Official Statistics

Asymmetries in international trade in goods statistics: methodology notes

Published 23 May 2023

Overview

The asymmetries publication provides an overview of the asymmetry between UK-reported trade in goods with equivalent trade reported by its partner countries, where complete data is available for a given flow and time period.

Data sources

The data used in the compilation of the Asymmetries publication is extracted from the United Nations International Trade Statistics Database, also known as Comtrade data.

All countries must collect their own trade data and submit data to United Nations Comtrade on a monthly basis. After collating this data, United Nations Comtrade applies an average exchange rate on an annual basis to convert all countries’ data into United States Dollars.

For imports, Comtrade Statistics report the ‘country of origin’ and for exports the ‘country of last known destination’ or ‘country of dispatch’.

Countries Included

For each flow, we have chosen to only include data for partner countries that have submitted data for each of the three years covered in this report. This is so that asymmetry comparisons can be made between years. It also ensures that we are not including asymmetries calculated using incomplete data in our totals.

However, to ensure that we are capturing instances of misclassification between countries, which is an important component of asymmetry, we do include data where one country has failed to submit data for a particular chapter, so long as both countries have submitted at least some data each year for the affected flow.

What are trade asymmetries?

Trade asymmetries are the differences between the published trade statistics of a given country (the reporter) and its partner countries’ equivalent ‘mirror flows’. Each country collects their own data on international trade in goods. Theoretically, each reporting country’s exports to a given partner country should match said partner country’s imports from the reporting country, and vice versa. For instance, UK-reported imports from France should match France-reported exports to the UK. In reality, it is often the case that figures do not match exactly; we refer to these differences as trade asymmetries.

There are numerous reasons for asymmetries, including discrepancies in methodology, valuation, and partner country assignment.

How do we measure asymmetries?

There are numerous measurements of asymmetries. The most popular approach is to calculate the absolute and relative asymmetry of the reporter and partner countries’ trade data.

What are the different asymmetry measures and how are they calculated?

Asymmetry

The asymmetry is defined as:

This is the simplest measure of asymmetry, which calculates the difference between the trade statistical value recorded by the reporter and their partner country, known as the mirror value. This measure can be positive or negative depending on which country is greater. For the purposes of this report the UK is considered the declarant or reporting country, ‘D’, and each other country will be considered a partner country, ‘P’. Therefore, positive measures of asymmetry indicate that UK figures are higher.

Absolute asymmetry

The absolute asymmetry is defined as:

The absolute asymmetry is the difference between the trade statistical value recorded by the declarant and their partner country, known as the mirror value, irrespective of which is greater. This measure can only take positive values since it is independent of the direction of the asymmetry and instead focuses on the magnitude.

In the publication we have chosen absolute asymmetry to be our leading asymmetry measure because we are most interested in the magnitude of the asymmetry.

Share of total absolute asymmetry

The share of the total absolute asymmetry is defined as:

The share of the total absolute asymmetry measures the contribution of the absolute asymmetry for a particular category, for example a HS2 chapter, on the absolute total of the asymmetries within a certain flow and year.

Relative asymmetry

The relative asymmetry is defined as:

The relative asymmetry is a percentage which compares the difference between the trade statistical values of the reporting and partner countries to the average trade value of both the reporting and partner countries. This measure makes no assumption about whether the reporter or partner country trade value is correct.

Unlike absolute asymmetry, this measure can be positive or negative depending on which country is greater. If the reporting country has a positive relative asymmetry, this means that they reported more trade than their partner country. If a reporting country has a negative relative asymmetry, this means that they reported less trade than their partner country.

For example, the declarant country ‘D’, reports $100 billion in imports from their partner country ‘P’, while Country ‘P’ reports $120 billion in exports to Country ‘D’. The relative asymmetry here is -18.2%. This means that, relative to the average between the two countries, Country ‘P’ is 18.2% lower than Country ‘D’. In other words, Country ‘D’ has reported 9.1% less trade than the average and Country ‘P’ has reported 9.1% more trade than the average.

How is the absolute asymmetry calculated at different levels of aggregation?

When calculating absolute asymmetry by partner country, flow, year, HS2 chapter and aggregates thereof, the order of applying the absolute function and grouping or aggregating matters.

When calculating absolute asymmetry by country, we have applied the absolute function after grouping by partner country, flow, and year, but not by HS2 chapter. This enables focus on the overall magnitude of asymmetry between ‘mirror flows’, regardless of any classification differences.

Conversely, when calculating absolute asymmetry by HS2 chapter, we have applied the absolute function after grouping by HS2 chapter, flow, and year, but not by partner country. This enables focus on the overall magnitude of asymmetry for each chapter, regardless of any discrepancies regarding partner country assignment.

Lastly, when calculating absolute asymmetry by country and HS2 chapter, we have applied the absolute function after grouping by HS2 chapter, partner country, flow and year. These figures are only available in the detailed data set, for our expert users.

When providing total absolute asymmetry figures in this report, or absolute asymmetries for groups of countries, we have aggregated absolute asymmetries calculated at the partner country level for each flow and year, rather than the country and HS2 chapter level. This is because for the purposes of this report we are most interested in the total magnitude of asymmetry between ‘mirror flows’, rather than classification differences within a ‘mirror flow’. As a result, absolute asymmetry totals calculated from the HS2 chapter or partner country and HS2 chapter tables will not match.

Chapter 99 adjustments and omission from analysis

This quote is taken from the ‘Read Me First (Disclaimer)’ section of the UN Comtrade website:

The values of the reported detailed commodity data do not necessarily sum up to the total trade value for a given country data set. Due to confidentiality, countries may not report some of its detailed trade. This trade will – however – be included at the higher commodity level and in the total trade value. For instance, trade data not reported for a specific 6-digit HS code will be included in the total trade and may be included in the 2-digit HS chapter. Similar situations could occur for other commodity classifications. Detailed data processed after 1 January 2006 and published in HS will sum up to the respective totals due to the introduction of adjustment items with commodity code 9999 and 999999.

This indicates that for all post-2006 data, the total reported trade by a country should match the sum of the chapter totals reported by that country. This was not always the case in the data we downloaded, with some small differences. We followed the method indicated above and added all such differences to each country’s chapter 99 totals.