Evidence on the structure of the global container shipping network, an essential determinant of the costs of trade, is scarce. This column uses satellite data to document salient features of the network, and the expansion of the Panama Canal as a natural experiment to examine the impact this improvement to one link of the network had on worldwide trade. The analysis suggests that the expansion of the canal increased world real income by $20 billion.
By Inga Heiland, Andreas Moxnes, Karen-Helene Ulltveit-Moe and Yuan Zi*
Container ships are the engines of global trade in merchandise goods. In the words of Marc Levinson, author of the acclaimed book, The box, “…the shipping container has made the world smaller and the world economy bigger” (Levinson 2006). Recent studies by Bernhofen et al. (2016) and Cosar and Demir (2017) focus on the seismic changes that the worldwide adoption of container shipping technology has brought about in international trade.
As documented by Rua (2014), by now nearly all countries have container ports, constituting the nodes of the global container shipping network. There is increasing evidence suggesting that connectivity is at least as important as geographical distance in determining freight costs. However, there is scarce evidence on the structure of the network, which is an essential determinant of the costs of trade. The networked environment also implies that a shock to a port – or a link – in the shipping network, such as improvements in shipping infrastructure, can affect shipping costs and trade flows for many countries.
In a recent paper (Heiland et al. 2019), we construct a unique and novel data set based on satellite data. The data set covers the movement of all container ships worldwide in 2016. Using these data, we document salient features of the container shipping network. We then demonstrate how information about the network can be used to investigate the impact of a local shock on trade costs, global trade flows, and real incomes across countries.
Specifically, we use the Panama Canal expansion in 2016 as a natural experiment. This allows us to identify the impact of the improvement of one link of the shipping network on worldwide trade. Exploiting route information inferred from the satellite data, we estimate the impact of the Panama Canal expansion on global trade, and quantify the trade and welfare effects of the shock through the lens of a quantitative trade model.
AIS data and the shipping network
Our empirical analysis of global container ship movements has become possible due to the rapid advent of the global Automated Identification System (AIS) over the last years. AIS reporting of vessel positions offers a degree of automation in data processing and aggregation that was not previously possible. AIS messages include information regarding vessel identity, physical appearance, voyage-related information such as draught and destination. Simply put, AIS data offer real-time information on the whereabouts of all ocean-going vessels.1
Using an exhaustive data set of all port calls made by container ships in 2016, we document novel facts about the container shipping network.
- First, container ships typically operate on fixed routes, i.e. they serve a stable set of ports, akin to buses serving a fixed number of stops in a city.
- Second, shipping activity is highly concentrated across ports, with some nodes (ports) in the network handling almost two orders of magnitude more ships than the median port.
- Third, the network is very sparse in the sense that only few countries have direct shipping routes to their trade partners. Less than 6% of all 22,650 pairs of countries with container ports are directly connected.
While the AIS data provides unprecedented detail about the movement of ships, one cannot observe the movement of the cargo itself (i.e. the actual route of a shipment from country i to country j). To make progress, we use the observed shipping network along with actual travel times between all direct port-pair links and apply standard graph theory to calculate the fastest route between any potential port pair. Consider, for example, a shipping network with direct links between New York-London, New York-Hamburg, London-Oslo and Hamburg-Oslo. The fastest route between New York and Oslo might be New York-London-Oslo if this route minimises the sum of travel times of each leg of the journey.
The fastest path calculations reveal that 52% of all country-to-country connections involve stops in more than two other countries in between. Therefore, besides adding to the distance travelled by a container, indirect routes expose bilateral flows to the shipping infrastructure of other countries.
The Panama Canal expansion
To demonstrate the importance of exposure to third-country infrastructure, we analyse the global trade effects of a large improvement in local shipping infrastructure in 2016: the expansion of the Panama Canal. After ten years of construction, the extended Panama Canal opened on 26 June 2016. The massive $5.25 billion construction project was a modern engineering marvel: it nearly doubled the capacity of the canal by adding a wider and deeper third lane.
We employ our information on shipping routes to explore how exporters and importers worldwide were differentially affected by this local change in the shipping infrastructure. We find that country pairs whose fastest connection passed through the Panama Canal prior to the expansion traded 9-10% more after the expansion compared to other country pairs.
Finally, we use a canonical model of trade to quantify the general equilibrium effect of the Panama Canal expansion. Based on our analysis, the expansion increased world real income by $20 billion. While the building costs were borne by Panama alone, the gains per capita were shared by many countries, due to the network structure of shipping.
Figure 1 Panama Canal exposure by country
Note: The figure shows the share of imports passing through the Panama Canal in total imports by country.
The way forward
There is a growing number of studies using satellite data for economic analysis (see Donaldson and Storeygard 2016 for a review of applications). So far, however, only a few recent papers have used shipping satellite data to explore issues related to trade.
Our analysis does not just highlight the importance of shipping networks, it also points to how shipping satellite data can be used within the field of international trade. We expect many more applications in the years to come.
*About the authors:
- Inga Heiland, Assistant Professor, Department of Economics, University of Oslo
- Andreas Moxnes, Professor of Economics, University of Oslo & Princeton University
- Karen-Helene Ulltveit-Moe, Professor of International Economics, University of Oslo; and CEPR Research Fellow
- Yuan Zi, Assistant Professor in the Department of Economics, University of Oslo
Bernhofen, D, Z El-Sahli and R Kneller (2016), “Estimating the effects of the container revolution on world trade”, Journal of International Economics 98: 36-50.
Cosar, K and B Demir (2017), “Containers and globalisation: Estimating the cost structure of maritime shipping”, VoxEU.org, 13 June.
Donaldson, D and A Storeygard (2016), “The view from above: Applications of satellite data in economics”, Journal of Economic Perspectives 30(4): 171–198.
Heiland, I, A Moxnes, K H Ulltveit-Moe and Y Zi (2019), “Trade from space: Shipping networks and the global implications of local shocks”, CEPR Discussion Paper 14193
Levinson, M (2006), The Box: How the Shipping Container Made the World Smaller and the World Economy Bigger, Princeton University Press.
Rua, G (2014), “Diffusion of containerization”, Finance and Economics Discussion Series 2014-88, Federal Reserve Board.
 Vessels send out AIS signals identifying themselves to other vessels or coastal authorities, and the International Maritime Organization (IMO) requires all international voyaging vessels with above 300 gross tonnage and all passenger vessels to be equipped with an AIS transmitter. This implies that all container ships carrying any significant amount of cargo are parts of our data universe.