My colleague and I employ U.S. Census Bureau data from cities of 10,000 or more to examine the impact of immigrants in American cities on self-employment and median income. The results show that self-employment has a statistically significant and positive impact on median income and immigrant population. When controlling for race populations, lagged immigrant population has a negative impact on self-employment, but removing the Hispanic control causes this relationship to become statistically insignificant. In other words, Hispanics, not other ethnicities, drive much of the self-employment in U.S. cities. An implication is that more attention to helping Hispanic business owners succeed and expand their businesses could benefit the general population of a city.
For more, the full academic journal article is here: http://www.jrap-journal.org/pastvolumes/2010/v47/jrap_v47_n2_a1_carpenter_loveridge.pdf