We decompose the employment effects of a credit guarantee scheme to small firms during a crisis and a recovery period. We explore firm-level variation in access to the program to show that net employment increases, but only during crises. Using matched employer-employee data, we are able to decompose this effect into hires and exits. We find that firms eligible for the program hire more, but also have more exits. Hired workers come mostly from unemployment and exiting workers do not immediately find a new job, showing that there is no direct reallocation of workers across firms. Newly hired and exiting workers do not differ on most observable characteristics, suggesting that immediate adjustments are more related to quantity rather than quality. Government support to small firms during crises does not seem to lead to job retention of "zombie" workers.