Abstract
Karl Marx and Max Weber were the first to write about the link between culture and the economy. Although both hypothesized different directions of causality – Weber believed that culture influenced the economy, whilst Marx believed the opposite – they were in agreement that the topics were closely related. Economists have been slow to study the hypothesized linkages between culture and economic performance, especially using scientifically falsifiable methods. Some of this apprehension is due to the fact that economic studies largely rely upon mathematical modeling and statistical testing to investigate potential relationships. However, culture is difficult or, as some would argue, impossible to measure. Applying Hofstede's cultural dimensions to a state-level panel dataset spanning from 1998 to 2003, this project attempts empirically to measure the impact of these dimensions on entrepreneurial activity in the United States. The results are mixed – with two dimensions displaying coefficients with the opposite signs to those expected – thereby generating some interesting questions for future study. This is the first known paper to measure the impact of culture on regional entrepreneurial activity across the United States.
Acknowledgements
The author wishes to thank Robert Lawson, Russell Sobel and seminar participants at the Tinbergen Institute's International Workshop on Analysis and Modelling of Regional Development, the 77th Annual Meeting of the Southern Economic Association, and the 34th Annual Conference of the Northeast Business and Economics Association for helpful comments on earlier drafts of this paper.
Notes
1. For detailed discussions, see Holden (Citation2002) and Brons (Citation2006).
2. Hofstede (Citation2001, p. 361) states that “[a] strong respect for tradition impedes innovation”.
3. The annual state-by-state dimension calculations are available from the author.
4. The culture measures are stationary in their final ‘per-immigrant’ form.
5. This assumption will be relaxed later on. Results from the revised hypothesis tests show that the assumption is not a crucial factor. The author thanks an anonymous referee for suggesting these revised hypotheses.
6. The panel-corrected standard errors assume that there is no serial correlation (or that it has been removed) (Beck and Katz, Citation1995). Author-calculated Lagrange multiplier tests show no statistically significant serial correlation.