To increase exports, governments focus on new or existing exporters to new export markets and on enhancing exports by existing exporters to existing export markets. Thus, understanding the relationship between first-movers and late-movers in export markets is key. First-movers generate information that can be used by late-movers in a market (Lederman et al. 2010). Exporters can learn from each other about product demand, consumer preferences, quality standards, regulations, and distribution networks at a given destination (Koenig et al. 2010). In a new paper (Haidar 2020), I study whether first-movers outperform late-movers at the product-destination level in export markets.
I use a unique disaggregated exporter-level customs dataset from nine origin countries. The analysis: (i) distinguishes old from new products at the origin-firm-product-destination level over time, (ii) orders precisely the entry of firms and products from origin to destination, (iii) looks at all (successful and failing) cases of exporters and exported products, and (iv) ensures that re-entry of intermittent products are not counted as new products when ordering the movers to a given product-destination market. The results show that late-movers outperform first-movers at the product-destination export market level.
I obtain data from the World Bank Exporters Dynamics Database (Cebeci et al. 2012). The raw data are from customs files from Burkina Fasu, Bulgaria, Egypt, Guatemala, Jordan, Mexico, Malawi, Peru, and Senegal. All non-oil-exporting firms and export transactions from these countries are included in the dataset.
The data include the following variables for each export transaction: exporter ID, HS-6 product ID, destination of shipment, value of exports , and year of transaction. The HS-6-digit level product classification illustrates the narrowness of product definitions and the richness of micro-level information available in the dataset. To test the quality of the data, I compare it with (i) UN-Comtrade data, and (ii) mirror data (what each destination reports as imports from each country of origin in the dataset). The customs dataset is highly correlated with both UN-Comtrade data and mirror data.
The descriptive statistics show that exporters do not shy away from experimenting with products and destinations. They highlight that Hausmann and Rodrik’s (2003) ‘self-discovery’ process holds not only at the macro level, but also at the micro level.
I define a first-mover as a firm that started exporting a given product to a given destination first and a late-mover as a firm that began exporting the same product to the same destination at least one year after the first-mover stepped in. In addition, I define a new product as an HS-6-digit code that was not exported by any existing exporter during the first two years of available data for any country in the dataset. This way I do not count new exporters of new products as first-movers to a given destination. Instead, I focus only on surviving exporters – i.e. existing exporters who introduced new products to a given destination – to avoid mixing new exporters (i.e. ones without prior experience) with existing exporters who step into a new market.
Following Melitz’s (2003) assumption that larger and more productive exporters would be willing to pay the market entry cost, one hypothesis can be that first-movers outperform late-movers. To test this hypothesis, I estimate a linear regression model in which I regress value, quantity, and prices of products exported at the exporter-origin-destination level on a dummy variable that equals one if an exporter is a first mover in a given product-destination market from a given origin. I also control for the number of years of experience that a given exporter has of exporting a given product at the time of exporting. By doing so, I address the concern that first-movers may tend to export low values and that not all exporters reach consumers simultaneously or quickly, as documented by Eaton et al. (2011). In addition, I control for the number of products that the exporter exports to the destination and the share of product in the exporter's overall export value. These counts include the observations they are attached to and are hence never zero, so no observations are lost by taking logs. I also include exporter-year, product-year, and origin-destination fixed effects in different estimations to control for shocks that may affect demand at the destination level as well as supply at the origin and exporter levels. Moreover, I cluster standard errors at the origin-product-destination level.
The results show that first-movers in a given market perform worse than late-movers in terms of export performance. One may argue that this result is expected because first-movers typically start small and then grow as they survive in a given market, or because first-movers are typically few and have a lot of product churning and low export volumes, as highlighted by Iacovone and Javorcik (2010). Thus, I also compare first-movers with late-movers with the same experience in a given product and also control for the number of products exported by the exporter as well as for the share of the product in total export value of the exporter. The first-mover coefficient increases and remains statistically significant. In particular, after controlling for experience, I observe that first-movers export an order of magnitude less than late-movers, suggesting that first-movers perform worse than late-movers and that my finding is not driven by the linear specification on experience or the relevance of the product. These results contradict the expectations that one would get from Melitz (2003).
One may argue that the export value performance differential between first-movers and late-movers is driven by either export quantities or export prices (i.e. quality) of exported products. The dataset includes export values and quantities, so I re-run the above regressions using quantities and prices as dependent variables. In different estimations, I let the dependent variable be the log of quantity exported by an exporter of a given product in a given year and the log of export price of a product by a given exporter in a given year. I find that the export performance differential between first-movers and late-movers is driven by export quantity, not prices. While there is a statistically significant difference in quantities exported by first-movers and late-movers, there is no first-mover effect on prices. These results contradict the prediction that first-movers may be exporting higher quality (i.e. higher price) products. These results show that first-movers do not necessarily exploit the export markets that they explore.
In a nutshell
This analysis uncovers that late-movers to a given market outperform first-movers in terms of export performance. This result holds in the presence of fixed effects that control for supply and demand shocks. It suggests that first-movers do not necessary internalise the informational externalities they generate. It also signals the absence of a ‘discovery advantage’ because, if there were one, then one would expect first-movers to sell more and grow faster than late-movers.
Further research can study the dynamics of first-movers’ and late-movers’ survival and growth once informational externalities cross the sector or country of origin dimension. Information about a given sector in a given country may inform about the same sector in neighbouring countries or other sectors in the same country.
Cebeci, T, A Fernandes, C Freund and M Pierola (2012), “Exporter dynamics database”, Policy Research Working Paper Series 6229, The World Bank.
Eaton, J, S Kortum and F Kramarz (2011), “An Anatomy of International Trade: Evidence From French Firms”, Econometrica 79: 1453-1498.
Haidar, J I (2020), “Late-Movers Outperform First-Movers in Export Markets”, Economics Letters 196.
Hausmann, R and D Rodrik (2003), “Economic development as self-discovery”, Journal of Development Economics 72: 603-633.
Iacovone, L and B Javorcik (2010), “Multi-product exporters: Product churning, uncertainty and export discoveries”, VoxEU.org, 1 August.
Koenig, P, F Mayneris and S Poncet (2010), “Local export spillovers in France”, European Economic Review 54: 622-641.
Lederman, D, M Olarreaga and L Payton (2010), “Export promotion agencies: Do they work?”, Journal of Development Economics 91: 257-265.
Makioka, R (2019), “The effectiveness of export promotion measures: A survey”, VoxEU.org, 09 October.
Melitz, M (2003), “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity”, Econometrica 71: 1695-1725.