In this paper, we develop a method to estimate markups using plant-level production data. Our approach relies on cost-minimizing producers and the existence of at least one variable input of production. The suggested empirical framework relies on the estimation of a production function and provides estimates of plant-level mark-ups without specifying how firms compete in the product market. We rely on our method to explore the relationship between markups and export behavior. We find that markups are estimated significantly higher when controlling for unobserved productivity; that exporters charge, on average, higher markups and that markups increase upon export entry. (JEL D22, D24, F14, L11, L60)

This paper uses representative panel data on 1,701 Bulgarian and 2,047 Romanian manufacturing firms to analyze how price-cost margins are affected by privatization and competitive pressure. Privatization is associated with higher price-cost margins. This effect is stronger in highly competitive sectors, which suggests that the creation of competitive markets and privatization go together. It also suggests that privatized firms reduce costs rather than increase prices, as in highly competitive markets firms are more likely pricetakers. Import penetration is associated with lower price-cost margins in sectors where product market concentration is high, but in more competitive sectors this effect is reversed.

In this paper we focus on a classic idea concerning span of control, which is that a prime driver is the scale of operations effect. We extend the theory concerning the scale of operations effect by allowing firms’ beliefs concerning a manager’s ability to evolve over the manager’s career. We empirically investigate the resulting testable predictions using a unique single-firm data set that contains detailed information concerning the reporting relationships at the firm. Our empirical analysis supports the notion that the scale of operations effect and learning are both important determinants of a firm’s span of control.