Results and Interpretation
Table 7.2 presents the results for the observations centered on the year 2000 (the baseline specifications). Columns 1 through 3 apply the productive entrepreneurship scores as a dependent variable, columns 4 through 6 use the unproductive entrepreneurship scores, while columns 7 through 9 apply the NEP scores.
Further, columns 1, 4, and 7 only include the IEI score, columns 2, 5, and 8 add the socioeconomic control variables, while columns 3, 6, and 9 include the regional dummy variables.The results are quite consistent and corroborate the correlations presented in Figs. 7.1-7.3. Specifically, higher IEI scores are consistently associated with lower levels of productive entrepreneurship, with 2 specifications statistically significant (including regional dummies resulting in insignificance at conventional levels). Columns 4 through 6 clearly indicate that higher IEI scores are correlated with higher levels of unproductive entrepreneurship, again with 2 of 3 specifications significant (here column 5—the inclusion of socioeconomic variables—results in insignificance). Finally, columns 7 through 9 indicate that higher IEI scores are associated with lower levels of net entrepreneurial productivity, with all specifications statistically significant.
The coefficients are difficult to interpret given the makeup of each of the indices, however beta coefficients (presented in brackets for each of the main independent variables of interest) can provide some indication of the magnitude of those coefficients. Specifically, a one standard deviation increase in the IEI score results in productive entrepreneurship declining anywhere between 0.19 and 0.46 standard deviations depending on specification, unproductive entrepreneurship increasing between 0.17 and 0.33 standard deviations, and NEP declining between 0.32 and 0.46 standard deviations.
In order to quantify these magnitudes, consider the median state in each category, which is Tennessee for productive entrepreneurship (with a score of 22.6), Wisconsin for unproductive entrepreneurship (with a score of 23.5), and Minnesota for NEP (with a score of -0.15). Here, a one standard deviation increase in the IEI score would result in Tennessee falling to anywhere between 27th and 30th in the overall ranking of states.
162 J. A. DOVE
Note Robust standard errors in parentheses. Constant suppressed in output tables. Beta coefficients included [in brackets] for the IEI variable, ***p < 0.01, **p < 0.05, *p < 0.1
Source Author’s creation
Wisconsin would increase in its ranking of unproductive entrepreneurship roughly 5 spots to 20th, while Minnesota would decline roughly 9 spots to 33rd in overall NEP ranking.
Finally, I consider how 10- and 20-year lagged values of the IEI scores influence entrepreneurial activity. Again, these lagged results are included for several reasons. First, while the IEI scores are highly persistent across time, it still may be the case that the initial findings are biased as a result of reverse causality. Lagged IEI scores can help alleviate this issue. Additionally, including lagged values can also help explain whether such policies create network effects and lead to persistence in the types of entrepreneurship that develop. Results for the 10-year lag are reported in Table 7.3 and 20-year lag reported in Table 7.4. Each table follows the layout of Table 7.2.
As can be seen, the results are extremely similar to those reported in Table 7.2. Specifically, the signs and significance are the same as those in Table 7.2, while the magnitudes of the beta coefficients are also similar across tables and specifications.
Overall then, these findings suggest that the greater availability and more liberal opportunities to employ and offer economic development incentives are negatively related to productive entrepreneurship and net entrepreneurial productivity and positively related to unproductive entrepreneurial activity.
These results are suggestive of several policy implications. First, proponents of economic development incentives are in fact correct that such incentives can stimulate entrepreneurial activity. However, it seems that it is actually unproductive entrepreneurship that is stimulated, which is in all likelihood the opposite of the intent or desire of those proponents. My findings would thus suggest that rather than being allocated in a manner that mimics market allocation, the opportunity to capitalize on such incentives stimulates greater rent-seeking behavior. Additionally, the results from the lagged IEI scores also indicate persistence and potential network effects that become entrenched within an economy due to the prevalence of such incentives and in line with the theoretical results of Coyne et al. (2010).Given the above two findings then, it would seem that the only effective way to limit these unintended outcomes and to stimulate greater productive entrepreneurial activities would be to better limit the scope of authority that many state and local governments are currently constitutionally granted to promote and aid private entities with the use of public resources. With greater limits in place this should, over time, change
164 J. A. DOVE
Note Robust standard errors in parentheses. Constant suppressed in output tables. Beta coefficients included [in brackets] for the IEI variable, ***p < 0.01, **p < 0.05, *p < 0.1
Source Author’s creation
ECONOMIC DEVELOPMENT
166 J. A. DOVE
Note Robust standard errors in parentheses. Constant suppressed in output tables. Beta coefficients included [in brackets] for the IEI variable, ***p < 0.01, **p < 0.05, *p < 0.1
Source Author’s creation
ECONOMIC DEVELOPMENT
the relative payoff between productive and unproductive entrepreneurship and result in greater productive opportunities, freeing up additional private capital to be channeled into these outlets, and ultimately crowding out unproductive entrepreneurial endeavors. Additional policy responses would be to better streamline the overall institutional environment so as to increase competitiveness and attract productive entrepreneurship. Though the specifics of this latter point are beyond the scope of this current paper, they would be worthy of future research.