Lies, damned lies and ...
The sanctity of trade statistics
Bilateral trade flow data are misleading. But a reported tweak will not help
According to a report in the Wall Street Journal, discussions are afoot on changing the way trade figures are tallied. The Bureau of Economic Analysis, the country’s main statistical body, calls this “completely inaccurate”. But in trade as elsewhere, the new administration seems prone to using statistics as a drunk uses a lamppost—for support rather than illumination.
The proposal reportedly involves stripping out some of America’s exports from the gross numbers. America sold $1.5trn of goods abroad in 2016, but of that $0.2trn were re-exports that left the country much as they had arrived. This type of trade has been growing, reflecting America’s role as a hub for North American trade. As a share of its combined exports to Mexico and Canada, re-exports rose from 12% to 20% between 2002 and 2016. Truckers and shippers benefit from this kind of trade. But critics see it as “padding”, obscuring gloomier trends in “made in America” exports.
So excluding re-exports from the total would provide Mr Trump with some more eye-popping figures with which to bash Mexico. A bid to tweak trade statistics need not be politically motivated, though. It could also reflect the (correct) realisation that standard measures of imports and exports do not always capture what is really being “made in America”.
Statisticians do sometimes adjust for re-exports, which can mask underlying trends. For example, they routinely strip out from Hong Kong’s figures its re-exports (a staggering $498bn-worth in 2016, compared with domestic exports of $13bn) to avoid double-counting China’s exports in world-trade totals.
Such adjustments are supposed to deal with the underlying gripe with re-exports: that they may not reflect a country’s value added. But tackling this properly involves a much deeper dig into the data. There is also foreign value added embedded in American exports, such as the Mexican parts in cars made in Michigan. The imports side is just as important. American imports from Mexico include both American value added and inputs from other countries. Accounting for all this is far more complicated than stripping out just one component.
Drilling down into bilateral trade relations, accounting for value added has big effects. But these data suggest that some might not be as large as often assumed. One commonly-cited factoid is that 40% of Mexican exports to America are embedded American content. New figures from the OECD put that figure at 14% (see chart).
That is still high enough to create a lot of American losers were America to sever trade relations. And the effect on the reported trade imbalance between America and Mexico is dramatic. Overall, however, switching to the more sophisticated value-added measure of trade flows would not provide political ammunition as powerful as ditching re-exports. On a value-added measure, the bilateral-trade imbalance between America and Mexico in 2011 was 43% smaller than the gross trade flows would suggest. The trade deficit with Canada would have become 39% smaller.
Focusing on value-added trade data is better than looking at the gross flows, but Mr Johnson questions whether the debate should focus on bilateral imbalances at all. When someone incurs a trade deficit with a bookshop and a trade surplus with his employer, neither matters in isolation—the overall balance is important. And for a country’s trade, that will be most determined by macroeconomic factors. Fiddling the figures might move the lamppost; it will still leave the future direction of trade in the dark.