The trade collapse of 2008-09 was characterized by a substantial overreaction of the trade in intermediate goods when compared to the trade in final goods. However, this overreaction was not the same for all intermediates and the initial drop and subsequent rebound was larger for more upstream inputs, that is, products further away from final demand. Figure 1 shows two intermediate goods that could both be used in the production of a car – a relatively upstream input –steel- and more downstream manufactured parts and components.  After the bankruptcy of Lehman Brothers in September 2008, indicated by the red vertical line, US imports of steel dropped much more dramatically than those of car parts.

In a recent EBRD working paper, I show that this observation can be explained by a supply-chain phenomenon called the bullwhip effect. The bullwhip is characterized by higher volatility of upstream industries relative to downstream industries when the final goods are subject to a demand shock. For example, when consumers suddenly stop buying cars, the sales and production volatility of their intermediate suppliers will be higher than their own. This phenomenon arises because of the way inventory adjustments take place along the supply chain.

Figure 1

Graphs depicting the development of US imports of car and components and the imports of steel.

Note:  Downstream depicts the development of US imports of car and components while Upstream depicts the imports of steel. Source: USITC and own calculations

So what is this bullwhip effect exactly? In a nutshell, the bullwhip effect assumes that firms, in order to avoid stock-out, adjust their inventory holdings in relation to the forecasted demand for their goods. When a huge demand shock – such as the one after the bankruptcy of Lehman Brothers – occurs, firms find themselves in a highly uncertain situation, which complicates estimating future demand for their product.The downstream firms revise their sales forecasts downward and decide to run down their inventories to a new desired level, thus reducing the orders to their suppliers. Due to the fact that the inventories are used to meet at least part of the demand, this reduction is disproportionate compared to the drop in sales of the final product. The suppliers in turn observe a much lower demand for their goods and they therefore also adjust their inventories and propagate the shock further upstream to their own suppliers, again in a magnified manner. After the final demand recovers, all supply chain participants not only produce to meet the new demand, but also to replenish their inventories to new higher levels, and the same type of mechanism unravels in the opposite direction. In extreme cases, this effect can even lead to an interruption of trade relationships.

However, under some circumstances the volatility of downstream suppliers might exceed that of upstream producers. A more upstream supplier is likely to be less specialized and involved in additional supply chains, while the more downstream product might be more specific to the final industry. To use again the example of car production, an upstream producer of steel will supply not only for the automobile industry, but also for a range of additional industries, such as machinery or construction. On the other hand, a downstream producer of car seats can only sell the products to the car industry. Therefore, when the demand for cars suddenly drops, the downstream producers will be fully exposed, while the impact can be mitigated for the more upstream industries which may keep selling to a range of other downstream producers. This result, however, will be conditional on the extent to which these other final industries are shocked as well. If the shocks are symmetric for all final industries, the volatility impact on the upstream industries will be largest regardless of the number of supply chains their products are used in.

My recent EBRD working paper shows, using monthly highly disaggregated US imports data, that the bullwhip effect has indeed been in place during the post-Lehman Brothers period. Most upstream imports faced increased volatility and a temporary loss of trading. However, in the majority of cases, trade resumed once the repercussions of the initial shock subsided. The results show that while the origin of trade did not matter for survival and recovery of a trading relationship, products that have been traded longer and whose market share was larger pre-crisis turned out to be more resistant to the bullwhip effect.

These results provide some interesting policy insights. They offer a possible explanation of why some countries’ trade might be more volatile than that of others and they establish a mechanism of international transmission of shocks. They also suggest that under some circumstances zero trade flows might occur in a subset of industries as a natural consequence of the production structure in which they operate. Such cases do not necessarily require a policy response. Some attention should be given to small and inexperienced exporters which seem more vulnerable during major demand shocks. However, unless the shock is protracted, a pro-active policy trying to help upstream producers might not be necessary.

A final question is whether bullwhip effects have any consequences for commodity prices. Commodities, being at the very upstream stages of production, should be most affected by major demand shocks. Unlike downstream products, however, their prices respond much faster to changes in demand. This might have some explanatory power as to why commodity prices plummeted at the time of the Great Trade Collapse.


Veronika Zavacka, 2012. “The bullwhip effect and the Great Trade Collapse,” EBRD Working Paper No. 148, European Bank for Reconstruction and Development, London.