these countries would have occurred at the start of the transition process. In contrast, the less developed countries, which are normally also slow reformers, postponed many reforms that drive enterprise restructuring; so one would expect that the peak of restructuring activity in less advanced countries occurs later than in more advanced countries. A possible explanation for this observed pattern in the BEEPS data is as follows. Most of the firms in the survey are SMEs from the new private sector. The pattern in the extent of deep restructuring in the BEEPS data is therefore largely driven by the extent of this activity among these firms. The pattern we observe suggests that this activity peaked in new private sector SMEs about 10-12 years after the start of transition. This could be either because of a standard pattern in new firm development, since the start of transition and the beginning of new private sector growth was synchronized across TEs, or because of the largely common timing of the transformational recession and the resumption of growth in TEs.
We turn now to regression analysis, and explaining the relationship between observed restructuring and the economic environment in particular. The early waves of BEEPS (1999 and 2002) are not very well suited to investigate the effect of competition in particular. A fundamental problem is that most measures of competition in the survey are contemporaneous, while restructuring refers to the period spanning three years before the survey. This is true of the market structure variables (number of competitors), price elasticity of demand, competition from imports, and indicators of pressure from competitors and customers for developing new products. In other words, competition is measured after restructuring activities have been undertaken, so it is difficult to establish a causal link from competition to restructuring (unless market structure is more or less stable). One may proceed under the assumption that market structure does not change much over three year periods, but the assumption is questionable at least. Attempts to estimate simple regressions with restructuring measures on the LHS and ownership and competition variables on the right-hand side (plus controls such as firm size, industry, location) show that it is very difficult to establish any robust result with respect to competition. Fortunately, the latest wave of BEEPS (2005) suffers less from this problem, because firms were also asked retrospective questions about number of