The Collapse in Relation to GDP
In an earlier post, I discussed the startling decline in US imports . Brad Setser has also reported on this phenomenon. This decline is not restricted to the United States, as noted in an OECD report released last week (h/t Torsten Slok):
Figure 3: Annualized q/q growth of GDP (black bars) and contribution of exports (red bars), and imports (green bars), all in percentage points. NBER recession dates shaded gray. Source: BEA, GDP advance release of 26 April 2009, and NBER.How Common Are Big Import and Export Comovements
The graph highlights the fact that the large negative contribution of the export collapse is roughly equal to the large positive contribution of the import collapse. At a recent conference, I was asked whether this was a common occurrence. My guess was that it wasn’t, but I thought I’d check. So here is a look at what (non-agricultural goods) exports and (non-oil goods) imports do over the longer haul.
Figure 4: Log non-agricultural goods exports (blue) and log goods imports ex-oil (red), in Ch.2000$ SAAR. NBER recession dates shaded gray. Source: BEA, GDP advance release of 26 April 2009, and NBER.Casual inspection of Figure 4 seems to suggest large comovement of imports and exports is not commonplace (this is somewhat different from the finding that HP-filtered imports and exports are procyclical, which pertains to “normal” fluctuations). Figure 5 provides a detail on the current and last recession, in growth rates.
Figure 5: Year-on-year growth rates of non-agricultural goods exports (blue) and of goods imports ex-oil (red), calculated as four quarter log differences. NBER recession dates shaded gray. Source: BEA, GDP advance release of 26 April 2009, NBER, and author’s calculations.I have two observations at this juncture.
- Comovements seem particularly pronounced in the previous and current recession.
- The current downturn is particularly sharp.
This latter point is illustrated by Figure 6, wherein I’ve showed the largest, correlated changes (negative changes in southwest quadrant; positive changes in northwest quadrant). What is interesting is that the previous large (q/q) changes were short-lived, and essentially reversed. The 2008Q4 and 2009Q1 are the only very large back-to-back changes I can find. (Unless the BEA’s guesses of what March imports and exports are really way off, then this is a remarkable outcome).
Figure 6: Quarter-on-quarter annualized growth rates of goods imports ex-oil (vertical axis) against non-agricultural goods exports (horizontal axis), calculated as log differences. Source: BEA, GDP advance release of 26 April 2009, NBER, and author’s calculations.Causes
So, we come to the question of what is causing this correlated and deep decrease in trade flows. A recent VoxEU post documented the decline in US and European trade flows, arguing that this decline is more likely associated with depressed economic activity and diminished access to credit, rather than to trade protectionism. I agree that thus far, this characterization seems correct. So, this leads to the other possibilities.
- Is it trade financing?
- Is it inventory decumulation?
- Is it vertical specialization?
(By the way, I don’t have a definitive answer; and these explanations are not mutually exclusive)
I think the downturn is in large part due to the lack of trade financing. But box 1.2 in the most recent OECD Economic Outlook Interim Report notes that it is difficult to explain the decline in trade growth using proxy measures for credit problems.
Figure 7: from OECD Economic Outlook Interim Report, Box 1.2. Two models are used to explain world trade. The first model is an indicator-based model that relates world trade to industrial production in the OECD countries and Brazil, China, India, Indonesia and the Russian Federation; export orders in the G7 countries (except Canada); the US tech pulse index; world semiconductor billings; and US credit standards. Monthly VAR models are used to forecast these variables over the very short term. The second model is based on the historical relationship between world trade growth, OECD GDP growth and US credit standards.The study used the US loan officer survey data as a proxy measure for credit availability; my guess is that this is a poor proxy, and that we canâ€™t rule out credit factors as an explanatory variable.
Is it inventory decumulation? Reader spencer suggests modeling imports as a function of lagged GDP, relative prices, and inventory changes (see comments to this post), I believe motivated by the view that big import changes are associated with inventory investment/disinvestment in this era of just-in-time supply management. Figure 8 suggests that there is a high correlation, especially in the post 1998 period. But the last two observations seem quite out of line with the previous observations, even post 1998. This suggests that inventory decumulation is not the only factor at work. It also suggests that a bounceback in imports is not in the cards, even if inventory accumulation resumes.
Figure 8: Quarterly change in goods imports ex-oil (vertical axis) against lagged inventory investment deflated by the GDP deflator (horizontal axis), in Ch.2000$. Blue circles, pre-1999; red circles, post-1998. Source: BEA, GDP advance release of 26 April 2009, and author’s calculations.Vertical specialization? As I have discussed in several previous posts  , much trade is now characterized by vertical specialization, where inputs are imported for incorporation in production in imports of exports. The decline in imports could then be driven by the decline in US exports, although it is hard to assess this channel since we do not have a good grasp of the extent of vertical specialization  .
Summing up, it seems to me that all three factors are likely in play. For me, though, the inventory channel does not seem to be important enough to imply a big bounceback in imports.
Originally published at Econbrowser and reproduced here with the author’s permission.