THE EXTENT OF OVERESTIMATION OF SMALL FIRM JOB CREATION - AN EMPIRICAL-EXAMINATION OF THE REGRESSION BIAS

Davis, Haltiwanger and Schuh (1993; 1996a; 1996b) suggested that the b elief that small firms are major contributors of new jobs is largely b ased on methodological flaws. In particular, their reasoning about the ''regression fallacy'', i.e., that temporary fluctuations in size sys tematically biases estimates in favor of small firm job creation, has caught on interest among researchers and concern among policy makers. In this article we attempt to estimate empirically the extent of overe stimation of small firm job creation due to the ''regression fallacy'' . It is concluded that the effect is very small and that correcting fo r it does not lead to qualitative change of the results. There may be good reasons to question to what extent small firms can lead economic development, and whether it is good or bad if they do create most new jobs, but concern for the ''regression fallacy'' does not seem to be a n important issue in this context.