ارزیابی تجربی از اثرات رفاهی دامپینگ متقابل

Can two-way trade in similar products lead to lower welfare than if such trade was banned? Theory answers yes. To empirically investigate this proposition we examine Swedish imports of bottled water. Assuming one-shot (Bertrand and Cournot) competition, we can use the estimates from a structural model of demand to uncover marginal costs. We simulate the effect on consumer and producer surplus of banning imports. We do not find convincing evidence that banning imports would increase overall welfare. Given our choice of market this suggests we should not be overly concerned with the welfare effects of two-way trade in consumer goods that are close to homogenous.

More than 60% of the non-sparkling mineral water sold in Swedish stores is imported. Sweden also exports mineral water; in an average month in 2001 Sweden exported 791 metric tons of bottled mineral water and imported 1305 metric tons.2 Does this shipping of water in both directions represent a good use of resources? The theory of trade under imperfect competition that has developed from the 1980s onward (see for instance Helpman and Krugman, 1985) demonstrates that trade may have adverse consequences on welfare under certain conditions. In particular, we know that when there is two-way trade in a homogenous good (“reciprocal dumping”) and important barriers to entry, the waste of transporting identical goods in opposite directions can dominate the positive, pro-competitive effect of trade. This was first shown in Brander (1981), and elaborated on in Brander and Krugman (1983). Is this result, that trade can hurt overall welfare, empirically relevant? Given the current debate about the merits of trade, and that a large share of world trade is generated by countries exporting and importing quite similar goods, this comes high on our “things we would like to know”-list. Particularly so since Feenstra et al. (2001) find that the global trade patterns for homogenous goods are consistent with a reciprocal dumping-type model with barriers to entry. Theory is clear that trade is typically welfare enhancing—even the staunchest critic of international trade is likely to agree that some trade is good. The question is: do we know of any cases when trade is bad? We can't know unless we look and we therefore wanted to choose an industry so as to stack the cards against the free trade case; bottled water suits that bill well—there is arguably little real product differentiation, low technological economies of scale, transport costs are high relative to the value of the product and Sweden is also an exporter of the same good.3 A highly concentrated industry structure suggests that there are considerable barriers to entry. Furthermore, a close substitute, tap water, is available at a price close to zero. This should limit the pro-competitive potential for imports. If trade does not reduce welfare even in this setting the pro-trade case should be strengthened. If we find that trade is indeed detrimental to welfare in this market it then becomes important to study less extreme cases. We use a structural model of competition on the Swedish market for bottled water to examine the welfare effects of bottled water imports. We make use of a detailed data set that includes monthly observations of prices and quantities of all brands of bottled water sold in Swedish stores during 1998–2001. To our knowledge no other paper has attempted a similar exercise—most of the empirical literature that examines trade under imperfect competition has focused on measuring intraindustry trade and examining how well such trade theory can explain actual trade patterns (see Hummels and Levinsohn, 1995) or simply on testing if there is a pro-competitive effect of lower trade barriers (see Levinsohn, 1993 or Tybout, 2003). A somewhat closer precursor is Berry et al. (1999) who examine the welfare impact of voluntary export restrictions (VER) on Japanese car exports to the US.4 Another paper who applies similar methods to international oligopoly is Irwin and Pavcnik (2004) who examine trade disputes between Airbus and Boeing. Even though one could argue that mineral water is a homogenous product, different prices of different brands demonstrate that brands are imperfect substitutes. Taking this seriously, we follow seminal work by Berry (1994) and Berry et al. (1995) and estimate a model of demand for differentiated products. We model demand as being a function of product characteristics, using a multinomial nested logit (MNL) specification. Similar MNL formulations are also applied by for instance Verboven (1996) on European car markets, by Slade (2003) on the UK beer market and Ivaldi and Verboven (in press) on European truck markets. The next section summarizes theoretical results on the welfare effects of trade that are relevant for our investigation. In Section 3 we describe the Swedish bottled water market, with particular attention to how it matches the assumptions of the Brander-Krugman model. In Section 4 we estimate demand. We proceed to calculate the implied markups in Section 5 and provide a counterfactual experiment of calculating consumer surplus, and global producer surplus, under the assumption that only domestically produced waters can be sold on the Swedish market. Section 7 concludes.

This paper has investigated whether transport cost losses from trade can outweigh the partial equilibrium gains from trade (stronger competition and more brands to choose from). Welfare on the Swedish market for bottled water is perhaps of limited interest in itself. It is a small market (on average Swedes spends less than the equivalent of 20 US dollars per year on this market) and the share of imports in the overall market is low. Rather, we would argue that the case is of interest because we attempted to select an industry where we would be as likely as anywhere to find a negative effect of trade. We thus attempted to evaluate the empirical relevance of the proposition that trade can lower welfare through wasteful transportation. The results show that even in this case we do not find convincing evidence against trade. Our results from this market therefore suggest that we should not be overly concerned with two-way trade in consumer goods that are close to homogenous. Clearly, as with any model based on simulations the work should perhaps more be seen as a reality based calculation rather than a prediction of what would actually happen if Sweden decided to ban imports. For instance it is likely that we would see some domestic entry into the still water market if the experiment had been undertaken in reality. In response to a banning of imports one could hypothesize that one of the large domestic brands of sparkling would launch a still version of their brand. Rather than hypothesize about such potential entry we note that free entry would make the pro-trade case even stronger. Thus our main conclusion would not be overturned by free entry. At a deeper level one may wonder about the welfare effects of marketing in this industry—are consumers' tastes for different brands just an artifact of marketing? This is clearly a large and interesting issue in its own right and one that we have abstracted from (see Bagwell, 2003 for an overview of the welfare effects of advertising). One conclusion that is not sensitive to our modeling is the fact that in terms of distance traveled per liter the difference between domestic production and imports is fairly small. One should thus exercise caution before equating international trade with more wasteful transportation than domestic trade. Lastly, one could argue that the drinking of bottled water uses up a lot of resources for little apparent benefit in a country where the tap water is of high quality. From an environmental perspective it is sure to be less important whether bottled water is imported than if bottled water rather than tap water is consumed.