One of the persistently challenges I consistently face in trying to model US import and export flows is the sensitivity of the results to the inclusion of time trends. Time trends are bothersome because they are, in one sense, a measure of our ignorance. That’s a worry as we consider the feasibility of global rebalancing    .
One can whittle away at the role of time trends by including in the standard formulation a supply factor, so that for instance US imports depend on US income, the US dollar real exchange rate, and foreign export supply capacity. (How to proxy that latter variable is a vexed question). But in addition, we know that trade costs have varied over time (see the posts on de-globalization:  ).
Here are two graphs that highlight the potential importance of trade costs. The first graphs US log US durables and nondurables ex.-oil minus log GDP against the average tariff rate for US, Japan and European Union. Notice that as tariff costs decline, the trade intensity of GDP increases.
Figure 1: Tariff factor (1+t) for US, Japan, EU, interpolated (blue line, left scale), log nondurables good imports ex.-oil minus log US GDP (red line, right scale), and log durables good imports to log US GDP (green line, right scale), flows in Ch.2005$. NBER defined recession dates shaded gray; assumes last recession ended 09Q2. Source: Yi (2003), BEA, 2010Q1 2nd GDP release, NBER and author’s calculations.
The graph is merely suggestive, as it doesn’t include other trade costs, including transportation, and implicit costs (legal, administrative, logistical, communication). What this graph suggest is that goods imports might not trend in the same way going forward as over the past decade and a half (remember, some of these imports are incorporated into exports).
One message I don’t want people to take from this is that it would be a good idea to increase trade costs (via explicit or implicit protectionism) to reduce imports. That’s because the US is going to need to export, going forward, in order for the world economy to rebalance. Effective exporting requires in part ability to import the low cost components. But in addition, we need open export markets to be able to export. That leads to the obvious question of what the corresponding graph looks for exports. Here it is:
Figure 2: Tariff factor (1+t) for US, Japan, EU, interpolated (blue line, left scale), log nondurables good exports minus log Rest-of-World GDP (red line, right scale), and log durables good exports to log Rest-of-World GDP (green line, right scale), trade flows in Ch.2005$. NBER defined recession dates shaded gray; assumes last recession ended 09Q2. Source: Yi (2003), BEA, 2010Q1 2nd GDP release, Federal Reserve, NBER and author’s calculations.
For more on income and price elasticities, and accounting (or at least trying to account) for supply factors as well as trade costs, see this paper.
By the way, some of these trade costs are not policy driven. Oil prices, for instance, are an important component of transportation costs; and movements in that price (in the paper, I use the relative price of oil as a proxy for transport costs) are largely outside of the hands of US policy-makers (“drill, baby, drill” enthusiasts notwithstanding).