Explaining variation in rates of entrepreneurship in the United States: 1899-1988.

Entrepreneurship is becoming an increasingly important phenomenon for researchers and policy makers to understand. It is important to job growth, accounting for eighty percent of new jobs created in the United States (Birley, 1986; Reynolds, 1987; Birch, 1979). It is a key factor in enhancing technological innovation, the growth of productivity and per capita incomes in industrialized nations (Baumol, 1986; President's Commission Report, 1984; Low & MacMillan, 1988). And, it is an important vocational choice, accounting for roughly one-tenth of the labor force at any point in time (Evans & Leighton, 1986), and approximately one-fifth of the labor force for some part of their careers (Steinmetz & Wright, 1989).

The importance of entrepreneurship to employment, innovation, productivity and income growth has led increasing numbers of researchers to investigate why entrepreneurship occurs. Most of this research has tended to take a "traits" approach to entrepreneurship. The "traits" approach views entrepreneurs as individuals with unique characteristics, the self-knowledge of which "motivates individuals to establish their own enterprises" (Aronson, 1991, p. 21).

While valuable, this approach has suffered from three weaknesses that hamper our understanding of entrepreneurship. First, the samples in "traits" studies are often selected on the basis of the dependent variable (entrepreneurial experience) and thus suffer from sample selection bias (Heckman, 1979; Carroll & Mosakowski, 1987). Second, much of this research looks at only entrepreneurs from a single industry or area at a time and therefore must generalize from small, unique samples (Mathias, 1967; Singh, 1985; Carroll & Mosakowski, 1987). Third, the "traits" approach often uses cross-sectional research designs which assume that rates of entrepreneurship are constant over time (Tuma & Hannan, 1984), despite evidence from longitudinal studies which show that rates of entrepreneurship are quite turbulent (Aldrich, 1990; Phillips, 1962; Aronson, 1991).

The weaknesses of the "traits" approach have led some scholars to view entrepreneurship in a different light. These scholars (who form the "rates" school) view entrepreneurship as a more environmentally determined phenomenon. To the "rates" school, the self-employed are not endowed with special abilities that differentiate them from individuals in wage and salary jobs, but instead are individuals who are responding to the environmental circumstances in which they find themselves (Aronson, 1991, p. 21). This school of thought argues that a greater understanding of entrepreneurship will come from longitudinal studies of rates of entrepreneurship (Aldrich, 1990; Stevenson & Harmeling, 1990).

While the research of the "rates" school has avoided many of the methodological problems encountered by scholars of the "traits" school, it has suffered from four weaknesses. First, much of the research in this school focuses on a single population at a time, thereby failing to consider the interdependence between different populations (Reynolds, 1992). Second, this research has rarely controlled for variables that "traits" scholars have found to be significant, making it difficult to rule out trait-based explanations for variations in rates of entrepreneurship. Third, much of this research has failed to incorporate the effects of changes in technology on rates of entrepreneurship. Finally, this research has failed to adjust for changes in the characteristics of the sample over time, such as that due to population growth.

This paper seeks to provide an explanation for why rates of entrepreneurship in the United States have varied over time in a way that overcomes the weaknesses of previous research conducted by scholars in the rates school. In this paper, entrepreneurship is operationally defined as the number of organizations per capita in the U.S. economy.(1) The paper presents and tests a Schumpeterian model for the variation in rates of entrepreneurship on data for the entire U.S. economy for the period 1899-1988. It then proceeds with a discussion of various control variables, which scholars in the field of entrepreneurship have found to influence entrepreneurship. Next, the model is tested on data measuring rates of entrepreneurship in the American economy since 1899, controlling for these variables. This is followed by a description of the test results and the conclusions.

The Schumpeterian Model

This paper argues that changes in rates of entrepreneurship in the U.S. economy follow a pattern explained by a Schumpeterian model of entrepreneurship. According to the Schumpeterian model (Schumpeter, 1947), entrepreneurship is essentially a function of the behavior of two groups of people: entrepreneurs, who create new resource combinations in response to exogenous technical change, and capitalists, who provide the resources for those new combinations. This model is shown in Figure 1.

According to Schumpeter (1939), the combination of factors used to produce a product or service is the production function. Entrepreneurship results from changes to this production function. Entrepreneurs see the inventions of scientists as creating opportunities for "new combinations" of factors of production (Schumpeter, 1939). That is, exogenous technological change is seen by entrepreneurs as providing the basis for a new good, a new method of production, the opening of a new market, the use of a new source of supply or the development of a new organizational form (Schumpeter, 1934).

To Schumpeter (1939), entrepreneurs, rather than managers in established firms, are the ones who see the opportunity for most of these new production functions. His reasoning is that new technological developments are often so different from the way in which existing technology works that they cannot be incorporated into the technological framework of existing firms. The conflict between the demands of the new technology and the production function of the established firm means that managers in established firms are less likely than entrepreneurs (who are not reliant on the existing production function) to accept the new technology (Schumpeter, 1939).

Since invention and technological change provide the opportunities for new combinations of factors of production, the rate of entrepreneurship in a society is, in part, a function of the rate of invention in that society. To Schumpeter, this rate of entrepreneurship is not constant over time because the underlying rate of invention that drives entrepreneurship is not constant.

There is a (discontinuous) flow of basic inventions related in an unspecified way to developments in science. These are largely exogenous to existing firm and market structures, and hence, to any measurable type of 'market demand' (Freeman, 1982, p. 40).

Therefore, rates of entrepreneurship are

. . . subject to explosive bursts as entrepreneurs realized the possibility of exceptional profits and growth arising from new combinations of technological and organizational change (Freeman, 1982, p. X).

A strong theoretical literature has now developed which argues the Schumpeterian view that technological change enhances new venture formation by creating new opportunities for combining resources in new ways to create new production functions (Aldrich, 1979, 1990; Astley, 1985; Wilken, 1979; Reynolds, 1991; Cross, 1981; Gould & Keeble, 1984; Tushman & Anderson, 1986; Brittain & Wholey, 1988; Mason, 1989; Romanelli, 1989; Dean, Meyer & DeCastro, 1993). In addition, empirical evidence has also been gathered which supports this view. Blau (1987) has shown that changes in technology are associated with changes in the rate of self-employment in the U.S.; Brittain and Freeman (1980) have shown that new organizational formation is influenced by technological change. These arguments lead to the first hypothesis:

H1: The relationship between the rate of technological change and the rate of entrepreneurship over time will be positive and significant.

If successful, the initial entrepreneur will earn entrepreneurial profits. This success will lead other entrepreneurs to follow, creating a wave of entrepreneurship in response to the invention or technological change. This occurs for three reasons. First, potential entrepreneurs will see the profits of the first entrepreneur and seek to imitate his or her production function to earn some of these entrepreneurial profits. Second, the success of the entrepreneur makes it easier to "do similar things in similar lines (Schumpeter, 1939, p. 100)," thereby enticing entrepreneurs to start related businesses. Third, "certain innovations . . . directly affect a wide variety of industries (Schumpeter, 1939, p. 100)." So, the success of the first entrepreneur will entice other entrepreneurs to start businesses based on the same technology in other industries.

Sociologists have provided empirical support for this imitative entrepreneurship pattern. They have found that low rates of entrepreneurship discourage subsequent organizational formation because of the lack of legitimacy for the businesses (Aldrich, 1990; Zelizer, 1978; Aldrich & Staber, 1988; Zucker, 1986), and the lack of training available in existing firms for prospective entrepreneurs to learn the entrepreneurship process (Stinchcombe, 1965; Hannan & Freeman, 1987; Hart & Denison, 1987; Aldrich, 1990; Romanelli, 1989). These arguments are consistent with the idea that when there are high rates of entrepreneurship in the economy, there are many firms for secondary entrepreneurs to imitate, thereby enhancing rates of entrepreneurship in subsequent periods.(2) This argument leads to the second hypothesis:

H2: The relationship between the rate of entrepreneurship at time (t-1) and the rate of entrepreneurship at time (t) will be positive and significant.

While it is the entrepreneur in the Schumpeterian model who creates new production function by establishing a new enterprise, the entrepreneur cannot establish that enterprise through his or her action alone. Rather, the entrepreneur requires the help of the capitalist.

The entrepreneurial act in the Schumpeterian model requires the entrepreneur to create new combinations of factors of production by bidding away resources from established producers (Schumpeter, 1934). This means that the entrepreneur must obtain funds to purchase these resources. To do this, the entrepreneur may use his or her own funds or the funds of others. In either case, funds are obtained from a capitalist. The entrepreneur must pay interest to the capitalist to obtain these funds since the capitalist must bear a risk for which he or she expects compensation (Schumpeter, 1934). If the entrepreneur is also the capitalist, the interest rate is considered the opportunity cost of the funds invested in the new venture. If the capitalist is an independent individual, the interest rate is considered the market interest rate for the funds.

The capitalist lends the money to the entrepreneur because he or she believes that the entrepreneur will use those funds to acquire resources to create a new production function that will yield a rate of return that exceeds the cost of those funds. There will always be potential entrepreneurs who cannot borrow money to create new production functions because the capitalists who lend money believe that the cost of funds to finance the venture will exceed the return from the creation of that production function.

This argument suggests that the number of entrepreneurs who receive financing will be a function of the cost of capital. At the margin, some potential entrepreneurs will receive financing when interest rates are low, but not when interest rates are high. There is now a fair amount of evidence for this argument (Cross, 1981; Pennings, 1982; Gartner, 1985; Bearse, 1982). Since the rate of interest varies over time, some potential entrepreneurs at the margin will receive financing from capitalists at certain times and not others, and the rate of entrepreneurship will be inversely proportional to the interest rate.(3) This argument leads to the third hypothesis:

H3: The relationship between interest rates and rates of entrepreneurship over time will be negative and significant.

Control Variables

A test of the Schumpeterian model of rates of entrepreneurship in the U.S. economy over time should not be conducted absent controls for other factors. Extensive research on entrepreneurship over the past thirty years has led to the accumulation of a large body of evidence for non-Schumpeterian factors that could bear upon rates of entrepreneurship. These variables need to be controlled to ensure that the results of a test of the Schumpeterian model are accurate. In the section below, these factors and the evidence for them are outlined.

The control variables were chosen using the following methodology. The research on entrepreneurship for the past thirty years was examined, and all the variables researchers have used to predict why people become entrepreneurs or why rates of entrepreneurship vary over time were catalogued. A search was then made for measures of these variables for the period 1899-1988. Those variables for which data were available were included in the study. While readers might criticize such an ad hoc approach to selecting control variables, this approach was necessary for two reasons. First, we do not yet have a theory of entrepreneurship which identifies all predictor and control variables. Second, time series data on variables that influence rates of entrepreneurship over time are limited over long time periods. In the absence of a theoretical basis for selecting control variables, and in the presence of data constraints, all possible control variables which could be found were included. As will be described later, the analysis was run with different combinations of control variables to test the robustness of the Schumpeterian model to the selection of controls.

Individual Level Variables

Psychological traits. Many psychologists argue that entrepreneurship can be explained by the possession of personality traits that differentiate people who start businesses from those who do not (Brockhaus, 1982; Brockhaus & Horwitz, 1986; Carsrud, Olm & Eddy 1986; Hornaday & Aboud, 1971; …