An Evaluation of the Small Sample Size Properties of Tests of Localization

<p>Urban economists refer to the clustering of firms that belong to the same industry across spatial space as localization of economic activity. To determine if an industry is localized, researchers use indicators of localization and tests of statistical significance. While the asymptotic distributions of these tests are well-known, there is little information regarding the small sample size properties of the tests. Often, agglomeration researchers choose tests without knowledge of their properties and relative performance. This paper evaluates the small sample size properties of discrete tests of localization using Monte Carlo simulations to determine which tests are the most reliable and accurate. Based on our results we provide some recommendations on which tests to use and the situations where they perform better. To our knowledge this paper is the first known study to do so.</p>

Name thesis:degree_name

M.A.

Level thesis:degree_level

Campus Access Thesis

Discipline thesis:degree_discipline

Moore School of Business

Year

2010

Author dc:creator

• Milnichuk, Matthew Young

Contributors dc:contributor

• Paulo Guimaraes

• Economics
• Social and Behavioral Sciences
• Agglomeration
• Discrete
• Localization
• Simulations
• Tests

Statement dc:rights

• © 2010, Matthew Young Milnichuk

Repository record dc:identifier

https://scholarcommons.sc.edu/etd/222

OAI identifier oai:identifier

oai:scholarcommons.sc.edu:etd-1223