Friday, June 10, 2016

Forecasting the Great Trade Collapse

One sentence summary: International trade can be forecast by only using data on prices and quantities, without any formal estimation; an application using data on U.S. imports supports this methodology by successfully forecasting the Great Trade Collapse and the corresponding recovery period.

The corresponding paper by Hakan Yilmazkuday titled "Forecasting the Great Trade Collapse" has been published at International Economics...

The working paper version is available here.

This paper introduces a simple methodology to forecast international trade. The main innovation is to calculate non-unitary expenditure elasticities of import demand implied by non-homothetic preferences in the previous year to be further combined with the current change in expenditure to forecast the current imports. Using U.S. data on aggregate expenditure and good-level imports, we test the performance of the methodology in forecasting international imports. The methodology is successful in forecasting not only the Great Trade Collapse and the corresponding recovery period but also the other periods in the sample. 

Non-technical Summary
During the recent financial crisis of 2008-2009, the decline in international trade has been more than the decline in domestic expenditure/income, the so-called Great Trade Collapse. Figure 1 shows the corresponding experience of the U.S. economy where the decline in U.S. imports is more than three times the decline in U.S. expenditure.

This has been a surprise for the trade literature, since trade and expenditure/income are supposed to move together, mostly according to the unitary expenditure/income elasticity of demand implication based on constant elasticity of substitution (CES) preferences that are commonly used in the literature. This may cause a problem, especially while conducting policy, because the policy makers simply would like to know how trade responds to changes in overall economic activity.

Using data on U.S. imports (including information on both quantities and unit prices) covering the quarterly period over 2000q1-2012q4, we calculate expenditure elasticities of demand for U.S. imports at the good level. Second, for each period, calculated expenditure elasticities of the previous year (at the good level) are multiplied with the current annual percentage change in U.S. expenditure to obtain current (ex post) forecasted annual percentage change in U.S. imports (at the good level). Therefore, two pieces of information are enough for forecasting (the annual percentage change in international trade): the expenditure elasticity measures coming from the previous year and the annual percentage change in expenditure in the current period.

The good-level results show that our methodology is highly successful in forecasting the Great Trade Collapse and the corresponding recovery period for the U.S.:

Our methodology is also successful in forecasting the overall U.S. imports:

as well as periods of the Great Trade Collapse and the corresponding recovery:

Regarding implications for policy makers, it is important to emphasize that this paper has considered an ex post forecasting strategy rather than an ex ante one, where the expenditure elasticity of import demand coming from the previous year has been considered as the estimation/historical segment, while the annual percentage change in overall expenditure in the current period has been considered as the exogenous variable in the out-of-estimation sample. Although an ex ante forecast has not been the objective of this paper, it can easily be achieved by combining expenditure elasticities of demand with any forecast/expectation regarding the future economic activity.

No comments:

Post a Comment