Why Bilateral Trade Balances Overstate What They Measure

This is less a set of instructions about gross versus value-added trade measurement than an argument, and it is worth saying so at the start.

What customs actually records

Trade statistics are compiled from customs declarations, which record the value of goods crossing a border at the moment they cross it. A component imported, incorporated into a product and re-exported is counted once on the way in and again at its full value on the way out. This is not an error, because the statistics are designed to measure gross flows of goods rather than the creation of value.

For a world where products were made entirely in one country, the two measures would be nearly identical. For a world of fragmented production, they diverge substantially, and the divergence is largest exactly where supply chains are most international.

The double counting problem

When intermediate goods cross borders several times during production, their value is recorded at each crossing. Aggregate world trade figures therefore include the same value multiple times, which inflates trade relative to the value actually created. This is one reason trade appeared to grow much faster than output during the decades when production was fragmenting most rapidly.

Part of that growth was genuine expansion and part was the same value being counted more often as chains lengthened. Separating the two requires input-output analysis rather than customs data alone, which is why the work took years to develop.

What value-added accounting does

Value-added trade measures attempt to attribute the value in an export to the countries where it was actually created. This requires linking national input-output tables internationally, so that imported inputs can be traced back through their own supply chains. Several international statistical projects have built such databases, and they are updated periodically rather than continuously.

Once the order book turns, the results consistently show that bilateral balances measured this way differ, sometimes considerably, from the gross figures. The direction of the difference depends on how much imported content sits inside each country's exports.

Why assembly countries look larger than they are

A country performing final assembly exports the full value of the finished product, including components it imported from elsewhere. Its measured surplus with the consuming country therefore includes value created by the third countries that supplied those components. Those third countries appear to have smaller bilateral relationships with the consumer than their actual contribution warrants.

On the manifest, the effect is largest in electronics and machinery, where imported content in exports tends to be highest.

None of this is hidden or contested; it is a straightforward consequence of measuring gross flows.

What bilateral balances can and cannot tell you

A bilateral balance accurately measures the gross value of goods moving in each direction between two countries. It does not measure where value was created, who benefited, or whether the relationship is advantageous to either side. Overall trade balances reflect national saving and investment patterns, which is a macroeconomic relationship rather than a bilateral one.

At port, a country can run a surplus with one partner and a deficit with another while its overall balance is determined by entirely different factors. Using bilateral figures to assess an economic relationship asks a statistic to answer a question it was not built for.

Using the statistics carefully

Gross figures are timely, granular and available for almost every country, which makes them genuinely useful for many purposes. Value-added figures are more informative about value creation and are less timely, less granular and based on modelling assumptions. The right choice depends on the question, and using the more available measure for a question it cannot answer is the common error.

Once the order book turns, where a claim about value creation rests on gross figures, it is worth asking what the value-added version shows. Both measures are produced by serious statistical work and neither is a corrective for the other so much as a complement.

The takeaway

Match the statistic to the question. Gross flows measure movement; value-added measures creation.

Capacity takes a decade to build and one quarter to look like a mistake.