Entrepreneurship in Low-Income Areas

  Full Report                              Research Summary

Entrepreneurship in Low-Income Areas

By Maurice Kugler, Marios Michaelides, Neha Nanda, and Cassandra Agbayani;
Columbia, Maryland  21044; 50 pages. Under contract number SBAHQ-15-M-0150.

September 2017   No. 437

This study describes entrepreneurship in low-income areas. It focuses on three things: the characteristics of self-employed workers in low-income areas, the income sources of these workers, and the characteristics of businesses operating in low-income areas.  The report was prepared for the Office of Advocacy by IMPAQ International.

Overall Findings

Low-income areas have fewer businesses than other areas.  Low-income areas are home to 2 out of every 9 workers, but only 2 out of every 11 self-employed workers, and only 2 out of every 30 businesses with employees

Businesses in low-income areas have fewer employees.  Among businesses with employees, businesses in low-income areas have an average of 12 employees while businesses in other areas have an average of 15 employees.

Low-income areas were identified using data from the U.S. Census Bureau’s American Community Survey. The survey divided the United States into 2,351 geographic areas. Average household income was calculated for each area, and one-fifth of the areas, those with the lowest average incomes, were classified as low-income.

Average household income in the low-income areas was $46,432, while average household income in the other areas was $80,281. The poverty rate in low-income areas was 22.2 percent, while the poverty rate in other areas was 11.1 percent. Of the 471 low-income areas, 249 were rural and 222 were urban. Of the 1,880 other areas, 1,058 were rural and 822 were urban.

Figure 1 shows the geographic distribution of low-income areas. Similar results were obtained when low-income areas were identified using poverty rates rather than average incomes.

The report shows that the self-employed in low-income areas tended to be younger and less educated than the self-employed in other areas. They were also less likely to be married and less likely to speak English. While 9.2 percent of the self-employed in low-income areas were black, only 3.8 percent of the self-employed in other areas were black. While 16.5 percent of the self-employed in low-income areas were Hispanic, only 8.8 percent of the self-employed in other areas were Hispanic. Only 2.3 percent of the self-employed in low-income areas were Asian, while 5.2 percent of the self-employed in other areas were Asian. These differences generally reflect differences in the characteristics of workers in those areas.

Self-employment was lower in low-income areas. Only 9.2 percent of workers in low-income areas were self-employed, while 10.9 percent of workers in other areas were self-employed. The pattern held within many demographic groups, including men, women, whites, blacks, most age groups, and all levels of education. No significant differences were observed for Asians, Hispanics, or the youngest age groups.

There were fewer businesses in low-income areas compared to other areas. Only 6.7 percent of businesses with employees were located in low-income areas, and they tended to be smaller, with an average of 12.0 employees, versus 15.1 employees in other areas. The businesses in low-income areas had an average payroll per employee of $33,071, while those in other areas had an average payroll per employee of $48,382. Similar patterns held within nearly every industry classification.

The self-employed who had incorporated their businesses tended to have higher incomes than those who had not. Only 30.6 percent of the self-employed in low-income areas had incorporated, while 38.9 percent of the self-employed in other areas had incorporated.

Figure 2 shows the relationship between worker type and income in low-income areas. Incomes in the previous 12 months were highest for the self-employed who had incorporated, at $68,605. Salary workers received $40,110, slightly more than the self-employed who had not incorporated, at $37,468. The unemployed received only $13,354. The incomes include both labor income and income from other sources, like investments and Social Security, but do not include the value of business equity or most benefits. Similar patterns were observed in a multiple regression analyses that accounted for differences in personal characteristics.

Data, Methods, and Limitations

The report relies on data from two sources, both of which are available to the general public. Data on personal and household characteristics, including self-employment, were from the 2013 American Community Survey (ACS), a nationally representative survey conducted by the U.S. Census Bureau. Data on business characteristics were from the 2013 County Business Patterns (CBP), a data series generated by the U.S. Census Bureau using multiple data sources, including surveys and administrative data.

The areas defined as low-income vary slightly between parts of the analysis. The smallest geographic unit available for the ACS is the Public Use Microdata Area (PUMA), and, for parts of the analysis focusing on individual characteristics, low-income areas are defined by classifying some PUMAs as low-income. The CBP provides information at the county level, and, for parts of the analysis focusing on business characteristics, low-income areas are defined by classifying some counties as low-income. While the areas defined as low-income vary, the geographic distributions of those areas are similar.

Entrepreneurs and entrepreneurship are primarily characterized in the report using descriptive statistics, such as the proportions of workers in various groups who were self-employed. In some cases, multiple regression analysis was used to explore the robustness of the patterns observed in the descriptive statistics. Some caution is warranted in interpreting the findings. Neither statistically significant differences in proportions nor statistically significant regression coefficients imply causal relationships between characteristics and outcomes.

For example, the report shows that only 9.2 percent of workers in low-income areas were self-employed, while 10.9 percent of workers in other areas were self-employed. Residence in low-income areas may have reduced self-employment through limiting the discretionary income of potential customers. However, residence in low-income areas may have increased self-employment by limiting the attractiveness of other types of employment. Residence in low-income areas may also be related to unobserved characteristics that affect both self-employment and area of residence, such as family background. Lower rates of self-employment in low-income areas do not imply that residence in low-income areas causes lower self-employment.

Consistent with the Office of Advocacy’s data quality guidelines, this report was peer reviewed. More information on the process can be obtained by contacting the director of economic research by email at advocacy@sba.gov or by phone at (202) 205-6533.

Discussion

Entrepreneurship can benefit local economies. However, rates of entrepreneurship vary across areas and demographic groups. Areas with low incomes tend to have fewer and smaller businesses, and demographic groups with lower self-employment tend to be concentrated in areas with low incomes.

The results presented in the report indicate that differences in self-employment between the areas defined in the report as low-income and other areas do not contribute significantly to differences between demographic groups. For example, about 4.9 percent of black workers and about 11.5 percent of white workers are self-employed. Black workers are concentrated in low-income areas, with about 33 percent of black workers living in low-income areas and only about 16 percent of white workers living in those areas.  About 4.6 percent of black workers in low-income areas are self-employed, and about 5.1 percent of black workers in other areas are self-employed. However, if black workers in low-income areas were self-employed at the same rate as black workers in other areas, the percentage of all black workers self-employed would rise by only 0.2 percentage points, and it would still be much lower than the percentage of all white workers self-employed.

In contrast, differences in self-employment across demographic groups contribute significantly to differences in self-employment between low-income and other areas. For example, if black workers in low-income areas were self-employed at the same rate as white workers in those areas, the percentage of workers in low-income areas who were self-employed would be 10.3 rather than 9.2. If black workers in other areas were self-employed at the same rate as white workers in other areas, the percentage of workers in other areas who were self-employed would be 11.4 rather than 10.9. The difference between low-income and other areas would be 0.5 percentage points smaller, so the difference in self-employment between black workers and white workers can account for over 30 percent of the difference in self-employment between low-income and other areas.

Programs that promote entrepreneurship may benefit low-income areas. Additional research that could aid in the development of effective programs would address questions like:
 

  • What is the effect of entrepreneurship on the local economy?
     
  • What are the barriers to entrepreneurship in low-income areas?
     
  • What programs are effective at encouraging entrepreneurship?

Generating convincing answers to those questions may entail a different type of analysis from that presented in the report. The most convincing answers to questions about causal relationships are generated using data in which some individuals or groups have been randomly assigned to the conditions of interest. Random assignment solves the problems with causal inference discussed in the preceding section by limiting the potential for correlations between the conditions of interest and personal characteristics. Some studies using such methods are described in the report.