The effect of agglomeration economies on firm deaths: A comparison of firm and regional based approaches

Research output: Contribution to journalArticlepeer-review

Abstract

This paper compares the merits of regional and firm based approaches for analysing the effect of agglomeration economies on firm deaths in Ireland. We aggregate a comprehensive data set on Irish firm deaths to Electoral Division (ED) level, the lowest geographical scale available. Estimates of the effect of agglomeration on firm deaths from a regional analysis at ED level using a cross-sectional spatial-autoregressive spatial error model are compared with firm-level estimates from a contemporary log-log model with spatially weighted agglomeration regressors. While estimates of the effects of agglomeration using these alternative methods is much discussed in existing literature rarely are the approaches or results compared. We show that contrasting results are found using the same data set dependent upon the unit of analysis used. Diversity lowers regional and firm deaths while specialisation raises regional deaths but lowers firm deaths. Greater urbanisation does not have a significant effect on firm hazard rates or equivalent regional estimates. While regional estimates provide evidence on the existence and nature of spatial dependence (negative in this case), firm estimates do not provide evidence that agglomeration in neighbouring regions is the source of this spatial dependence. No empirical analysis to our knowledge directly compares regional and firm based approaches for analysing the effect of spatial agglomeration economies on firm deaths.

Original languageEnglish
Pages (from-to)3358-3374
Number of pages17
JournalUrban Studies
Volume56
Issue number16
DOIs
Publication statusPublished - 1 Dec 2019

Keywords

  • firm deaths
  • localisation
  • spatial agglomeration
  • specialisation
  • variety

Fingerprint

Dive into the research topics of 'The effect of agglomeration economies on firm deaths: A comparison of firm and regional based approaches'. Together they form a unique fingerprint.

Cite this