Small Business Clusters in Oklahoma: MAR or Jacobs Effects?

Mihai NicaEERS 2010

TL;DR

This paper investigates whether the growth of small businesses in Oklahoma is driven by industrial specialization (Marshall-Arrow-Romer effects) or diversity (Jacobs effects). Using spatial statistics and regression analysis on county-level data, the study finds stronger evidence for Marshall-Arrow-Romer effects, suggesting that specialization plays a more significant role in small business cluster formation and growth.

What Problem Does This Paper Address?

The formation and growth of small business clusters are crucial for economic development, yet the underlying mechanisms are not fully understood. Two prominent theories, the Marshall-Arrow-Romer (MAR) and Jacobs effects, offer competing explanations. MAR theory posits that regional specialization and economies of scale drive growth, while Jacobs theory emphasizes the role of industrial diversity and competition. This paper aims to empirically test which of these theories better explains the dynamics of small business clusters, specifically in Oklahoma, to inform regional economic policy.

Methods

The study employs a mixed-methods approach combining standard statistical analysis with spatial statistics. Key techniques include:

  • Herfindahl-Hirschman Index (HHI): Used to measure industrial concentration (specialization) at the county level. A higher HHI indicates greater specialization.
  • Regression Analysis: Used to examine the relationships between variables like industrial concentration, employment, and the number of small businesses (NSB) over time. Both static (cross-sectional) and dynamic (panel) models are utilized.
  • Spatial Statistics (LISA - Local Indicator of Spatial Association): Applied to identify statistically significant spatial clusters of high/low NSB and high/low industrial concentration. This helps to understand the spatial patterns and interdependencies of small business activity.

The novelty lies in applying these spatial analysis tools specifically to small business cluster dynamics and comparing the MAR and Jacobs theoretical frameworks within this context.

Datasets

The primary dataset used is the County Business Patterns (CBP) database from the U.S. Bureau of the Census for 1998 and 2005. This data provides information on the number of small businesses (defined as those with fewer than 500 employees) and employment by county and industry. Geographically referenced files from the Minnesota Population Center were also used for spatial analysis.

Key Insights and Ideas

  • Specialization Dominates Diversity: The most significant finding is the preponderant influence of Marshall-Arrow-Romer (MAR) effects. This suggests that areas with a higher degree of industrial specialization tend to foster more small business growth.
  • Twin-Peaked Distribution: The distribution of small businesses exhibits a twin-peaked pattern, suggesting the existence of two distinct "clubs" of economic activity that diverge over time, rather than a convergence towards a single equilibrium.
  • Spatial Clustering is Key: Spatial analysis reveals significant clusters of high NSB and high industrial concentration, particularly around major cities like Oklahoma City and Tulsa, reinforcing the idea that agglomeration economies are at play.
  • Growth Divergence: The study points to a strong NSB growth divergence process, where areas with already high NSB tend to grow faster, rather than a catch-up effect for lagging regions.

Key Results

  • MAR Static Effects: The study found significant MAR static effects for 2005, indicated by a positive and significant relationship between industrial concentration (HHI) and NSB.
  • MAR Dynamic Effects: Significant MAR dynamic effects were also found for the period under scrutiny, suggesting that past industrial concentration influences current NSB growth.
  • Negative Relationship with Diversity: In some regressions, a negative relationship was observed between industrial concentration and NSB growth, suggesting that areas with higher specialization exhibit lower employment and NSB, which is interpreted as a possible indication of negative MAR effects (though this seems to contradict the main finding, it highlights the complexity).
  • Spatial Clusters and Growth: High growth (HH) clusters for NSB were found around major cities, often coinciding with high industrial concentration. Counties in Low-Low (LL) spatial clusters (indicating low NSB and low concentration) showed lower NSB growth compared to other clusters.
  • NSB Growth vs. Employment Growth: A significant positive relationship was found between initial employment and NSB growth, but not necessarily with overall employment growth, suggesting that new small businesses might be fragmenting the labor market rather than creating entirely new jobs.

Limitations and Future Work

  • Inconclusive for 1998 Static Effects: While MAR static effects were evident for 2005, the results for 1998 were inconclusive.
  • Spatial Cluster Influence: The influence of base-year spatial clusters on NSB growth was found to be minimal.
  • "Clubbing" Phenomenon: The observed twin-peaked distribution and growth divergence suggest a "clubbing" phenomenon, where distinct groups of regions follow different growth trajectories. Further in-depth analysis is needed to understand if this is a normal consequence of population dynamics or an endemic issue.
  • Policy Implications: The findings suggest that policies aimed at fostering industrial specialization might be more effective for promoting small business clusters in Oklahoma, though the complex interplay of factors warrants careful consideration.

The paper concludes that while both MAR and Jacobs effects are discussed in the literature, the evidence in Oklahoma leans heavily towards MAR effects, indicating that specialization is a key driver for small business cluster formation and growth.

Summary updated

Abstract

One of the main factors that act as catalysts for small business creation seems to be the existence of business clusters. This study uses both standard and spatial statistics methodologies and tools to analyze the dynamics of small business clusters and the degree to which it is influenced by Marshall-Arrow-Romer or Jacobs effects. The study contributes to better understanding of the influence that industrial diversity or specialization has on the growth of the number of small businesses and, consequently, on small businesses cluster formation. Maybe the most significant finding of this study is that the evidence suggests that Marshall-Arrow-Romer effects are preponderant. Indeed, the study finds significant MAR dynamic effects for the period under scrutiny, as well as Marshall-Arrow-Romer static effects for 2005.
BibTeX
@inproceedings{nica2010small,
  author = {Mihai Nica},
  title = {{Small Business Clusters in Oklahoma: MAR or Jacobs Effects?}},
  booktitle = {EERS},
  year = {2010},
}
Figure