Domestic migration and new business creation in the United States

Abstract

While an extensive line of research links international migration to entrepreneurship, less attention has been devoted to mobility within a country. This study examines the relationship between new business creation and migration between states in the United States, building on prior research associating education and entrepreneurship. In addition to tracking general migration patterns among the states, the study focuses on young, single and college-educated segment of the population. This segment may be less risk averse and more geographically mobile, and brings knowledge resources to the geographic areas where they choose to live. While overall levels of net migration did not predict new business creation, prior net migration of the young, single, college-educated was a statistically significant predictor. States that are attracting young, single, highly educated people are more entrepreneurially dynamic than would be indicated by overall levels of migration.

Bien qu'un axe de recherche important établisse le lien entre migration internationale et entreprenariat, peu d'attention a été portée sur la mobilité à l'intérieur d'un pays donné. La présente étude examine la relation entre la création d'entreprises nouvelles et les migrations entre les états aux Etats-Unis, en s'appuyant sur de précédentes recherches associant éducation et entreprenariat. Outre le fait de suivre les modèles migratoires généraux entre les états, l'étude se concentre sur le segment de la population qui est jeune, célibataire et a fait des études universitaires. Ce segment tend à être moins réticent à prendre des risques et plus mobile géographiquement, et il apporte des ressources de connaissances aux zones géographiques dans lesquelles ces personnes décident d'habiter. Alors que les taux globaux de migration nette ne permettaient pas de prévoir la création d'entreprises nouvelles, la migration nette préalable des jeunes célibataires ayant une formation universitaire a constitué un prédicteur statistiquement significatif. Les états qui attirent des personnes jeunes, célibataires et ayant fait des études poussées sont plus dynamiques d'un point de vue entrepreneurial que ce que les niveaux globaux de migration permettaient de prédire.

Keywords:

• migration
• entrepreneurship
• geographic mobility
• human capital
• creative class

Mots clés:

• migration
• entreprenariat
• mobilité géographique
• capital humain
• classe créative

Acknowledgements

This research was supported by a Langroise Faculty Scholarship Endowment grant.

Notes

^aTabled means and standard deviations are for untransformed variables.

^bYoung, single, college-educated.

^cPercentage of population aged 25 or more with at least a bachelor's degree.

^dLog10 transformation used in analyses.

*p < .05; **p < .01; n = 51.

^aStandardized regression coefficients are tabled for each independent variable.

^bYoung, single, college-educated.

^cPercentage of population aged 25 or more with at least a bachelor's degree.

*p < .05; **p < .01; ***p < .001; n = 51.

1. Because the migration could have occurred any time between 1995 and 2000 and the census was collected in 2000, some of the migrants could have been as much as five years younger at the time of their inter-state move.

2. All variables were examined for normal distribution before inclusion in the analyses. The measure of state population, as well as the Hispanic/Latino portion of the population, is positively skewed, and the appropriate log transformations were performed. The average wage salary variable showed positive skew created by the extreme value for Washington D.C. (average wage salary of $69,312). The overall net migration variable was negatively skewed. This was caused by two outliers: New York (874,248) and California (755,536). The distribution of the net migration of the young, single, college educated was positively skewed because of only one extreme outlier, California (73,037). Similarly, one extreme outlier (Washington D.C.) was found in the distribution of the percent unmarried variable. The distributions of these measures were corrected by assigning the univariate outliers a value equivalent to three standard deviations above the mean. This preserves the ordering of these states on the measures, while allowing the observations not to be so deviant that they perturb the correlation (see Tabachnick and Fidell 1983). The state economic freedom measure was not available for Washington D.C., so the mean for this measure was used for that observation.

3. Net migration is a function of both in-migration and out-migration. Ancillary analyses separated these two components of migration for the young, single, college educated. Both types of movement were seen as critical in predicting subsequent business creation (data not tabled).

4. Because our literature review refers to international migration as well as domestic movement, ancillary analyses also examined the relationship between the two. Several interesting relationships were found. The correlation between overall levels of net domestic migration and the percentage of the state population that was born in a different country was −.28 (n.s.), while the net movement of the young, single, college educated was positively correlated with the stock of foreign born (r = .52, p < .01). In terms of flows of immigration from other countries, inward international immigration (relative to state population) between 1995 and 2000 was not related to overall domestic migration patterns (r = −.04, n.s.) during that time, but was significantly related to the net domestic migration of the young, single, college-educated cohort (r = .55, p < .01). That is, the young, single, college-educated cohort is generally attracted to the same states as international immigrants. However, the movement of the young, single and college educated was significantly correlated with subsequent levels of entrepreneurship (r = .46, p < .01), while the flow and stock of international immigration, though positive, were not statistically significant predictors (r = .21, n.s.; r = 17, n.s.).