Abstract
Exporters’ performance in a given market may affect their exports to the rest of the world. Importers base their future transaction decisions upon the information revealed by exporter’s performance in other countries. This paper estimates significant effects from these information spillovers on the export patterns of fourteen developing countries, and somewhat smaller effects for a sample of exporters from six developed countries. On the other hand, it is in developed countries’ markets that the largest information spillovers are generated. Indeed, increases in market share in the United States allows for significant increases in exports to the rest-of-the world associated with information spillovers. But developing country markets could also generate important amounts of information for regional exporters. Hong Kong is the top market in terms of generating information for other East Asian exporters, and the Argentinean and Chilean markets play an important role for exporters from other Latin American countries.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Martin et al. (2005) use this variable to capture the impact of information on the probability of war.
Note, however, that these are likely to be correlated (the correlation coefficient is 0.5 in our sample).
Nevertheless, when simply running a correlation of bilateral phone calls on the sum of exports and imports of bilateral trade in newspapers for a sample of twenty two OECD countries (for which the availability of bilateral phone call data is comprehensive) we found a correlation of 0.8, suggesting that both proxies are capturing similar variations in the data.
The use of the internet has changed this in recent years, but it was not an important instrument for exchange of information during the period under examination here (1969–1997).
To see this, accept for the moment that trade flows are exclusively determined by factor endowments. Two countries with identical factor endowments will not trade with each other, but may have a significant exchange of information.
Note that the random component of the utility function does not interact with quantities consumed, which significantly simplifies the algebra.
Note that quantity consumed and the value of consumption are now identical by construction.
See Anderson et al. (1992).
Note that in equilibrium \( a_{{g,c}} = 1 \mathord{\left/ {\vphantom {1 {{\left( {1 - s_{{g,c}} } \right)}^{{1 \mathord{\left/ {\vphantom {1 \beta }} \right. \kern-\nulldelimiterspace} \beta }} {\left( {{\sum\nolimits_{k \ne c} {a^{\beta }_{{g,k}} } }} \right)}^{{1 \mathord{\left/ {\vphantom {1 \beta }} \right. \kern-\nulldelimiterspace} \beta }} }}} \right. \kern-\nulldelimiterspace} {{\left( {1 - s_{{g,c}} } \right)}^{{1 \mathord{\left/ {\vphantom {1 \beta }} \right. \kern-\nulldelimiterspace} \beta }} {\left( {{\sum\nolimits_{k \ne c} {a^{\beta }_{{g,k}} } }} \right)}^{{1 \mathord{\left/ {\vphantom {1 \beta }} \right. \kern-\nulldelimiterspace} \beta }} } \) where \( s_{{g,c}} = {d_{{g,c}} } \mathord{\left/ {\vphantom {{d_{{g,c}} } {d_{g} }}} \right. \kern-\nulldelimiterspace} {d_{g} } \) is the market share of variety c for good g. Note also that this system cannot be solved as there are more unknowns than independent equations.
We exclude agriculture products from the sample due to their high levels of protectionism that will significantly distort bilateral trade flows and that are difficult to measure due to their complexities (tariff quotas, specific tariffs, seasonal tariffs, etc...). According to tariff data available at http://www.worldbank.org/research/trade for a sample of 82 developing and developed countries agriculture ad-valorem tariffs are two to three times larger than manufacturing tariffs and those numbers exclude tariff quotas, specific and seasonal tariffs which are much more predominant than for manufacturing products.
Trade data sources are from national sources compiled by the United Nations in Comtrade’s data base. For a description of data sources and variable construction see the Appendix.
Unlike in the linear regression case, where the fixed effects can be eliminated by taking first differences, there is no simple transformation in the case of censored regressions (see Hsiao 2003, pages 243–249).
To see this, note that if the market share in country k affects exports to country mc, then it is also true that country c market share in country mc (and therefore exports of c to country mc) will affect exports of c to country k (and therefore the market share of country c in k).
A different specification of import demand information spillovers was used using export levels instead of market shares and results, which were robust to these alternative specification, are available from the authors. Note that when using levels we face a dynamic panel. In order to avoid the complexities of estimating dynamic panels with censored data, we dropped all observations for which reported exports were below 1,000 dollar and we used an Arellano–Bond estimator for dynamic panels.
We also tested for the presence of information spillovers within two-digit SITC categories. Results suggested the presence of two-digit spillovers for developing countries. For developed countries, the sign of the coefficient turned out to be negative. This is most likely due to the high correlation between the two-digit and the three-digit information spillover variables, making these two difficult to disentangle.
One may also wonder whether the importance of information spillovers from other countries declines as the bilateral relationship between the exporter and the importer becomes more significant. In order to test for this we interacted the bilateral market share variable with the information spillover. In the pooled sample and in the sub-sample of developing countries exporters we found a negative sign, suggesting that the need for information from other countries declines as the bilateral relationship is strengthened. For the sub-sample of developed countries exporters in the other hand, we obtained a positive coefficient.
See Portes et al. (2001) for a similar result.
The changes in sign for distance are driven by data for the United States and Canada as can be seen from the exporter-specific regressions in Table 2.
In the exporter-specific regressions we assume that the random effects are product and importing country specific, but we don’t allow them to vary through time.
Note that in the pooled sample, Mexico is the first Latin American country in terms of information spillovers (before Argentina, Chile and Colombia).
References
Anderson, J. and vanWincoop, E., “Gravity with gravitas: a solution to the border puzzle,” American Economic Review, vol. 93(1), pp. 170–192, 2003.
Anderson, S., de Palma, A. and Thisse, J.-F., Discrete Choice Theory of Product Differentiation. MIT: Cambridge, MA, 1992.
Casella, A. and Rauch, J., “Anonymous market and group ties in international trade,” Journal of International Economics, vol. 58, pp. 19–47, 2002.
Feenstra, R., “Border effects and the gravity equation: consistent methods for estimation,” Scottish Journal of Political Economy, vol. 49(5), pp. 491–506, 2002.
Feenstra, R., Hanson, G. H., and Lin, S., “The value of information in international trade: gains to outsourcing through Hong Kong,” NBER working paper no 3928, 2002.
Gould, D., “Immigrants links to the home country: empirical implications for U.S. bilateral trade flows,” Review of Economics and Statistics, vol. 76(2), pp. 302–316, 1994.
Head, K. and Ries, J., “Immigration and trade creation: econometric evidence from Canada,” Journal of International Economics, vol. 31(1), pp. 4762, 1998.
Head, K. and Mayer, T., “Everything you always wanted to know about CES (but were afraid to ask),” mimeo, University of British Columbia, 2001.
Hsiao, C., Analysis of Panel Data, 2nd edition. Cambridge: Cambridge University Press, 2003.
Hummels, D., Toward a Geography of Trade Costs, mimeo, Krannert School of Management, Purdue University, 1999.
Martin, P., Mayer, T., and Thoenig, M., Make trade, not war, mimeo, University of Geneva, 2005.
McLaren, J., “Supplier relations and the market context: a theory of handshakes,” Journal of International Economics, vol. 48, 121138, 1999.
Portes, R., Rey, H., and Oh, Y., “Information and capital flows: the determinants of transactions in financial assets,” European Economic Review, vol. 45(4), pp. 783796, 2001.
Rauch, J., “Networks versus markets in international trade,” Journal of International Economics, vol. 48, 139150, 1999.
Rauch, J. and Trindade, V., “Information and globalization: wage co-movements, labor demand elasticity, and conventional trade liberalization,” NBER working paper no 7671, 2000.
Rauch, J. and Trindade, V., “Ethnic Chinese networks in international trade,” Review of Economics and Statistics, vol. 84(1), pp. 116130, 2002.
Veall, M. and Zimmermann, K., “Pseudo-R2 measures for some common limited dependent variable models,” Journal of Economic Surveys, vol. 10(3), pp. 241–259, 1996.
World Bank, Reputation in manufactured goods trade, industry series paper no 21, Washington, 1989.
Zlotnik, H., “International migration 1965–96: an overview,” Population and Development Review, vol. 24(3), pp. 429–467, 1998.
Acknowledgements
We would like to thank Simon Evenett, Robert Feenstra, Carsten Fink, Jean-Marie Grether, Olivier Lumenga-Neso, Jaime de Melo, Francis Ng, Garry Pursell, Maurice Schiff, Isidro Soloaga, David Tarr, Anthony Venables, and participants in seminars at the USITC and the World Bank for very helpful comments and suggestions. The views are the authors’ and do not necessarily reflect those of the institutions with which they are affiliated.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
Variable construction
The variables are constructed as follows. The market share of the exporter of good g from country c to the importing country mc at time t:
where x g,c,mc , t are exports of good g from the source country c to the importing country mc at time t; and m g,mc,t is defined as total imports of country mc of good g at time t (note that ideally one would need to include consumption of domestically produced goods, but this data does not exist at the level of disaggregation with which we are working and would only be available for limited time periods for most countries at more aggregate levels, e.g., ISIC 4 digit). All trade data is measured in constant 1987 dollars.
The element j of the vector of information flows, I c,tI , are defined as:
where \( \Psi _{{k \to {\text{mc}},t}} \) is the bilateral exports of newspaper from country k to country mc at time t.
Data sources
Trade data sources are from the national sources compiled by the United Nations in Comtrade’s database. The analysis is carried using manufacturing products trade data given of SITC rev.1 classification at the three digit level. The 60 importing countries in our sample are given in the first column of Table 3. The 20 exporting countries are listed in Table 2.
Data on newspaper trade are also provided by the Comtrade’s database (SITC rev. 1, code 8922).
Distance data are calculated using geodesics distance between countries’ capitals. The variables for common language and sharing of a border have been obtained from Andrew Rose’s web page at: http://faculty.haas.berkeley.edu/arose/.
Rights and permissions
About this article
Cite this article
Nicita, A., Olarreaga, M. Information and Export Performance. J Ind Compet Trade 7, 95–111 (2007). https://doi.org/10.1007/s10842-007-0008-8
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10842-007-0008-8