Exporting markups Joakim Gullstrand Karin Olofsdotter Department of Economics, Lund University
Background and motivation Some recent studies explain variations in firms export prices by recognizing exporting firm s ability to set prices and price discriminate across markets. Empirical evidence of the importance of price discriminating behavior. Fabiani et al (2005) suggests that more than 80 percent of firms in the Euro area apply price-discriminating strategies. The law of one price rejected in a large number of studies suggesting arbitrage cost and segmented markets. Gullstrand, Olofsdotter and Thede (2011) show that firms with large variations in prices across export markets have higher markups. Simanovska (2011) suggests that firms set higher prices and have higher markups on richer markets.
Background and motivation Price discrimination and segmented markets received fairly little attention in the international-trade literature. Lately, variations in markups have been introduced in heterogeneous-firm settings. Melitz and Ottaviano (2008) Model where markups vary across firms and export destinations. Firms with lower cost levels charge lower prices and have higher markups. Firms will have lower markups and prices on markets characterized by higher competition. Bernard, Eaton, Jensen and Kortum (2003) Firm s markup and price will be higher on markets where it can exert more market power.
Background and motivation (cont.) Empirically, De Loecker and Warzynski (2012) find that exporters have higher markups than non-exporters. Gullstrand, Olofsdotter and Thede (2011) focus on the heterogeneity among exporters price-setting behavior and variations in markups. In this paper, we investigate variations in markups across exporters operating on different foreign markets. In particular, we estimate firm-product-destination specific markups and explore how these correlate with other characteristics of the export markets.
Estimating markups The point of departure is the firm s optimal price setting strategy leading to a markup definition of price over marginal cost: mmmmmm = μ ii P ii c ii = 1 + s iiθ ii η ii 1 where P ii : price of firm i s product at time t c ii : marginal cost s ii : firm s share of the market θ ii : conjectural-variation variable η ii : market elasticity of demand
Estimating markups (cont.) To identify markups, we consider a general model consistent with an imperfectly competitive market structure. Approach based on Hall (1988) and relies on the insight that cost shares of factors of production are equal to their revenue shares only if markets are characterized by perfect competition. Imperfect competition, on the other hand, drives a wedge between the cost and revenue shares, as measured by the markup.
Estimating markups (cont.) Follow the methodology by De Loecker and Warzynski (2012). In particular, they derive the following expression for the markup: μ ii = θ ii X α ii X where θ ii X : output elasticity of the variable input X ii α ii X : the share of expenditures on input X ii in total sales While expenditure shares are directly obtained from the data, output elasticities have to be estimated. estimate a flexible production function controlling for unobserved productivity shocks building on the works of Olley and Pakes (1996) and Levinsohn and Petrin (2003).
Estimating markups (cont.) We estimate a value-added production function with high- and low-skilled labor together with fixed capital where material input is used as a proxy for productivity shocks. Export sales and input shares are divided across the export destinations by the volume in exports to the different countryproduct combinations. Makes it possible to estimate production functions and output elasticities for each firm-product-market combination. The estimated output elasticities, θ, are used to compute the preferred markups: L μ iiii = θ L iiii L α L iiii 1
Data Firm-level data for exporting firms in the Swedish food processing sector for the period 1997-2006 provided by Statistics Sweden. The food processing industry the fourth largest manufacturing industry in Sweden. Export behavior of firms in this industry is found to resemble the behavior of firms in other countries and industry settings (see, e.g., Greenaway et al., 2010). Data includes firms export volumes and quantities by products (defined by the CN at the 8-digit level) to each country in the world. Country information from CEPII gravity dataset. Unbalanced panel with 6503 firm-product-destination-year observations.
Table 1. Descriptive figures of exporting firms in the Swedish food processing industry Variable Mean (Std. Dev.) Definition Source Markup 1.15 (2.72) Defined as in methodology Own calculation. section TFP 1.33 (0.40) As in Aw et al (2003) Own calculation. Export volumes 1.27e+7 (9.95e+7) Exports volumes of each Statistics Sweden product to each destination Share of high-skilled 0.17 (0.08) Share of total workforce with Statistics Sweden workers at least 3 years tertiary education Total number of 16.98 (15.01) Firms total number of foreign Statistics Sweden export destinations market penetrations Total number of 29.13 (35.52) Firm s total number of Statistics Sweden products exported exported products Foreign-owned firm 0.33 (0.47) Firms owned by foreign firms Statistics Sweden by at least 50 % Own foreign firms 0.52 (50) Firms that owns at least 50 % a foreign firm. Statistics Sweden
Table 2. Results from log-linear specifications with estimated markups as dependent variable All goods GDP of destination -0.007 (.24) GDP per capita of destination 0.06 (.00) Distance to destination 0.035 (.07) TFP 0.578 (.00) Export volume (product and destination) 0.153 (.00) Share of high skilled workers 0.077 (.00) Total number of export destinations -0.050 (.00) Total number of products exported 0.054 (.00) Foreign-owned firm -0.028 (.03) Own foreign firms 0.058 (.00) R 2 adjusted 0.47 n of observations 6503 Specification: LS with dummies for 10 industry belongings, 5 product types, 21 location dummies, 14 regional export dummies, 9 year dummies
Table 2. (cont.) All goods Consumption goods Intermediate goods GDP of destination 0.711 (.34) 0.491 (.53) 0.491 (.53) GDP per capita of -0.774 (.31) -0.540 (.49) -0.168 (.32) destination TFP 0.293 (.00) 0.287 (.00) 0.319 (.01) Export volume (product 0.151 (.00) 0.139 (.00) 0.205 (.00) and destination) Share of high skilled 0.170 (.00) 0.165 (.00) 0.124 (.19) workers Total number of export destinations -0.004 (.87) -0.002 (.97) 0.013 (.87) Total number of products exported 0.037 (.00) 0.038 (.00) 0.027 (.61) R 2 (within) 0.20 0.22 0.18 n of observations 6503 5262 1234 Specification: Firm-product-destination fixed effects, year dummies