Costs and outcomes of collaborative relationships: interlocal collaboration effects on green job creation in Florida Metropolitan Areas

ABSTRACT

This study investigates how the costs of different network structures influence the effectiveness of collaborations in achieving their desired outcomes. Building from the ICA framework, an explanation is advanced for how the performance of collaborative arrangements is influenced by costs imposed on participating actors that result from the structure of the network and the actors positions within the network. Fixed-effects panel regression analysis is applied to collaborations on green economic development in Florida metropolitan areas over an eleven-year period. The findings demonstrate greater collaboration not only produces better outcomes, but that the effect of collaboration is influenced by threshold effects of collaboration costs on participants. Over-time high-cost mechanisms can reduce commitment to collaboration or lead to partially connected collaboration. We conclude that achieving better performance through collaborative networks may require governments to seek optimal patterns and positions with predictable and acceptable collaboration costs, rather than simply scaling up collective efforts.

KEYWORDS:

• Collaborative governance
• institutional collective action
• inter-municipal collaboration
• network performance
• collaboration risk
• green economic development

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. For the identification of metropolitan area in Florida, this study uses the Metropolitan Statistical Areas (MSA) defined by the Office of Management and Budget (OMB).

2. Based on the identification of green job provided by the U.S. Bureau of Labor Statistics (BLS), we employ the 6-digit North American Industry Classification System (NAICS) industry classification codes.

3. The information extraction (IE) techniques used in this study basically follow the same procedure with the named entity recognition (NLP) techniques in natural language processing, which first identify the entities to be extracted such as person and location, then perform the systematic review of filtering and refining the extracted entities.

4. For the definition of green economic development and its elements, we use diverse references of the United Nations Sustainable Development (UNSD), the United Nation Environment Program (UNEP), the Bureau of Labor Statistics (BLS), and the Energy Efficiency and Conservation Block Grant (EECBG) programs. To build a standardised concept list in which all relevant words consist of the concept of green economic development, we utilise Carley’s systematic coding schemes (Carley 1997) to filter 1,076 relevant words out of 1,315,776 words in references. For crawling unstructured text and converting it to machine-readable text, we use the ConText and AutoMap software.

5. The transitive triplets effect is defined by the number of transitive patterns in i’s relations, which is defined as${\mathrm{s}}_{i2}^{\mathrm{n}\mathrm{e}\mathrm{t}}\left(\mathit{x}\right)\mathit{\hspace{0.17em}}=\mathit{\hspace{0.17em}}\sum _{\mathrm{j},\mathrm{h}}{\mathrm{x}}_{\mathrm{ij}}{\mathrm{x}}_{\mathrm{ih}}{\mathrm{x}}_{\mathrm{jh}}$.

6. Although the original Berardo and Scholz hypothesis employs the indegree centrality as to measure the popularity of actor, interlocal agreement as a formal network assumes the presence of undirected network, thus makes no difference among indegree, outdegree, and total degree centrality in value. The popularity effect (Berardo and Scholz 2010) is defined as, ${\mathrm{s}}_{i1}^{\mathrm{n}\mathrm{e}\mathrm{t}}\left(\mathit{x}\right)=\mathit{\hspace{0.17em}}\sum _{\mathrm{j}}{\mathrm{x}}_{\mathrm{ij}}{\mathrm{x}}_{+\mathrm{j}}=\mathit{\hspace{0.17em}}\sum _{\mathrm{j}}{\mathrm{x}}_{\mathrm{ij}}\sum _{\mathrm{h}}{\mathrm{x}}_{\mathrm{hj}}$.

7. To test whether the random-effects (RE) estimator is biased, we employ the Breusch and Pagan Lagrangian multiplier test for random effects and the Hausman specification tests.

8. At the same time, however, it is limited that the homophily effects among networking local governments are not effectively controlled in the model, which can cause clustering errors in analysis.