European Grains Exports

Data

We use COMEXT monthly data for all non-Intra EU trade flows. We consider all EU countries reporting as EU members at a given time.¹ We focus on 3 products: Wheat (HS:1001), Maize (HS:1005) and Rice (HS:1006). However, the extra-EU exports flows are extremely limited for Rice (mainly nich markets for Camargue rice, or Po region rice for Risotto etc) and we will neglect this product in the following section. The EU maize exports follow also specific pattern, and are mainly directed to the Mediterranean region, with some increased activities going to China in the more recent periods. The period used in the statistical analysis covers January 2006 to August 2020. Looking at this large period allows us to identify some seasonal trend with more robustness.

The monthly, quaterly, country and regional data are availble in Excel at DATA_EU_MONTHLY_CEREALS.xlsx. The file includes data for all regions and is not restricted to the examples used in this note.

Regional level Analysis

Overall, Africa is the main destination of European wheat export (extra-EU), ranging from 70 percent to 97 percent of destination markets depending on the year. EU maize exports are lower and less concentrated towards Africa (from 20 to 80 percent market share destination depending on the year).

Recent evolution

Figure displays the last few months of EU exports to four destinations: Non-African countries, African countries (both North and South of Sahara) and the region South of Sahara. The upper panel shows the wheat market while the lower panel illustrates to maize market. Colors indicate the quarter of each month and we represent the 14 years estimated linear trend on the each graph. Each panel uses a homogenous scale across region allowing to quickly visualize the size of each market.

Monthly Exports by region

Comparison with the last three years

While we can see in Figure that monthly behavior is quite erratic, in particular for the last few months, some data points are above, and other below the trend. If we aggregate the number by quarter, we can get a simpler picture: Figure . However, the result seems counter-intuitive since the Covid-19 episod appears to be associated with increased exports from the EU to most markets. Still, we observe these results for Q1 and Q2, while Figure was showing a decrease for the two first months of Q3 (July & August).

Quaterly Exports compared to reference

The graph data are described in the table below:

Quaterly data

HS4	REGION	QUARTER	average 2017-2019	2020
1001	Non-Africa	1	1.896	4.293
1001	Non-Africa	2	1.855	3.575
1001	Africa	1	3.802	7.319
1001	Africa	2	3.729	5.532
1001	Sub-Saharan Africa	1	1.503	3.152
1001	Sub-Saharan Africa	2	1.551	2.288
1005	Non-Africa	1	0.556	1.319
1005	Non-Africa	2	0.606	0.816
1005	Africa	1	0.092	0.548
1005	Africa	2	0.137	0.120
1005	Sub-Saharan Africa	1	0.014	0.011
1005	Sub-Saharan Africa	2	0.001	0.041

Decomposition by component

Beyond the purely descriptive figures and tables of the previous section, it is useful to do a complete decomposition of the time series between the trend, the “seasonal”, here monthly, component and a residual. Studying the residual during the first months of 2020 will give us an estimation of the random shock associated with Covid-19. We use a standard multiplicative model and we estimate the monthly fixed effects using a log linearized model. The trend is estimated using a twelve months lagged average. Indeed, the common approach of using a centered mobile average will not allow us to estimate the random shock post-February i.e. six months before the last observation. This approach has some limitation since the trend is estimated in a purely statistical way without controlling for actual level of harvest in the previous year, either in the exporting or importing country. Still, this approach is more robust than the simple difference computed from the reference period as proposed above. Figure provides an overview of such decomposition for all European exports for wheat. We can identify the actual level of exports for each month over the last 14 years, the estimated trend, the seasonal component and the residual.

Overview of decomposition

Focusing on the monthly component in Figure , we can identify the export pattern of the EU. Wheat exports are mainly concentrated in March, April and September, and to a lower extent in February and October. Maize exports are more intense in October and November.

Monthly Effects

Still, the most important part of the analysis is to look at the random, or residual terms, for the first months of 2020. These results are provided in .

Shocks in 2020

We have drawn a line at “0”, i.e. the level where no positive or negative shocks are observed. The profile for wehat is particularly interesting since February and March are associated with a strong positive shocks on all markets, but in particular in Africa, South of the Sahara. The period April to June is associated with a level of trade consistent with pre COVID-19 expectations while July and August appear to be below average.

Key conclusions

We can provide a few explanations to this pattern:

1. 2019 was a particularly good year for European wheat production (estimated at 132 million MT, compared 115E6MT in 2018, and similar to the 2014-2015 production level). Therefore, the EU had relatively large level of inventories and a potential to export in 2020,

2. The negative forecasts for Ukraine and Russia, as well as in North Africa, have triggered some early purchases in 2020. This movement was accelerated in March with the threat of export restrictions. Therefore, several African countries may have decided to strengthen their inventories in Q1 and Q2, potentially incentivized to hoard some grains, and have adjusted their orders in Q3,

3. For Maize, like in China (see below), the EU may have replaced the US at the margin.

Country level Analysis

A systematic analysis for all African countries is not always possible since there are a relative large number of zeros (i.e. no imports for a given month) at the country level. A quaterly approach may be more promising but we will not explore this approach hereafter.

Still, focusing on some large countries, that have a more sustained trade activities, we could look at a few countries, and see in particular if the monthly pattern is mainly driven by the supply side (European stocks and harvest cycles) or to which extent the actual demand side, and the harvest pattern of the importers is important. In particular, looking at countries located in the Northern and Southern hemispheres could inform us. S

Selected countries and recent changes

As previously, we can look at the recent trend for some countries in . Let’s note that Algeria harvest wheat in Q3, and South Africa in Q4.². So, it appears are more limited during harvesting period, but starts in the next quarter (to target a given level of inventory?). Nigeria does not have a significant production of wheat - if any - and shows a pretty smooth level of imports over the various months. Algeria shows slightly highly level of imports in the first few months of this year, compared to the previous years, but a bit lower activity in the last two months. South Africa has a slightly similar patter, while Nigeria have relatively high level of imports this year. This overall behavior seems to be consistent with our previons conclusions: European exports have replaced less reliable exporters in Q1 and Q2 when a number of countries have strenghthened their inventories due to local weak harvest - actual or expected - e.g. Algeria, or doing actual hoarding. However, the Q3 has allowed to adjust inventories, while the EU stocks level are low and the wheat spring harvest is starting.

Monthly Exports by country

---

1. So, United Kingdom is included during some years but not in 2020. Since, this country is not a major exporter extra-EU of the commodities, it has limited impacts.↩︎

2. see hyperlinks to crop calendar↩︎