What determines the order in which firms are privatized? Do governments strategically sequence privatization? An answer to this question is important for understanding the behavior of governments and firms in transition economies and for determining whether empirical studies of the effects of privatization need to take into account the potential selection bias brought about by strategic sequencing. This issue is also relevant for countries such as China, India, and Mexico, which have large state sectors and are currently pursuing privatization.

This is the first study that examines both theoretically and empirically how competing government objectives may give rise to different privatization outcomes. To obtain testable predictions about which factors affect the sequencing of privatization, we develop new—and adapt existing—theoretical models of sequencing strategies for the following government objectives: maximizing sales revenue and public goodwill, increasing economic efficiency; and reducing the political costs associated with layoffs.

Regarding the maximization of privatization revenues, we show that it is a reasonable strategy for a government pursuing this objective to privatize more profitable firms first. The same outcome will arise if the government objective is to generate public good will from free or subsidized transfers of firms to citizens. Regarding efficiency, one strategy is to privatize inefficient firms first, so as to induce major restructuring and improvement in enterprise performance. Assuming that private firms are more efficient in responding to information, the government may also want to privatize firms in industries that face the greatest uncertainty in terms of demand and cost shocks. Privatizing more profitable firms may also reduce the political cost associated with layoffs, as our model shows that such firms are likely to have fewer layoffs than less profitable firms.

The Czech government carried out one of the most extensive mass privatization programs. To determine which objectives the government was pursuing, we examined firm-level data. Those data provide strong evidence that the Czech government privatized more profitable firms first. This outcome is consistent with the hypotheses that the government sought to maximize privatization revenues and public goodwill and minimize the political costs of unemployment. The fact that labor market conditions were not an important determinant of privatization allows us to rule out the hypothesis that reducing the political costs of unemployment was an important priority for the Czech government (not a surprising result in view of the low unemployment rate). We also find that the privatization process was consistent with the hypothesis that firms likely to be more responsive to changes in demand conditions were privatized first. However, inefficient firms are not being privatized first, suggesting that the government did not pursue the objective of productive efficiency. These conclusions provide insight into privatization strategies that may be adopted when different, sometimes competing, objectives are pursued by the government.

Our results have important implications for studies evaluating the effect of privatization on firm performance. Several such studies measure gains from privatization by comparing the performance of privatized firms to firms remaining in the public sector. However, such comparisons are valid only if firms are randomly chosen for privatization. If the government selectively privatizes more profitable firms (as our results suggest), it should not be surprising to see these firms perform better than firms that remain public, even if privatization has no effect on firm performance. A similar statistical problem arises in studies examining the effect on firm performance of the length of time since privatization. Our result that more profitable firms are likely to be privatized early implies that unobserved firm characteristics that make firms more profitable may be correlated with the length of time the firm has been privatized.