Empirical tests of the 'unemployment push' hypothesis that unemployed workers start up new enterprises to a greater extent than employed workers to escape unemployment have produced contradictory results. This is at least partially a consequence of studies neglecting the industrial organization of regions and not incorporating adequate variables to represent market opportunities. Using data of US establishments in retail and service industries with low entry barriers, we find little or no evidence for the 'unemployment push' hypothesis. US regions with relatively high unemployment rates do not show relatively strong subsequent increases in the number of establishments across a broad range of industries. Possible exceptions are used merchandise stores and automotive repair shops.