The increasing integration of the global economy over the past few decades has strengthened research and policy interest in understanding how exporters set their prices for internationally traded goods. Of particular interest is whether exporting firms have market power and are thus able to price discriminate across destinations by varying their markups. As the discriminatory pricing behavior of firms generates deviations from the Law of One Price, it may indeed help explain why international markets are not perfectly integrated.
Recent empirical work demonstrates that firm-level markups are variable. For instance, they respond to trade liberalisation (De Loecker et al. 2016), to exchange rate fluctuations (Berman et al. 2012, Chen and Juvenal 2016), and they vary with per capita income (Simonovska 2015). But surprisingly there is no evidence on how the markups of exporting firms vary across export markets depending on trade costs such as bilateral distance or ad valorem tariffs. Nor is there any evidence on how product quality shapes the response of markups to changes in trade costs.
Markups, trade costs, and quality heterogeneity
In new research (Chen and Juvenal 2019), we explore theoretically and empirically how exporters adjust their markups across destinations depending on distance, tariffs, and the quality of their exports. In our theoretical framework, we extend the monopolistic competition model of Martin (2012) where trade costs are assumed to be ad valorem (i.e. applied as a percentage of the producer price per unit traded) but also per unit (i.e. defined as a constant cost per unit traded). Allowing for per unit trade costs enables us to generate variable markups that depend on trade costs and quality (Crozet et al. 2012, Irarrazabal et al. 2015).
Our model shows that for a given quality, markups increase with per unit trade costs such as bilateral distance. Instead, markups fall with ad valorem trade costs such as tariffs. Moreover, the effects of trade costs (i.e. distance and tariffs) on markups are heterogeneous and smaller in magnitude for higher quality exports.
Intuitively, these predictions arise because the elasticity of demand with respect to the export price varies with trade costs and quality. On the one hand, the demand elasticity falls with bilateral distance, especially for lower quality exports. As the demand in more distant markets becomes less elastic to changes in the export price, exporters find it profitable to raise their prices (by raising their markups) to compensate for the lower demand they face due to higher transportation costs. But they raise their markups less for higher quality exports. On the other hand, the elasticity of demand increases with ad valorem tariffs, especially for lower quality exports. Firms therefore lower their markups in countries with higher tariffs, but they reduce them less for higher quality exports.
Markups, trade costs, and quality heterogeneity in the data
We use a very rich data set of firm-level exports of Argentinean wines between 2002 and 2009. For each export transaction we observe the name of the exporting firm, the country of destination, the date of shipment, the Free on Board value (in US dollars) and the volume (in liters) of each wine exported. The crucial feature of our trade data set is its level of disaggregation. For each wine exported, we observe its name, type (red, white, or rosé), grape, and vintage year. In other words, wine exports are reported at the individual product level.
We divide the value by the volume exported at the firm-wine-destination-quarter level to compute export unit values as a proxy for export prices. Given the level of disaggregation of the data, unit values can plausibly be interpreted as prices.
The second advantage of our data set is that we observe an external measure of quality. We rely on two well-known experts wine ratings, the Wine Spectator and Robert Parker, to measure the quality of each wine at the name-grape-type-vintage year level. In both cases the quality ratings are reported on a (50,100) scale with a larger value indicating a higher quality. Once we merge the trade data with the quality ratings of the Wine Spectator which has the largest coverage of Argentinean wines, we observe 237 multi-product wine producers shipping 8,361 different wines with heterogeneous levels of quality. As we only include wine producers in the sample, each wine is exported by one firm only.
As in any study of price discrimination that relies on price data, the challenge is to disentangle the variation in markups from the variation in marginal costs. Thanks to the granularity of our data, we can identify the variation in markups by comparing the unit values of a given product exported by a given firm at a given point in time across destinations. Since product-specific marginal costs do not vary across destinations, the variation in prices across markets captures the variation in markups.
Figure 1 Destination-specific mean markups
a) Bilateral distance
Notes: This figure presents destination-specific mean markups against (a) (log) bilateral distance and (b) (log of one plus) mean tariffs. The mean markups are measured by the country fixed effects obtained from regressing (log) unit values on product-quarter and destination country dummy variables.
Figure 1 plots the mean markup charged by Argentinean wine producers in each destination country against trade costs: (a) the bilateral distance between Argentina and each destination country and (b) the mean tariff imposed by each country on Argentinean wine exports. Although these figures do not account for the effects of other country-level characteristics (such as GDP or GDP per capita) in explaining markups, they show that markups tend to be higher in more distant markets, and (slightly) lower in countries with higher tariffs. For instance, markups are highest for Luxembourg, which is a distant country, while they are low for Uruguay, which is close to Argentina. Conversely, markups are low in high-tariff countries like Jordan, but relatively high in low-tariff countries such as Canada or Japan.
In a first step, we estimate the direct effects of distance and tariffs on the variation in markups across markets. Consistent with the patterns in Figure 1, we find that the elasticities of markups with respect to distance and tariffs are equal to 0.021 and to -0.086, respectively. On average, a doubling of distance therefore increases markups by 1.47%, while a doubling of tariffs (from their sample mean) reduces markups by 1.08%.
In a second step, we allow for variable trade cost elasticities across quality levels and we find a great deal of heterogeneity. For instance, at the 5th percentile of the quality distribution (i.e. for a low-quality wine), the magnitude of the distance and tariff elasticities is large at 0.052 and -0.227, respectively. In contrast, at the 95th percentile (i.e. for a high-quality wine), the two elasticities are both close to zero.
Figure 2 illustrates the main results. The markups of lower quality exports respond strongly to changes in bilateral distance and tariffs (i.e. the magnitude of the elasticities is large). Instead, the markups of higher quality exports vary little or not at all with changes in trade costs (i.e. the elasticities are small in magnitude or even insignificant).
Figure 2 Trade cost elasticities and product quality
a) Bilateral distance
Notes: This figure presents (a) Bilateral distance and (b) tariff elasticities by quality level (95 percent confidence intervals reported as dashed lines).
We also find that the heterogeneous effects of trade costs on markups are stronger for exports to richer destinations where consumers generally have a stronger preference for higher quality goods. They are also stronger for higher quality firms, larger firms, and exporters who own a large share of the export market. As high-performance firms tend to charge higher markups, they are indeed better able to adjust their markups across countries and quality levels in response to changes in trade costs.
Our results are important for several reasons. First, they provide strong evidence that the variation in firm-level export prices across markets can be explained by markup variation conditional on quality. Due to market power, firms price discriminate across destinations, but they also price discriminate more aggressively for lower quality exports. Second, as the markup of a given product with a given quality varies across export markets depending on distance and tariffs, our results confirm that trade costs also play a key role in generating deviations from the Law of One Price, and they thus matter in explaining the degree of international market segmentation. Lastly, we expect our findings to matter in explaining aggregate export prices and markups. On the one hand, we show that our results continue to hold more generally for Argentinean manufacturing industries other than wine. On the other hand, we find that the heterogeneous effects of trade costs on prices and markups are mainly attributable to the high-performance firms that contribute to the bulk of aggregate exports.
Berman, N, P Martin and T Mayer (2012), “How do different exporters react to exchange rate changes?”, Quarterly Journal of Economics 127(1): 437-492.
Chen, N and L Juvenal (2016), “Quality, trade, and exchange rate pass-through”, Journal of International Economics 100: 61-80.
Chen, N and L Juvenal (2019), “Markups, quality, and trade costs”, CEPR Discussion Paper 14204.
Crozet, M, K Head and T Mayer (2012), “Quality sorting and trade: firm-level evidence for French wine”, Review of Economic Studies 79(2): 609-644.
De Loecker, J, P K Goldberg, A K Khandelwal and N Pavcnik (2016), “Prices, markups, and trade reform”, Econometrica 84(2): 445-510.
Irarrazabal, A, A Moxnes and L D Opromolla (2015), “The tip of the iceberg: a quantitative framework for estimating trade costs”, Review of Economics and Statistics 97(4): 777-792.
Martin, J (2012), “Markups, quality, and transport costs”, European Economic Review 56(4): 777-791.
Simonovska, I (2015), “Income differences and prices of tradables: insights from an online retailer”, Review of Economic Studies 82(4): 1612-1656.