Predicting firm creation in rural Texas: A multi-model machine learning approach to a complex policy problem

Area of Science:

• * Rural entrepreneurship and economic development.
• * Applied machine learning for socioeconomic analysis.
• * Comparative studies of firm creation.

Background:

• * Rural and urban areas face increasing political and economic divisions.
• * Entrepreneurship is a key strategy for rural economic revitalization, job creation, and resilience.
• * Limited research exists on firm creation specifically within rural contexts compared to urban settings.

Purpose of the Study:

• * To identify predictive factors for firm creation in rural America.
• * To address gaps in entrepreneurship research regarding comparative variable importance, focus on urban/high-tech firms, and application of modern machine learning.
• * To provide insights for policymakers on fostering rural economic growth.

Main Methods:

• * Application of four machine learning methods: subset selection, LASSO, random forest, and extreme gradient boosting.
• * Analysis of a novel dataset examining rural Texas counties from 2008-2018.
• * Comparative framework to assess the predictive importance of various socioeconomic and industry-specific factors.

Main Results:

• * Socioeconomic factors such as age distribution, ethnic diversity, social capital, and immigration are more predictive of rural firm growth than broadband access or patents.
• * Industry strength (oil, wind, healthcare, elder/childcare) and the number of local banks also predict firm growth.
• * Predictive factors for rural firm growth differ significantly from those in urban areas.

Conclusions:

• * Rural entrepreneurship is a distinct phenomenon requiring tailored strategies and focus.
• * Machine learning models effectively identify complex, multifactorial drivers of rural firm creation.
• * Findings offer practical guidance for policymakers aiming to stimulate rural economies.