|This thesis consists of four essays within the field of international trade economics, as well as an introduction chapter. All four papers are empirical studies of trade flows in food products using transaction data. The first chapter introduces the data used in the papers, and provides a brief review of the general literature. During the last decades, the focus in international economics has shifted from studies of trade flows using aggregated data to studies of firm behavior. Historically, the starting point for the analysis of trade flows has been the gravity model of international trade. All four papers in this thesis use a set of regressors commonly used in this literature. The papers aims to go beyond the existing literature in dealing with transaction-level data for trade flows of highly perishable food products. Three of the papers relate to export of salmon, while one of the papers investigates the import of apples. All transaction data is based on customs declarations at the firm level. The first paper studies the effect of trade costs on the export of Norwegian salmon. Trade of salmon is a rapidly growing industry, and fresh salmon is a highly perishable product. Trade in such products tend to be highly vulnerable to trade costs, e.g. in the form of transportation and transaction costs. The paper studies the evolvement of export growth within the industry for the period 2003-2009. In total, 483,956 individual transactions are studied. Two different versions of the gravity model of trade are estimated to study trade growth. In addition, the paper studies how different trade costs affect the extensive and intensive margins of trade. The extensive margin is defined as the number of exporters, while the shipment (transaction) frequency is used as a measure to capture the intensive margin of trade. A Poisson model is used to estimate the extensive margin, while both a Poisson model and a Negative Binomial model are used to estimate the intensive margin. I find a significant negative effect from transportation costs on trade values. Transportation costs are measured, both as geographical distance from Norway to the destination market, and as the internal size of the destination market. In addition, I find that shipments towards densely populated areas, large markets, and shipments with air transport are positively related to trade values. Transportation costs are shown to choke off both the extensive and intensive margins of trade. Trade to markets within the EU negatively affect the extensive margin, while the opposite effect is found for the intensive margin. This result could indicate that it is the largest exporters that are dominant in the European markets. Further, the results indicate that much of the negative effect from distance on trade is an aggregation effect. The second paper investigates the duration of trade relationships, and hit-and-run behavior in Norwegian salmon export. In the literature on international trade, much has been said about why firms start to export, less has been said about which factors may induce termination of trade relationships. In this paper, we investigate trade duration by two different approaches. First, we use a Cox model to estimate hazard rates to study the probability for termination of trade relationships. Second, we estimate the probability for a firm to choose different lengths of the trade relationships by using a multinomial logit model. In the first approach, trade duration is calculated by the number of subsequent years a trade is observed between the trading partners. In the latter approach, we categorize trade duration by the number of transactions. We define a hit-and-run strategy as a trade relationship that is only observed with one single transaction. Our findings reveal a large presence of short-lived trade relationships, and that estimated survival rates are heavily dependent on the level of aggregation. We show that trade relationships are shorter in large markets served by many firms. Hence, keen competition seems to be a substitute to deeper relationships. Hit-and-run strategies are characterized by large initial trading volumes, and by large transportation costs between the trading partners. The third paper studies the choice of invoicing currency for Norwegian salmon exporters. In today’s seafood markets, salmon is the species with the most varied transaction modes. Unilateral contracts with different specification, standardized future contracts, and a number of other transaction modes, are used in addition to traditional spot transactions. The exporters’ choice of invoicing currency affects which part in the trade relationship that takes on exchange rate risk, and can thus be an important factor for an exporter’s competitiveness. The paper discusses the empirical patterns of use of different invoicing currencies observed in the data, and uses a multinomial logit model for estimating the firms’ choice of invoicing currency. I find that all common invoicing strategies from the literature of international trade are present in the industry. The exporters use local currency pricing (LCP) for 47 % of the exported quantity, and producer pricing (PCP) in 19 % of the exports. As vehicle currencies, the producers use both euros and American dollars. A Norwegian exporter that invoice the trades in Norwegian kroner are not subject to exchange rate risk. I find that for the choice between PCP and LCP factors such as the economic size of the destination market, total import of salmon in the destination country, the frequency of trades from the exporter to the destination market, the size of the exporter, and trades to countries within the EU increases the probability for invoicing in Norwegian kroner. The fourth, and final paper, investigates the import prices obtained by Norwegian firms importing apples. Apples are a particular interesting product, as they can be imported free of tariffs during one season of approximately half of the year, while being subject for import tariffs for the reminder of the year. In the paper, we set out to investigate if the largest and most specialized firms obtain the lowest import prices, if invoicing currency is important for the prices, and how gravity variables affect the prices received by the Norwegian importers. The import prices vary markedly between the different firms. We find that largeness and specialization result in significantly lower import prices. We find that it is costly for the firms to use local currency pricing in the transactions. Increased transportation costs result in higher import prices.