Marcinèin, A. and van Wijnbergen, S. (1997). The impact of Czech privatization methods on enterprise performance incorporating initial selection-bias correction Economics of Transition, 5(2):289--304.

Governments with transitional economies have applied different privatization methods, from an almost free distribution to the direct sale of state assets. While a free distribution was believed to ensure the political feasibility of the programme and its fairness, direct sales, (or more generally, standard privatization methods) had a significant advantage in creating concentrated ownership structures as the prerequisite to corporate control and restructuring. Many economists believe that the two goals of mass privatization, political feasibility and the creation of proper ownership incentives, contradict each other and recent empirical comparisons of enterprises seem to support this view. However, all empirical works have been based on the weak assumption that privatization methods were applied on a randomly selected sample of enterprises, which then allowed for a direct comparison between these enterprises. Our main claim is that governments actually selected enterprises non-randomly and therefore, the resulting selection-bias must be incorporated into the analyses. To show this, we apply a Heckman two-step regression method on a sample of 559 Czech enterprises. The main point of this paper then is that performance is influenced by the selection process and that the combination of vouchers with outsider owners is preferred over 100 per cent voucherization.