Assume 50 per cent of China’s export value equates to embedded imports. You could think of this amount of exports and imports as the “currency hedged” trade position – i.e., in loose terms, hedged against RMB currency appreciation. E.g. it would reasonably hedged if these exports and imports were all denominated in US dollars; less so perhaps to the degree there is a currency mix. But in general, on this portion of trade flows, the lower RMB value of export revenues would tend to be offset by the lower RMB cost of embedded imports. RMB exposure at the macro level would tend to be immunized on the “hedged” trade position as defined. Exporters would not be forced to raise their prices as much on this portion of trade, for a given RMB appreciation.

Then define the “un-hedged” trade position as total gross trade flows less the “hedged” position. This might also be referred to as the “value-added” trade position. By construction, the exchange rate sensitivity of the “un-hedged” position should equal the exchange rate sensitivity of the total trade position.

The “un-hedged” trade position has its own trade surplus, which is equivalent in size to the total trade surplus. This may be a little bit artificial, in that it associates the trade surplus, an aggregate net measure, with some specific gross components. But it seems like a reasonable decomposition for purposes of looking at foreign exchange risk.

The “un-hedged” trade position could be broken down further into the trade surplus plus the rest. This could be viewed as the “value-added” mismatched position (trade surplus) plus the “value-added” matched position.

The size of the un-hedged position and its components should then be assessed for foreign exchange risk relative to the size of GDP. Foreign exchange risk for GDP will then be a function of the size of the total un-hedged position and the profile of its mismatched and matched components, relative to GDP.

This seems to be roughly consistent with your own interpretation, regarding the importance of the size of China’s export sector to the conclusion of the article. Also, (correct me if I’m wrong), it seems to fit in the sense that your value-added percentage transformation in the second graph appears to preserve the size of the actual trade surplus by construction. I think the gap between your export value added and non-processing imports is what I’ve called the “un-hedged” trade surplus, which is equivalent to the actual (total) trade surplus.

Moreover, given the equivalence of the “un-hedged” trade surplus and the actual (total) trade surplus, it suggests in a strange way that the FX sensitivity of the trade surplus on its own is somewhat independent of the issue of import or domestic content. The size and proportion of the trade surplus, relative to total trade flows, represents a sort of upper bound for the embedded import content of exports and a lower bound for the domestic content. Again, I think this reinforces your modified statement regarding the conclusion of the paper. If this is a reasonable approach to FX risk on trade, what I haven’t started to consider is the relative GDP exposure to FX risk for each of the two components of the “un-hedged” trade position. Each adds some FX sensitivity, but I’m not sure how to think about that.

This is all very roundabout, perhaps, but the general idea seems consist with your statement:

“Exchange rate appreciation should have a smaller impact on China than on countries with similarly sized export sector and higher-domestic value-added. But given China’s geographic size, China’s export sector is actually quite large relative to its economy.”

(I put this together quite quickly; may well have goofed some point of logic. E.g. it would see, not work very well to the degree that imports were invoiced in currencies that were implicitly pegged to the RMB. I don’t know enough about the trade flows to know this. But it seems to contradict the starting premise of the article’s assumption about trade FX sensitivity.)