Page 112 - 1-Entrepreneurship and Local Economic Development by Norman Walzer (z-lib.org)

The Drivers of Regional Entrepreneurship in Rural and Metro Areas 101

edge Clusters and Entrepreneurship in Regional Economic Development Confer-
ence. Minneapolis, Minnesota.
Rappaport, Jordan. 2003. Moving to nice weather. Federal Reserve Bank of Kansas City
Research Working Paper, 03–07.
Bureau of Economic Analysis–Regional Economic Information System (BEA–REIS),
1969–2004. U.S. Department of Commerce. CD.
Rubin, Sarah. 2001 Rural colleges as catalysts for community change, the RCCI ex-
perience. Rural America 16(2): 12–19.
Small Business Administration. 2005. The innovation-entrepreneurship NEXUS: A na-
tional assessment of entrepreneurship and regional economic growth and development
www.sba.gov/advo/research/rs256tot.pdf (February 20, 2007).
Sutaria, Vinod, and Donald Hicks. 2004. New firm formation: Dynamics and deter-
minants. Annals of Regional Science 38: 241–62.
U.S. Bureau of the Census, Department of Commerce. 2004. Educational attainment
in the United States: 2003. www.census.gov/prod/2004pubs/p20-550.pdf (Febru-
ary 20, 2007).

APPENDIX: EMPIRICAL ANALYSIS

An empirical model of entrepreneurial breadth and depth was estimated to
analyze the relationship of various community characteristics and the
quantity and value of entrepreneurial activity at the county level. Based on
the five core categories of hypothesized entrepreneurial drivers suggested by
existing research, an empirical model was estimated where the entrepre-
neurial breadth and depth measures were included as dependent variables
and independent variables were included to measure human capital,
amenities, financial capital, infrastructure, and other features of the local
economic landscape.
The empirical model was estimated in linear form, with results of the
three regressions reported in table 5.3. Variance inflation factors were less
than 2 suggesting that multicollinearity is not a significant issue in estima-
tion. Adjusted R-squares range from 0.09 to 0.32, satisfactory levels for
cross-sectional analyses, and F-statistics for all equations are significant at
the 0.05 percent level. The Hausman Specification Tests on the results from
initial ordinary least squares regressions detects a simultaneity problem be-
tween the dependent variables and the explanatory variables. A two-stage
least squares (2SLS) estimation method was implemented to reduce the ef-
fects of simultaneity and resulted in coefficient similar in sign and signifi-
cance to OLS results. The White Test does not indicate heteroskedasticity in
the data and residual plots show few outlying observations. Nevertheless,
we still tried weighting the 2SLS equations by population, resulting in co-
efficients of similar sign and significance to their unweighted equivalents.
Given these findings, we focus on the unweighted results.