The Distribution of Strike Size:Empirical Evidence from Europe and North America in the 19th and 20th Centuries

TL;DR

This paper analyzes the distribution of strike sizes in Europe and North America over two centuries, finding that mixtures of lognormals and Pareto laws effectively model the data, especially in the upper tails.

Contribution

It provides empirical evidence on strike size distributions using long-term data, highlighting the applicability of mixture models and Pareto laws in this context.

Findings

Mixtures of two or three lognormals fit the full data well.

Pareto power law approximates the upper tail of the distribution.

Lognormal and Pareto models are nearly indistinguishable in the upper tail.

Abstract

We study the distribution of strike size, which we measure as lost person days, for a long period in several countries of Europe and America. When we consider the full samples, the mixtures of two or three lognormals arise as very convenient models. When restricting to the upper tails, the Pareto power law becomes almost indistinguishable of the truncated lognormal.