifying invariant features of within-group inequality in consumption across groups

TL;DR

This paper analyzes consumption expenditure data across countries and years, revealing invariant distributional features that are largely lognormal with a power law tail, and introduces a stochastic model to explain these patterns.

Contribution

It identifies invariant features of consumption distributions across diverse groups and proposes a stochastic multiplicative model to explain the observed invariance.

Findings

Consumption distributions are mostly lognormal with a power law tail.

Distributions coincide under normalization by mean and log scaling.

Core distributional features are invariant across social and economic groups.

Abstract

We study unit-level expenditure on consumption across multiple countries and multiple years, in order to extract invariant features of consumption distribution. We show that the bulk of it is lognormally distributed, followed by a power law tail at the limit. The distributions coincide with each other under normalization by mean expenditure and log scaling even though the data is sampled across multiple dimension including, e.g., time, social structure and locations. This phenomenon indicates that the dispersions in consumption expenditure across various social and economic groups are significantly similar subject to suitable scaling and normalization. Further, the results provide a measurement of the core distributional features. Other descriptive factors including those of sociological, demographic and political nature, add further layers of variation on the this core distribution. We present a stochastic multiplicative model to quantitatively characterize the invariance and the distributional features.