International trade has played a major role in the global recession especially in transmitting demand contractions across national boundaries (Freund 2009). According to standard trade-linkages reasoning, countries “catch” recessions from their major export partners while transmitting recessions to their major import suppliers.
This simple economic logic directs attention to a set of facts that has heretofore attracted little attention in the global debate.
- US exports are disproportionately sold to the EU and Canada; US imports are disproportionately sourced from China, Japan, and the rest of Asia.
- Imports from China dominated the US (real) import decline, while exports to Canada figured disproportionately in US export declines.
- US imports peaked in October 2007, with a secondary peak in October 2008, and a trough in February 2009. US exports, by contrast, continued to rise until much later in June 2008.
- The timing of US import peaks by sector is markedly different from that of US export peaks by sector. The export peaks cluster around the general peak in mid-2008. The import peaks, by contrast, both came much earlier (end of 2007) and show a great deal of sectoral dispersion.
- US imports related to housing construction – especially wood and construction equipment – showed relatively early downturns.
- The downturn in US imports of motor vehicles and parts has been particularly deep; it began in March 2007 when oil prices were still rising sharply.
These facts suggest that the transmission of the recession to the US (via declines in US exports) is mediated by a very different set of goods than the transmission from the US to other nations (via US imports).
Table 1. Changes in real US trade, not seasonally adjusted
Source: Authors’ calculations. For comparison with the disaggregated data, seasonally adjusted data show peaks for real imports in April 2008, real exports in August 2008, a trough for both exports and imports in May 2009, and a slight uptick in both series in June.
Detailed trade data analysis
Using a new dataset on disaggregated real trade and trade prices, we identify features of the current business cycle visible in US trade data.1 High-frequency trade data must be used with caution in an environment of price instability, such as the current recession, which was preceded by a sharp run-up of commodity prices followed by a sharp deflation. Thus, it is important to distinguish between nominal and real trade data (Francois and Woerz 2009; Ferrantino 2009).
The timing of US real import peaks by sector is markedly different from that of US real export peaks by sector. US export peaks tend to cluster around the general peak, reflecting the synchronisation of the global peak in GDP in mid-2008. US import peaks, by contrast, show a great deal of sectoral dispersion, with some sectors turning down much earlier than others. This suggests that the decline in US import demand, as well as its depth, was significantly influenced by particular sectoral weaknesses in the US economy, in particular the timing of the housing and oil bubbles.
These facts are summarised in Figure 1, in which each colour-coded bubble represents a sector. The horizontal location of the bubbles represent the timing of the peak, the vertical location represents the cumulative decline in real trade since the peak. Bubble size represents the dollar value of trade in 2008.
Figure 1. US trade decline by sector
Most notably, the demand for imported motor vehicles and parts peaked in March 2007, well before the peak in aggregate imports. This suggests some response to rising gasoline prices. Inputs into autos and/or construction, including aluminium, iron and steel, and plastics, also began to show real import declines in 2005 and early 2006. By contrast, US imports of computers and industrial machinery and crude and refined fossil fuels) did not decline until relatively late in the cycle.
The particular linkages through which global declines in demand are transmitted across countries are driven by pre-existing geographic and sectoral trading patterns, as well as by the particular structural weaknesses in each country and its trading partners. One would expect countries to “catch” recessions from the countries to which they disproportionately export and transmit recessions to the countries from which they disproportionately import. The pre-existing patterns of specialisation with these trading partners will in part determine the sectoral distribution of the transmission of the business cycle though trade. Overlaid on these patterns are contractions in the specific sectors for which disequilibria were particularly important in triggering the downturn and the upstream and downstream linkages of those sectors.
Geographically, US exports go disproportionately to the EU and Canada, while US imports come disproportionately from China, Japan, and the rest of Asia. More precisely, from 2006-2008, the EU and Canada accounted for 42.9% of US exports and 34.0% of US imports on a nominal basis, while Asia accounted for 35.5% of US imports and only 25.9% of US exports. China alone accounted for 16.1% of US imports and only 5.9% of US exports.
The transmission of the recession to and from the US is mediated by a very different set of goods. From July 2008 to the trough, we identify 13 categories of goods for which US real imports fell by $1 billion or more, and 10 categories of goods for which US real exports fell by $500 million or more. They are largely non-overlapping categories. For example, US import declines are dominated by electronics, crude oil, and other consumer goods, while US export declines are concentrated in intermediate and capital goods. There are large two-way declines in US trade in passenger cars, motor vehicle parts, and organic chemicals. Goods imported from China dominated US import declines, while goods exported to Canada figured disproportionately in US export declines.
Table 2. Largest real US import declines, July 2008-February 2009
Source: Authors’ calculations.
Table 3. Largest real US export declines, July 2008-January 2009
Source: Authors’ calculations.
Stories of the current cycle – construction, automotive, and petroleum
The collapse of US housing associated with the financial crisis shows up clearly in US construction imports, which began to decline much earlier than US imports in general and have fallen more deeply. US real imports of sawn or chipped wood, of the type used in construction, peaked in May 2005 and declined by 62.9% through May 2009. This peak is 29 months earlier than the general peak in US imports. The corresponding price series peaked earlier, in March 2005, and has declined by a cumulative 32.5% through May 2009. A simultaneous decline in prices and quantities is a clear indicator of a decline in import demand, induced by the declining demand for construction. Similarly, US real imports of equipment such as bulldozers, graders, and shovel loaders, which have multiple uses but are important for construction, peaked in May 2006, 19 months before the general peak, and declined by 81.5% in the subsequent three years.
Import prices and quantities for inputs related to housing turned down relatively early compared to more direct indicators of the state of the housing market, such as new housing starts (peaking in May 2005), housing units under construction (September 2005), the Case-Shiller Composite-20 home price index (July 2006), and prices of new one-family homes under construction (March 2007). Construction firms anticipated the bubble, restricting their purchases of imported wood inputs at least as early as their construction activity and well before the decline in housing prices. Since construction plays an important role in the business cycle (Leamer 2007), this suggests that import data on construction inputs may be an important tool in anticipating the business cycle. The impact of the decline in US construction on US trading partners was highly geographically concentrated. In a typical year, about 80% of US imports of sawn wood come from Canada, while about 50% of US imports of bulldozers and similar equipment come from Japan.
Figure 2. Prices of sawn wood imports and the Case-Shiller housing index
Price shifts for oil are often responsible for big swings in nominal trade data. These shifts were particularly sharp in the run-up to the current recession. Prices of US imports and exports of crude and refined mineral and petroleum products peaked in mid-2008 and bottomed early in 2009. In real terms, trade in mineral fuels has been relatively resilient, consistent inelastic demand. In the year ending in May 2009, US real exports of mineral fuels rose 47.6%, while US real imports declined only 10.3%. These represent different stages of the production process – US imports consist about half of crude petroleum and half of refined products, while US exports consist largely of refined products.
Real trade in motor vehicles and parts has decreased more sharply than trade as a whole. US real imports for the category have fallen by 55.1% since their peak in March 2007, while US real exports have fallen by 60.4% since June 2008. This fits in with the general pattern observed above – US imports were reduced by the increase in gasoline prices prior to the recession, while the decline in US exports corresponds to the general drop in global demand and the emerging crises of GM and Chrysler. US real imports of passenger cars have recovered slightly since February 2009. This is unlikely to be a seasonal effect, since prior to the recession the seasonal peak for US auto imports was usually in the fourth quarter and could possibly be related to anticipation of the “cash-for-clunkers” program under which US consumers received government financial incentives in the summer of 2009 to trade in low-mileage vehicles for more fuel-efficient ones.
Figure 3. Real US imports and exports of vehicles
Figure 4. Collapse of vehicle imports and exports, Feb 2008 – Feb 2009
We analyse the mechanism of transmission of the global recession to and from the US using real monthly merchandise trade data. The channels of transmission arise both from historical patterns of comparative advantage and peculiar features of the US and global recessions. The US's import decline was a major vector for transmitting the recession to Asia, as this drop in demand came well before the official recessions began in Asia. Peaks in US imports show a higher degree of sectoral dispersal than peaks of US exports, reflecting the particular patterns of weakness in the US, such as imports of inputs into construction and the relatively synchronous timing of the recession in US trading partners.
In particular, contractions of US imports of wood and construction equipment, used as inputs into housing, showed relatively early downturns. The downturn in US imports of motor vehicles and parts, which began when oil prices were still rising sharply, has been particularly deep but appears to have reversed in recent months. The downturn in US auto exports was related both to the global contraction and to the situation immediately preceding the bankruptcies of General Motors and Chrysler. By contrast, large nominal swings in exports of mineral fuels were largely price-induced, and mask relatively stable US demand for crude oil and foreign demand for US refined products.
Disclaimer: This piece reflects solely the views of the authors and does not reflect the views of the US International Trade Commission or any of its Commissioners.
Freund, C. (2009). “Demystifying the collapse in trade”, VoxEU.org, 3 July.
Francois, J. and J. Woerz (2009). “The big drop: Trade and the Great Recession”, VoxEU.org, 2 May.
Ferrantino, M. J. (2009). “The global trade contraction: How much is 2008-09 like 1929-33?”, USITC Executive Briefings on Trade, April.
Leamer, E. (2007). “Housing IS the Business Cycle”, NBER Working Paper 13428.
1 Specifically, we have applied the BLS indices for US export and import prices indices to the US trade data released by Census. Cf. Census' monthly release on real U.S. trade, aggregated to six end-use, and the multicountry aggregate real trade data produced by CPB.
2 Principal suppliers and markets are defined as the top two, plus any others with a 10% market share, in 2008.