How does international trade affect the risk exposure of firms and countries? The literature is split on this question. Some papers suggest trade enhances volatility, while others argue that trade dampens it.
Greater participation in international markets, by increasing the role of large firms as well as sectoral concentration, may magnify a country’s vulnerability to idiosyncratic supply shocks (di Giovanni and Levchenko 2009, 2012). However, cross-country diversification can also dampen macroeconomic volatility by reducing exposure to domestic demand shocks (Caselli et al. 2015). Focusing on micro-demand shocks, Kelly et al. (2013) document that firms with a more diversified customer base display a lower volatility, which suggests trade opportunities may indeed reduce volatility by increasing diversification.
The apparently conflicting results in the literature show that the structure of exporters’ sales, within as well as across destinations, must be analysed together with the various shocks that hit firms and countries in international markets to understand the trade-volatility nexus. The ultimate impact of trade on volatility depends on the nature of shocks that prevail in international markets, and the extent to which exposure to such shocks is diversified within and across firms.
In Kramarz et al. (2019), we contribute to this debate with an integrated analysis of the sources of volatility of firms and countries in international trade markets. We model the universe of trade relationships between French exporters and their European partners, observed over a period of 15 years. The richness of the data, together with a new empirical strategy, allows us to provide a comprehensive decomposition of the different sources of trade volatility at both the firm level and in the aggregate. We find that individual shocks hitting the foreign partners of French exporters are a key component of individual and aggregate volatility.
A key feature of the data is that we observe firm-to-firm exports, which allows us to recover a variety of shocks at the root of the growth of trade flows. Trade flows may be driven by a change in the competitiveness of French exporters. The competitiveness itself originates either from ‘aggregate’ shocks that are common across firms in an industry, or from idiosyncratic shocks to firms’ productivity. Firm-to-firm trade also varies due to foreign ‘demand’ shocks that are either aggregate in nature (i.e. common to all buyers within a destination) or idiosyncratic, hitting the importing firm or eventually the match it forms with a particular French exporter.
A novel empirical strategy allows us to estimate the different shocks driving firms’ trade growth. These estimates are then combined with the micro-structure of trade flows to quantify the importance of various shocks for the volatility of exports at different levels of aggregation. Computations reveal that a substantial share of the individual and aggregate volatility of exports is driven by individual shocks hitting foreign buyers.
Volatility in the small
The first set of results is about the sales’ volatility of individual exporters. We find that aggregate shocks, whether they hit the supply or the demand side of the market, do not generate much volatility. Their variance is indeed small, in comparison with microeconomic risks. By contrast, both microeconomic supply and demand shocks generate a substantial amount of volatility. In the baseline estimation, they respectively represent 45 and 50% of firms’ volatility.
That microeconomic supply shocks are an important source of firm-level volatility is not surprising as such shocks cannot be diversified through operational hedging. What is more surprising is the substantial contribution of microeconomic demand shocks to firm-level volatility, which could have naturally faded away when the firm diversifies its portfolio of clients. These shocks are quantitatively important because (1) their estimated volatility is high, and (2) exporters’ sales are highly skewed towards their main partner. In our data, more than 90% of exporters derive more than half of their sales in one destination from a single partner (Figure 1, triangle lines). Such high concentration of French firms’ exports explains their strong exposure to idiosyncratic demand shocks.1
Figure 1 Cumulated distributions of sellers’ degrees
Number of destinations per seller
Number of buyers per seller-destination
Note: The figure shows the cumulated distributions of the sellers’ number of export destinations (top panel) and their number of partners within a destination (bottom panel). The circle lines are based on each seller’s overall exports. The distributions labelled “Top X% Sales” are computed from a sub-sample restricting the amount of each firm’s sales to the X first percentiles of the distribution of sales when transactions are ordered by the decreasing share of the destination (top panel) / buyer (bottom panel) in the firm’s total sales. Interpretation – around 25% of sellers serve a single European destination and more than 40% serve a single partner in their typical export destination. Less than 20% of sellers serve at least 7 destinations and more than 5 buyers per destination (Circle lines). 60% of sellers recover at least 90% of their export sales from a single destination (diamond line, left panel). More than 70% of sellers make at least 90% of their bilateral exports from a single importer.
Volatility in the large
The second set of results is about the volatility of aggregate exports. Unlike in the small, we find that aggregate shocks explain a substantial part – about 40-60% – of aggregate volatility. Nonetheless, individual shocks still account for the lion’s share of aggregate volatility. This result is reminiscent of Gabaix (2013) and di Giovanni et al. (2014), who show that a large share of the volatility of the US and French GDPs is driven by shocks to individual firms. The reason is that the distribution of sales among firms is extremely concentrated. In our export data, the 10% largest exporters thus account for 90% of French exports in Europe. Because of this extreme concentration, idiosyncratic shocks affecting the right tail of the distribution do not wash out in the aggregate. The novelty of our findings is that we can further dig into the underpinning of this ‘granular’ volatility. We separate granular fluctuations coming from idiosyncratic shocks to producers and from idiosyncratic shocks to their foreign partners. Depending on the specification, we find that individual foreign demand shocks explain up to 30% of the volatility of bilateral exports and 20% of fluctuations in aggregate intra-EU exports.
The prevalence of demand-driven granular fluctuations in trade data reveals a lack of diversification in exporters’ sales, including among superstar exporters. This result is a bit surprising as we would expect large exporters to be better diversified and therefore less strongly exposed to idiosyncratic demand shocks. They export to a larger number of destinations and tend to serve more buyers, within a destination. Figure 2 shows that this is true in our data, although the lower exposure to idiosyncratic demand shocks of these firms is not sufficient to compensate for their granular impact in exports.
Figure 2 Volatility, by decile of exporters’ size
Note: This figure represents the median volatility of individual exports across deciles of exporters’ size. Exporters are grouped into size bins based on their initial exports, with bin 1 corresponding to the 10% smallest exporters. For each decile, the figure reports the median volatility of export sales (“Volatility seller”), the median volatility attributable to idiosyncratic demand shocks (“Buyer-driven vol.”) and the median volatility induced by idiosyncratic supply shocks (“Seller-driven vol.”). See Kramarz et al. (2019) for details.
The extreme concentration of trade flows among a few large suppliers and a few large foreign buyers thus makes trade vulnerable to idiosyncratic shocks hitting large players in international markets. Whereas the concentration among domestic suppliers is now well established, our results suggests that concentration of foreign buyers matters as well and should be taken into account to fully capture the dynamics of aggregate trade and production.
Caselli, F, M Koren, M Lisicky, and S Tenreyro (2015), “Diversification through Trade”, NBER Working Paper no. 21498.
di Giovanni, J, and A A Levchenko (2009), “Trade Openness and Volatility”, The Review of Economics and Statistics, 91 (3), 558–585.
di Giovanni, J, and A A Levchenko (2012), “Country Size, International Trade, and Aggregate Fluctuations in Granular Economies”, Journal of Political Economy, 120 (6), 1083-1132
di Giovanni, J, A A Levchenko, and I Mejean (2014), “Firms, Destinations, and Aggregate Fluctuations”, Econometrica, 82 (4), 1303–1340.
Gabaix, X (2011), “The Granular Origins of Aggregate Fluctuations”, Econometrica, 79 (3), 733–772.
Kramarz, F, J Martin, and I Méjean (2019), “Volatility in the Small and in the Large: The Lack of Diversification in International Trade”, Journal of International Economics, forthcoming.
 Interestingly, such a lack of diversification is observed in supposedly well-integrated markets, namely good markets within the EU. Since a firm’s number of foreign clients is a decreasing function of the destination’s market potential, we conjecture that the lack of diversification, and its impact on the volatility of exports, is even stronger outside of the EU.