The project investigates the process through which firms enter into and maintain relationships with buyers in foreign markets. It will provide microfoundations for understanding how macroeconomic shocks such as exchange rate changes, trade liberalizations, and financial crises affect trade patterns. It will also provide an empirical description of the formation of new cross-border business relationships, and of the ways in which exporters and importers adjust their behavior as their relationships mature and they learn about themselves and each other. The research will be based on newly-available data sets that describe all legal merchandise trade between Colombia and the United States. These data will be merged with establishment-level information on the exporters and importers who do business with one another. The resulting data set will allow project participants to model exporter-importer search and matching, the dynamics of post-match shipments, and the conditions under which buyer-seller matches break up. In exploratory work with these shipments data, project investigators have documented some striking patterns. In a typical year, one-third to one-half of Colombian exporters are new, that is, they were not exporting to the U.S. in the previous year. The great majority of these new exporters ship tiny amounts, service a single buyer, and disappear from the U.S. market in the following year. But a small fraction survive and grow very rapidly, both in terms of sales per buyer and in terms of number of buyers. To explain these patterns the investigators have developed a dynamic search and learning model of exporter behavior. In this model, producers who hope to sell in a particular foreign market devote resources to identifying potential buyers there. When they find one, the value of the resulting sales conveys a noisy signal about the appeal of their products in this market. Taking stock of this new information, firms update their beliefs about the scope for export profits, and adjust the intensity of their search for more buyers accordingly. Thus, in addition to responding to exchange rate and cost shocks, firms' exports reflect their endogenous efforts to add new buyers to their client base. Project investigators plan to extend their work in three ways. First they will generalize their proto- type model to (a) allow firms to learn from the experience of competing exporters, and from their own experiences in other (non-U.S.) markets, and (b) allow the history of each business relationship to affect its future evolution (duration and sales trajectory). Once these generalizations have been made, they plan to move toward formal econometric estimation of the model's parameters. Second, rather than viewing matches as purely random, they plan to treat them as reflecting assortative matching, and to characterize econometrically the "mating" process, drawing on the marriage literature. Finally, to inform their modeling efforts, the investigators intend to conduct interviews with personnel at Colombian manufacturing firms with export experience, as well as with personnel at firms in the U.S. that are part of the potential client base.
Intellectual Merit. The research will provide a framework for assessing quantitatively how macroeconomic policies affect the export participation of individual producers over time, reconciling observations at the transactions level with aggregate data.
Broader Impact. The framework will address questions about how individual firms react to, and benefit from, the opportunity to trade. It will provide a set of tools to assess a wide set of policy questions concerning macroeconomics and trade policy, and innovation. The models developed may also shed light on the way that intra-national business relationships form and evolve.
This project uses shipments-level trade data to develop and estimate new models of exporter behavior. Its main findings are summarized below. (1) The project begins with a descriptive analysis of individual merchandise shipments from Colombia to the United States, 1992-2009. Several striking patterns emerge. First, most new exporters drop out of the U.S. market within a year, especially those that begin with small-scale shipments. But those who survive this shakedown period have much lower exit rates in the future. Indeed, surviving members of new cohorts tend to expand their sales very rapidly, gaining market share as they mature. Second, most buyer-seller matches are short-lived, lasting less than two years, on average, so successful exporters must frequently replenish their portfolio of clients. Third, although the typical Colombian exporter ships to 1 or 2 U.S. clients per year, a handful of large sellers deal with 30 or more U.S. importers. These latter exporters accounts for a substantial fraction of total shipments. (2) Project researchers next develop and estimate a model that is consistent with these facts. It is based on the conjecture that firms' exporting behavior reflects search and learning processes. Producers who are interested in a particular foreign market devote resources to searching for potential buyers there. The foreign buyers they find either reject their products or establish business relationships with them. In doing so, these buyers generate noisy signals to the producers who found them about the local popularity of their products. Business relationships generate export revenues, but they eventually become unprofitable or fail for exogenous reasons. So producers who don’t invest in finding new foreign buyers tend to drop out of the export market. In deciding how intensively to search for new buyers, they take stock of market-wide conditions and their history of successful and unsuccessful encounters. Producers also recognize that establishing successful relationships reduces their future search costs. In this sense, the model allows for network effects. Fit to the shipments data described above, this model quantifies the costs of finding buyers. For example, the typical firm that has not yet successfully exported must pay about $51,000 to find one new potential client per year. However, network effects push search costs down dramatically. For example, among firms that have established their first exporting relationship, the typical cost of finding one new client per year is only $3,900. Finally, through a combination of highly appealing products and lucky early draws, a handful of firms is able to get past the costly initial stage of building a client base. These firms account for a large portion of aggregate exports. Importantly, without a network effect our model is unable to replicate this feature of the data. The analysis concludes with some experiments that quantify the effects of export promotion policies and exchange rate fluctuations on export growth, with versus without learning and network effects. Network effects prove particularly important in replicating two key features of aggregate export dynamics: inertia, and fluctuations unrelated to exchange rate shocks. (3) Project participants go on to develop a variant of their basic model that characterizes "born to export" firms in developing countries. This model is inspired by the experiences of Bangladeshi apparel producers, most of whom have focused exclusively on foreign markets since their birth because western-style apparel is in scant demand domestically. The analysis is also meant to characterize other "export-only industries" in developing countries, including, for example, cut flowers in Colombia and Kenya and Chilean salmon farmers. Several messages come out of this analysis. First, it is particularly risky to begin exporting goods that lack a domestic market. Entrepreneurs wishing to do so must first sink substantial investments in productive capacity, rather than simply re-direct some of their existing capacity toward foreign markets. Further, they must commit to these investments without the benefit of production experience or feedback from domestic consumers. Thus modest exporting incentives are unlikely to generate much of a response in these industries. Second, since the potential entrants to an export-only industry lack idiosyncratic experiences in domestic markets, they are likely to hold similar beliefs about their prospects in export markets. This makes them tend to move as a herd—or not at all—in response to market-wide shocks, be they informational (e.g. signals generated by pioneer firms) or economic (e.g. changes in foreign demand). Third, when they do move, export-only industry entrants commit their entire capacity to foreign sales, so these industries are capable of generating dramatic export surges. Finally, once they emerge, new exporters in export-only industries are relatively likely to remain in foreign markets. This is because they cannot reorient their production toward domestic consumers when market conditions deteriorate abroad. These predictions are borne out by comparisons of Bangladeshi apparel exporters with apparel exporters in countries where western-styled apparel can be sold domestically.
