Where does the Equant Analytics data come from?
Our trade data is collected from UN Comtrade, Eurostat, OECD, WTO, National statistics offices and IMF Direction of Trade statistics.
How do you make sense of all these data sources?
Goods data: We primarily use Comtrade and IMF sources for our goods data. These are the global sources of trade data and OECD, Eurostat and World Bank sources are very similar to Comtrade and IMF. We have tested each data source for its reliability and, because there are only marginal differences between the Comtrade, OECD, Eurostat and World Bank sources, opt to use Comtrade because it has the most comprehensive coverage.
Service data: We average OECD, Eurostat and United Nations data and present them in US Dollar values for consistency. The currency conversions are done using the OECD’s currency conversion matrix.
How often is the data collected?
The data is collected monthly. Even annual data for the last calendar year tops up during the course of the current year on a monthly basis so this process makes sure that the data is as current as possible.
How do you define sectors in your data?
We use the Customs and Excise HS Codes from 1996 onwards for our data. These have changed over time and we have re-classified as these have changed taking an average across the old and new classifications to ensure that all data is as accurate as possible.
We provide conversion matrices for comparison with SITC and SIC codes.
How do you cope with currency fluctuation in your data?
All our data is presented in US Dollars as this is how it is presented in the original source. This means that currency volatility is embedded in the data.
Do you report values and volumes?
We report trade values. We can provide volumes on request.
How lagged is the data?
Annual data: lagged by up to seven months but topped up with monthly data as it comes in. Comtrade does not provide full monthly data for the previous calendar year until July of the current year.
Monthly data: lagged by up to five weeks. The monthly data from Comtrade is lagged by four months but national statistical offices report data for the previous month in the current month. This, along with a momentum forecast, is used to provide data up to the end of the last calendar month.
The margin of error on a one month projection is 1%.
How do you derive your outlook for trade in the future?
We provide a momentum projection for trade flows. We are not forecasting based on assumptions about GDP, demand, prices, geopolitics or currency fluctuations. We use Gaussian probability distributions to predict the momentum of trade over a ten-year and three-year period (monthly and annually). This gives us both a long-term trajectory and short-term volatility in the projection.
This approach is not the only approach that could have been taken but has the advantage of simplicity and calculable accuracy in the near term (up to three months or the current year). As it is not based on assumptions about other variables, it does not include a risk that those assumptions, for example about the economic cycle, may be wrong.
What do you do where data is not reported or irregularly reported?
We use a mirroring technique derived from the OECD and World Trade Organisation. The process is standard process used by trade economists globally. The process is as follows:
- If country A has no data, then we will take the data for a particular flow from its partner, country B.
- If country A and country B both have data and the difference between them is 5% or less, we average the data
- If country A and country B both have data and the difference between them is 50% or less then we take an average of the data weighted in favour of the better reporting country (tested against historical data in that trade flow)
- If country A and country B both have data and the difference between them is more than 50% then we take the data for the better reporting country.
Two things make this process unique to Equant Analytics:
- We have applied the mirroring across all trade flows in the entire dataset (including services)
- The process we apply makes import values in a trade flow equal to export values.
Why do you need to conduct this mirroring process to refine the data?
Trade data is not regularly reported by all countries, it is often lagged and zeroes in the data are ambiguous. If one country reports zero trade with a particular partner or for a particular sector, this does not mean that there is no trade. The mirroring technique makes sure that if a zero is recorded, it is because there is no trade, not because it is poorly reported.
Why is the Equant Analytics data more comprehensive/accurate?
The mirroring process makes sure that as many gaps and irregularities in the data are available as possible. This makes it more reliable than other sources that do not engage in this process.
Why is the mirroring process not standard practice?
The sheer scale of processing power has made this process across an entire global dataset costly and ineffective up to this point. It is, however, standard practice – the OECD and the World Bank developed the mirroring on bi-lateral trade routes where data is poor in one country.
Which countries are particularly poor reporters?
Data is irregularly reported in:
- Middle East and North Africa
- Sub-Saharan Africa
- Many central European/ central Asian countries (Russia, Kazakhstan, Turkmenistan, Georgia, Afganistan for example)
Data is reported with a greater time lag in:
- Many Latin American Countries
- Some European countries (Greece, Malta, Cyprus, Spain, Ireland, for example)
What does this mean about how the data is interpreted?
The trade flows that we report are the best possible estimates for trade between one country and another. We cannot say that the data is perfect, but the mirroring process yields more data because, for example, a reliable data country, such as Germany, will be trading with less reliable countries and this will give us a value for that flow.
Why the data is still volatile (such as data for the MENA region): because the mirroring process still leaves some gaps in the data. We leave these as zeroes because the flow between two partners is not showing any results. Any estimation to smooth the trends would be making assumptions that there has been trade where this may not be the case.
How to interpret flows between two weak reporters: We only report the data that exists and so, for countries where reporting is generically weaker, we may be under-estimating values or not reporting them at all. For this reason, there is very little data on trade between MENA countries, for example, but there is data for trade between Sub-Saharan African countries and MENA (SSA tend to be better reporters than MENA).
What is the EADR?
The Equant Analytics Divergence Ratio (EADR) is the percentage difference by country between what is reported in the raw data and what we obtain through the mirroring process. It is a measure of a country’s reporting reliability.