Data and Empirical Specification

Data for this study come from a number of sources, the most important of which are related to measures of the various types of entrepreneurship as discussed above and also economic development incentives.

In order to measure productive entrepreneurship, Sobel (2008) includes measures of per capita venture capital investment, per capita patents, the growth rate of self-employment activity, all firm establishment birth rates, and the birth rate of firms with 500 or more employees. These variables are averaged over several years, centered on 2000. Four measures of unproductive entrepreneurship are also applied, three of which evaluate the number of lobbying and political organizations residing within each state's capital and one measure of legal quality within a given state.

The subcomponents of each measure of productive and unproduc­tive entrepreneurship are then indexed through a Borda Count, which normalizes these variables into two comparable measures ranging between “1” and “48,” with a higher score indicating relatively more of the particular type of entrepreneurship. From there, it is possible to derive a “net entrepreneurial productivity score” (NEP) which is the difference between the return to productive relative to unproductive entrepreneur­ship. Thus, a positive NEP indicates that a particular state has relative more productive to unproductive entrepreneurship. NEP scores range between a low of “-47” and a high of “47.”

Data for state economic development incentives is drawn from Patrick (2014a) who develops an Incentives Environment Index (IEI).

The three specific constitutional restrictions evaluated include credit clause restrictions (which dictate how and if a state or local government may use state credit to aid private enterprise) and if certain approval requirements exist before such aid can be extended. Current appropri­ations clauses shape whether and how a state government may use cash subsidies, land grants, public loans, or fund other similar activities. Finally, stock clauses limit the nature of the financial relationship between the public sector and private firms. They specifically govern the nature of private-public partnerships, investment in seed capital, and various other types of public cooperation or ownership. Thus, the index is based on six sub-indices (credit clause restrictions, current appropriation restrictions, and credit restrictions for state governments).

Based on these variables, I would expect that if more liberal state development incentive options were being channeled in a manner that approximates market allocation, then both productive entrepreneurship and NEP would be positively associated with the IEI index, while unpro­ductive entrepreneurship would be negatively correlated.

then I would anticipate the opposite result. Figures 7.1, 7.2, and 7.3 provide a simple correlation between these three entrepreneurial variables and the IEI index.

Figure 7.1 shows the relationship between productive entrepreneur­ship and the IEI index, Fig. 7.2 includes unproductive entrepreneurship, while Fig. 7.3 uses the NEP scores. Here, Figs. 7.1 and 7.2 show a clear negative correlation between both productive entrepreneurship and the IEI as well as the NEP and IEI. Further, Fig. 7.2 shows a clear positive correlation between unproductive entrepreneurship and IEI. Thus, as the availability of economic development incentives becomes more permissible and liberal, it suggests that they are associated with growth in unproductive entrepreneurship and lower levels of productive entrepreneurship. Overall, this leads to lower levels of net entrepreneurial productivity.

In order to more thoroughly tease out causation, I evaluate the following cross-sectional econometric model:

[Entrepreneurship] _i = α+βj. [IEI] _i + δj^^, |3_2 + e_i (1)

Here [Entrepreneurship] _i represents each of the three measures of entrepreneurship (productive, unproductive, and the NEP) for each

Fig. 7.1 Productive entrepreneurship and economic development incentives (Note Entrepreneurship measure from Sobel [2008], Incentives index from Patrick [2014a]. Source Author’s creation)

Fig. 7.2 Unproductive entrepreneurship and economic development incen­tives (Note Entrepreneurship measure from Sobel [2008], Incentives index from Patrick [2014a]. Source Author’s creation)

Fig.

state i, while [IEI] _i represents the incentives index for each state i. B_F’ is a vector of control variables, typical to the literature and includes the percentage of the population that is male, the percentage of the population with a bachelor’s degree or higher, the unemployment rate by state, real per capita GDP, and regional dummies for each of the US Census regions (Northeast, Midwest, West, and South) to help control for as many unobservable regional characteristics that might influ­ence entrepreneurship as possible. These latter variables are all averaged between 1995 and 2000, which loosely corresponds to the years aver­aged in calculating measures of entrepreneurship. Table 7.1 provides the summar y statistics for each of these variables.

One final issue to address is the potential reverse causality that might exist between the types of entrepreneurship and economic development incentives. While no particularly strong instrument exists, as a second best I also include 10- and 20-year lagged values of the incentive index. This is done for two reasons. First, it can act as a second-best solution to issues of reverse causality, and second it can help indicate the extent to which

Table 7.1 Summary statistics

Source Author’s creation

persistence in the IEI might lead to even more productive or unpro­ductive entrepreneurship, suggesting that network effects as discussed in Coyne et al.