Creating a policy environment for entrepreneurs.

• Author: Garrett, Thomas A.

Entrepreneurship is often viewed as a catalyst for economic growth. Through innovation, hard work, and a willingness to accept financial risk, the entrepreneur takes advantage of previously undiscovered opportunities for arbitrage and profit (Kirzner 1997). (1) This quest for profit, and the possibility of personal and financial failure, aid in ensuring that an economy's resources are used efficiently. Successful entrepreneurs provide employment opportunities to others, generate innovation, spur economic growth, and contribute to state and local governments in the form of tax revenue (Gwartney, Holcombe, and Lawson 2004; Kreft and Sobel 2005). Because of this perception of the benefits generated by entrepreneurship, a large literature has focused on the factors that influence the decision of an individual to become an entrepreneur and the conditions under which entrepreneurship prospers.

Previous research on entrepreneurship has examined the roles of various demographic, human capital, and financial considerations in the decision to become an entrepreneur. Rees and Shah (1986), Gill (1988), and Hamilton (2000) stressed the importance of the earnings differential between entrepreneurship and paid employment. Liquidity constraints on entrepreneurship were addressed by Evans and Jovanovic (1989), Evans and Leighton (1989), and Holtz-Eakin, Joulfaian, and Rosen (1994a, 1994b). Personal and job satisfaction differentials between entrepreneurship and paid employment have been addressed in Taylor (1996), Blanchflower and Oswald (1998), and Blanchflower (2000). In addition, Blanchflower, Oswald, and Stutzer (2001), Georgellis and Wall (2000a), and Beugelsdijk and Noorderhaven (2004) examined the importance of social factors, or latent entrepreneurship, in explaining differences in entrepreneurship across countries and regions, respectively.

This article examines the influence of government policy on rates of entrepreneurship across U.S. states, a topic that has been receiving increasing attention. Recent research has explored the influence of several state-level policies on entrepreneurship, such as personal income tax rates, bank deregulation, and bankruptcy laws. (2) We extend this literature by considering other policies, such as corporate income tax rates and state minimum wages. Furthermore, the flexibility of our empirical model accounts for potential nonlinearities between the policy variables and the rate of entrepreneurship in a state.

We obtain estimates of the effects of government policies on entrepreneurship by exploiting the differences in entrepreneurship and policies across the 50 U.S. states during the 1990s. Throughout, we define the rate of entrepreneurship as the share of the working age population (16-64) who are proprietors. We exclude farm proprietors, as does previous research, on the basis that the decision to become a farm proprietor depends upon different factors than the decision to become a nonfarm proprietor. As summarized by Table 1, there were substantial differences in state rates of entrepreneurship at the beginning and the end of the period. For example, in 1990 there were two states--Mississippi and South Carolina--whose rates of entrepreneurship were less than half that of Alaska, the most entrepreneurial state. (3) The decade saw significant upward movement in entrepreneurship: The average of state rates of entrepreneurship went from 13.5 percent in 1990 to 15.8 percent in 2000, and all but two states saw higher rates of entrepreneurship in 2000 than in 1990.

One of our objectives is to determine whether the geographic pattern of entrepreneurship is related to the geographic pattern of policy environments. In 1990 and 2000, New England and the West were the most entrepreneurial regions, with the South and Great Lakes regions lagging. The geographic pattern of changes in entrepreneurship is less clear than the difference in the levels of entrepreneurship. Although some of the entrepreneurial states in New England and the West saw the largest increases in entrepreneurship, some of the lagging states, particularly in the South, also saw large increases.

The Empirical Model

Our empirical model extends that of Georgellis and Wall (2000a) by adding a vector of explanatory variables that controls for the policy environment:

(1) [E.sub.it] = [α.sub.i] + [τ.sub.t] + β' [X.sub.it] + θ' [Z.sub.it] + γ' [G.sub.it] + [ζ.sub.it].

In equation (1), the dependent variable [E.sub.it] is the rate of entrepreneurship in state i during year t. The parameter [α.sub.i] denotes state fixed effects and [τ.sub.t] denotes year effects. The vector [X.sub.it] measures average demographic characteristics, and the vector [Z.sub.it] measures business conditions. The policy environment is captured by the vector of policy variables, [G.sub.it]. Finally, [ζ.sub.it], is the error term. Data sources and summary statistics for all variables used in the estimation are provided in Tables 2 and 3.

The demographic variables in [X.sub.it] measure the age, gender, and racial compositions of state employment, categories across which rates of self-employment differ a great deal (Georgellis and Wall 2000b). For example, men are nearly twice as likely as women to be self-employed, and blacks are less than one-third as likely to be self-employed as whites or Asians. Our vector of business conditions, [Z.sub.it], includes the state's unemployment rate, per capita real income, industry employment shares, real proprietor's wage, per capita real wealth (as proxied by dividends, interest, and rent), and the real median house price weighted by the rate of home ownership. These last two variables control for differences in the levels of assets that the average person has to support an entrepreneurial venture.

Care needs to be taken when interpreting the estimated coefficients for the variables in [X.sub.it] and [Z.sub.it]. These variables might simultaneously measure differences across states in the supply of entrepreneurs and the demand for the products that are more likely to be produced by entrepreneurs. For example, more than 10 percent of self-employed women in 1997 were in the child-care business, while virtually no men were (Georgellis and Wall 2000b). On the one hand, a state with a relatively high share of females might have a relatively high supply of child-care providers, and therefore have more self-employed women. On the other hand, the state also has relatively more women demanding child-care services, thereby making the state a relatively lucrative market for self-employed child-care providers. Therefore, because supply and demand cannot be separated by the variables in [X.sub.it], and [Z.sub.it], we include them only as controls and do not interpret their coefficients.

An exception is the unemployment rate, which is a measure of the health of a state's economy. A low unemployment rate suggests relatively low risks and high returns for entrepreneurial ventures, thereby pulling a higher share of the population into entrepreneurship. In Parker (1996), however, a high unemployment rate indicates the number of people with limited opportunities for wage-and-salary employment who might be pushed into self-employment out of necessity. Thus, the sign of the coefficient on the unemployment rate has been interpreted as a measure of the relative strengths of the pull and push effects of the unemployment rate.

The Policy Environment

The variables of greatest interest in this article are the four measures of state policy in the vector of policy variables, [G.sub.it]. This vector includes measures of bankruptcy laws, personal income taxes, corporate income taxes, and the minimum wage.

Homestead Exemption

State bankruptcy laws allow those filing for personal bankruptcy to exempt some of their assets and income from creditors. The exemptions can include some or all of the value of the person's home, pension, and a host of other assets. Because an entrepreneur's home is likely to be his or her most valuable asset, recent studies have focused on the possibility of a link between the homestead exemption and levels of entrepreneurship (Berkowitz and White 2004; Fan and White 2003; Georgellis and Wall 2006). These studies have posited two opposing effects. The first effect arises because a potential entrepreneur views the level of the homestead exemption as insurance against the failure of an entrepreneurial venture. If one's home is not subject to distribution to creditors, a potential entrepreneur is more likely to take on the increased risk of being an entrepreneur instead of being a wage-and-salary employee. In addition to this wealth-insurance effect, however, the homestead exemption creates a credit-access effect. Banks and other creditors are aware of bankruptcy exemptions and adjust the availability of credit accordingly. Thus, by making credit more difficult to come by, the homestead exemption might reduce the number of entrepreneurs.

Our homestead exemption rate is a measure of the percentage of the average person's homestead that would be protected from creditors in the event of personal bankruptcy. In creating the variable, we need to account for several state-level differences in the treatment of homesteads during bankruptcy proceedings. The primary source of these differences is the homestead exemption--the amount of a home's value that is exempt from bankruptcy proceedings. Cross-state differences in the homestead exemption are summarized in the first data column of Table 4. These differences are significant: In 1997, six states did not allow for any amount of the value of a person's home to be exempt from distribution to creditors, but eight other states placed no limit on the amount that could be exempted.

The homestead exemption rate is constructed to allow also for the fact that some states permit fliers to use the federal exemption level and that some states allow married fliers to double the exemption...