Trade Liberalization and Exporter Dynamics – LICODYN

We develop a theoretical model of exports with learning. Firms face uncertainty initially, which explains why some exit early despite self-selection, but also why some expand early in the country and product dimensions. The model predicts differences in entry, growth and exit rates by product-country depending on their general as well as product-country-specific export experience. It also predicts the impact of lower tariffs on exports to other product-countries.
We test these predictions on data covering all French exporters in 1993-2006. We match customs transactions by firm-country-product-year to company data from tax packages. We also build a dataset of applied tariffs applicable to French exporters for each potential destination country and for around 5,000 products.
We then estimate the effect of exporting age on export growth, entry and exit by firm-country-product-year controlling for observed and unobserved heterogeneity. We exploit the heterogeneity and sequencing of tariff reductions to identify their causal effect on exports to third countries.

The theoretical model predicts that young exporters differ in their export market entry, growth and exit behavior. It characterizes how these three indicators depend on age in a given market, defined as a product/destination country pair, and on the firm's export experience in general. Empirically, these differentials are statically significant and substantial even when controlling for a large set of confounding factors. The results confirm that learning in one market extends in part to dynamics in other markets. This dynamic complementarity between markets implies that bilateral trade agreements have positive spillover effects on third countries and products not covered by the agreement. These spillover effects predicted by the model are confirmed empirically when the impact of tariff cuts following the GATT Uruguay Round is studied using micro-data. Econometric analysis confirms significant spillover effects on third countries for entry and exit, but not for export growth, in line with the model. In the future, it would be interesting to study whether intra-European trade has been affected by the Brexit, the US-China trade war or the EU-Japan free trade agreement through similar mechanisms.

Recent research has provided direct evidence of learning mechanisms in the sales of multinational company affiliates. Along with our results, this suggests that learning and experimentation are a feature of the dynamics of cross-border activity, even among large firms.

In our view, this prevalence of experimentation and learning justifies a more macroeconomic outlook. To the best of our knowledge a structural, general equilbrium model of exports or Foreign Direct Investment where uncertainty is correlated across markets is still missing at the time of writing. There are formidable computational challenges to estimating such a model, as the use of very disaggregated data (firm-product-country-year) implies a state space of exploding cardinality (number of firms x number of countries x number of products x number of years). However, we believe that our reduced-form work provides a rationale for undertaking such work.

A structural estimation framework would of course make it possible to quantify how much learning shapes trade patterns. Another important advantage would be to allow for ex ante estimation of trade policy spillovers in various counterfactual trade liberalization scenarios (such as a potential trade agreement between the EU and India, for example).

Another potential research avenue would be to examine cross-product trade policy spillovers, which are also predicted by our model. This could be particularly interesting to understand the interaction between goods and services trade policy. This is an area of interest for policymakers given the long-term trend in 'servitization' of goods producers in developed economies. This may lead to a higher estimate of the gains from liberalizing services trade.

Albornoz, F.; Calvo Pardo, H. F.; Corcos, G.; Ornelas, E. Sequentially exporting products across countries. J. Int. Econ. 2023, 142, 103735.

Money Macro and Finance Annual Conference, University of Cambridge, 2021.

CESifo Global Economy Workshop, Conference, Munich 2020.

Recent Trends in Firm Organization and Trade Dynamics Conference, CEPR-Banca d’Italia Sciences Po, Rome 2019.
FREIT EIIT Conference, Purdue University, 2018.

This research proposal aims at a better understanding of the dynamics of exports at the level of individual firms and their trade policy implications.
When firm-level trade data became widely available in the late 1990’s, they revealed a high turnover of exporting firms and a substantial contribution of entrants to aggregate export growth in the medium term. These patterns are at odds with now-standard trade models such as the Melitz (2003) model and some of its extensions. In particular existing theories based on self-selection and fixed export startup costs struggle to explain the high failure rates of new exporters (Ruhl and Willis, 2017). These theories have also been proved hard to reconcile with slow export growth and delayed entry in new foreign markets. Because aggregate exports partly respond to changes in trade costs through the number of exporters, the failure to explain these firm dynamics may lead to biased estimates of the gains from trade agreements.
Recent efforts to explain exporter dynamics have focused on ex ante uncertainty and learning (Freund and Pierola, 2010; Albornoz et al., 2012; Nguyen, 2012; Arkolakis et al., 2016). This research programme builds on two intellectual traditions. Firstly, a tradition in Industrial Organization theory recognizes the importance of learning to explain age-dependent growth and survival of firms in a market (Jovanovic 1982). Secondly, an empirical tradition in management science identifies knowledge about foreign markets as a key driver of internationalization strategies (the socalled ‘Uppsala model‘ of Johanson and Vahlne 1977).
In this project we will also consider learning mechanisms to explain firms’ export dynamics. A starting point will be previous work published in the Journal of International Economics (Albornoz et al. 2012). In that paper we offer a learning model of firms’ export growth, entry and exit and find support for their predictions in microdata on Argentinian exporters in the 2002-2007 period.
This approach will be extended in several directions. A first objective will be to extend the model to multi-product exporters. Just as understanding the number of exporters and the number of markets they access matters to understand aggregate exports, expansion of a firm’s product line is another important engine of export growth. The extended model would account for potential learning about the profitability of currently exported products in new markets as well as the profitability of new products in current export destinations.
A second objective is to explore the rich trade policy implications of such learning mechanisms. To the best of my knowledge these have never been studied. Learning implies that bilateral trade liberalization affects exports to third markets: liberalizing countries may be used as testing grounds for future exports to non-liberalizing countries; perhaps more surprisingly, firms may start exporting to close non-liberalizing countries before entering a distant liberalizing country. These trade policy spillovers matter for our understanding of gains from trade agreements and for their design.
Part of the project will therefore focus on deriving the theoretical response of aggregate trade flows to liberalization in third countries from a leaning model. Another part of the project will be to estimate the impact of bilateral trade liberalization on export dynamics in third markets. To conduct that empirical analysis we will match customs and balance-sheet data on French firms with highly disaggreated tariff data for the period 1993-2010. The analysis will exploit variations in Uruguay Round tariff reductions across products, countries and over time.
Finally, we will propose a survey of this exporting firm dynamics recent literature. In particular we will discuss ways to discriminate between competing theories of exporting firm dynamics.