Analysis of the Labour Market in Metropolitan Areas: A Spatial Filtering Approach

Tomasz Kossowski,Jan Hauke

doi:10.2478/v10117-012-0017-5

Abstract

Analysis of the Labour Market in Metropolitan Areas: A Spatial Filtering Approach The power of today's computers allows us to perform computation on massive quantities of data on the one hand and produces enormous amounts of analysis output on the other, as noted by Griffith in his 2003 book. Besides, visualisation and spatial filtering (the core of considerations in Griffith's book) have a chance to be widely used in research practice, especially in geosciences and, more precisely, for georeferenced data. Following the idea proposed by Patuelli et al. (2006, 2009), we analysed the labour market in Poland, focusing on metropolitan areas and their surroundings. The analysis was performed on a data set for the unemployment rate in the 2,478 Polish communes. We took into account spatial autocorrelation and used spatial filtering techniques to construct components of an orthogonal map pattern. As shown in Tiefelsdorf & Griffith (2007), the spatial filtering techniques could be employed in both, parametric and semi-parametric approaches. In this paper we adopted a parametric one.

Highlights

In his significant book (2003:1), Griffith noted that: 1) „At least since the dawn of civilization data have been analyzed as numerical figures to support a decision or to understand a part of reality.” 2) „One consequence of the massive quantities of data collected and analyzed today is the enormous amount of analysis output.” 3) „Much of the data collected today are georeferenced, or tagged to the Earth’s surface (...)”
We focused on metropolitan areas and their surroundings
In this article we presented an analysis of the labour market in Poland, with particular emphasis on selected metropolitan areas

Summary

Introduction

In his significant book (2003:1), Griffith noted that: 1) „At least since the dawn of civilization data have been analyzed as numerical figures to support a decision or to understand a part of reality.” 2) „One consequence of the massive quantities of data collected and analyzed today is the enormous amount of analysis output.” 3) „Much of the data collected today are georeferenced, or tagged to the Earth’s surface (...)”. Commonly used computers are sufficiently powerful to perform calculations for quite large collections of data employing ever more complicated numerical methods and sophisticated algorithms. There are many spatial econometric procedures for a statistical analysis of georeferenced data available in the literature. The method is based on spatial weights matrices measuring the spatial dependence between values of georeferenced variables. Owing to the bias of statistical efficiency and the problem of independence assumption, it is not advisable to use the ordinary least squares (OLS) method with the data. In the procedure spatial filters are computed (Griffith 1996, 2000). This technique is based on the computational formula of Moran’s I statistic

Data and methods

Results

Conclusions