Method of Analysis
Our analysis is based on a simple log-log regression model, which is often used to gauge relationships between the percentage changes of two measures, using historical price data of commodities futures traded on the Chicago Mercantile Exchange (CME) as well as annual exports for each country-crop pair, based on data from the USDA Foreign Agricultural Service. We selected the “Top 5” countries in pulling data from Gro’s web app, which are those countries with the highest levels of a metric, such as exports, over the time period chosen. We chose the 20 years from 1999 to 2019. We also selected a sampling of countries from the middle of the pack to reflect a range of areas that react to world prices rather than set them. Although we looked at five representative midtier countries, a similar analysis can be done for any country in the world.
While there is a huge disparity between the exports of the largest exporters and the exports of all other countries, price spikes have been shown to coax bushels out of the midtier producers.
We determined the percentage rise in CME prices that is associated with a 1% increase in a country’s exports. The smaller the price gain needed to prompt a rise in exports, the more responsive the country is to price signals, and the more willing it is to boost exports in the event of world shortages.
For example, Indonesia is less reactive to changes in CME prices than, for example, Turkey. A 1% rise in Indonesia’s wheat exports was associated with a 36% increase in CME futures prices, while a 1% export gain by Turkey was associated with a 29% price jump, our analysis showed.
The same association was found when CME prices fell, and countries’ exports decreased.
While our analysis showed only associations between prices and export volumes, we think it is reasonable to conclude in statistically significant cases that marketers in certain countries have altered their export behavior on the basis of price. Furthermore, we think it's likely that their reactions to higher prices will continue in the future.
The Top 5 countries were used except when interactions between commodities (e.g., corn production and soybean production) or between countries (e.g., Paraguay production and Brazilian production) may corrupt the results, in which case the Top 5 is adjusted. The two resulting categories are referred to as “major countries” and “minor countries.” The commodities investigated are corn, soybeans, and wheat, which constitute a large portion of trading interest on the CME and across the globe.
A caveat: CME contracts are used more widely by agribusinesses in the US than in other countries. This biases the results toward underestimating exports’ price effects for non-US shipments.
High positive effects indicate the country requires strong price rallies in order to increase exports, while small effects indicate the country’s exports are more responsive to changes in futures price.
What We Found
The countries we examined fell into three general categories.
Countries with a strong need to export a crop:
The first category consists of countries whose export quantities don’t change much regardless of whether world prices are high or low. These countries may routinely produce more than is needed for domestic consumption, and therefore export much of the crop. For corn, these countries include Argentina (for which a 1% rise in exports is associated with a 56% increase in CME prices) and South Africa (28%). For soybeans, they include Brazil (44%), Paraguay (65%), and the US (83%). For wheat, they include Canada (74%), Indonesia (a minor producer, 36%), and Turkey (minor, 29%).
In the case of Turkey, wheat production makes up just 2% to 3% of the world’s total, but the country consistently exports around a third of its crop. Such small producers can add up. Wheat production is more widely spread across countries than corn or soybean production, a factor that may help explain why there are weaker relationships between any individual country’s exports and the prices of wheat futures.
Countries that export only when the price is right:
The second of the three categories contains countries whose export levels are highly sensitive to world prices. A high enough price incentivizes them to route the crop away from domestic consumers and toward buyers abroad. In corn, these include Brazil (for which a 1% rise in exports is associated with an 11% increase in CME prices), India (a minor producer, 16%), and Australia (minor, 7%). Among minor soybean producers, only India (11%) exhibited a statistically significant relationship between CME futures prices and exports. Candidates in wheat included the US and Australia, but while both are major exporters, neither’s exports had a strong enough statistical relationship to CME prices to justify inclusion.
Countries with high production and domestic-consumption capacity:
The third category consists of countries that have high domestic-consumption capacity, allowing them to either ship tonnage abroad or route it to domestic users at home. The US is the sole claimant to this title in the corn market, as Argentina is in the wheat market. The soybean market has no stand-out country in this regard. These nations showed large, negative relationships between CME prices and each country’s exports for each commodity, meaning an increase in a commodity’s CME price led to a decrease in that commodity’s exports. This is likely achieved by the country exhibiting either large enough production to influence futures prices or healthy enough domestic usage, or both. The US data showed that a 1% increase in exports is associated with a 42% decline in CME corn prices, perhaps due to a combination of its domestic-consumption capacity and the influence of its crop size. In the case of Argentina, a wheat export gain of 1% is associated with a 29% price drop on the CME. Argentina’s negative relationship between CME prices and exports was unexpected as Argentine exports do not hold much influence over wheat futures, thus unobserved factors may be corrupting its results.
The Case of Argentina
Countries’ varied abilities to boost exports become especially important when there is a shock to world supplies. Argentina’s drought of 2017/18 was one such occasion. The country’s soybean and corn crops contracted by 31% and 22%, respectively, compared with the previous year.
Argentina typically provides about a tenth of the world’s exports of soybeans and a sixth of those of corn. But in 2017/18, soybean exports plummeted 70% from the previous year, while corn exports dropped 13%.
Futures prices on the CME rose sharply, and other countries responded with a combined 16% jump in soybean meal exports, to 33.5 million tonnes; a 9% increase in soybean oil exports to 4.0 million tonnes; and an 11% gain in whole-soybean shipments, to 146 million tonnes. In addition to standard suppliers like Brazil and Paraguay, minor exporters of soybeans entered the fray. Notably, Russia in 2017/18 increased soybean exports by 138% to 517,000 tonnes.
During Argentina’s historic 2017/18 drought, Russian soybean exports more than doubled.
It is not only important to price-forecasting that a particular country has a large crop-production program. It is just as critical to understand how the country’s domestic market is structured.
Understanding that some countries’ high domestic-consumption capacity allows for a more intuitive price-to-quantity relationship, while other countries’ low domestic-consumption capacity forces their crop to export with much lower sensitivity to price, helps a risk manager predict price. It does this by revealing that certain countries’ crops (e.g., Ukraine corn, US corn, and Indonesia wheat) will increase their contributions to global exports only when prices are attractive. This allows the risk manager to keep attention on insensitive exporters’ crop sizes regardless of price levels (e.g., for the case of Canada wheat, Argentine corn, US soybeans, and Paraguay soybeans) while remaining aware of the effect of price changes on exporting countries with a high price sensitivity. This type of analysis can be readily done with data available in Gro.