Use of preferential tariffs by UK businesses for imports of goods: 2012 to 2023

• We estimated a two-year average of £118.7 billion of goods imports from the EU into Great Britain by UK businesses that were eligible for preferential tariffs for 2022 and 2023; of these, £107.4 billion used available lower preferential tariffs, with a preference utilisation rate (PUR) of 90%.

• £19.6 billion of goods imported into the UK from the set of non-EU countries analysed were eligible for preferential tariffs in the two-year average for 2022 and 2023; of these, £15.6 billion used available lower preferential tariffs, with a PUR of 80%.

• PURs differ across multiple factors, including business sizes, transportation routes, country of origin, and time since agreement entered into force.

• We created a detailed dataset on UK businesses imports of goods and tariffs using administrative data for 2012 to 2023; we analysed this data using a logistic regression model to investigate the relation between various factors and the probability of using a preferential tariff.

• We found a positive association between the odds of using preferential tariffs and potential duty savings, direct experience of importing under preference, and longer time since agreement entered into force (holding other factors constant and compared with reference categories).

• We estimate a strong negative association between using an indirect transportation route and the odds of using preferential tariffs; the estimated direction of the correlation between using an intermediary and the odds of using preferential tariffs varies according to type of intermediary used.

Preferential tariffs are reduced rates of customs duty that are negotiated with specific partner countries through preferential trade agreements (PTAs). They can reduce the cost of imports by allowing eligible goods to enter the UK at lower duty rates than the standard Most Favoured Nation (MFN) rate. MFN rates are the normal non-discriminatory tariffs that apply to imports from any World Trade Organisation (WTO) member country. 

Importers must make a formal claim to use the reduced rate of customs duty and must ensure that the goods can be treated as originating in the signatory country they are importing from. The information required should be made available by the exporter.

Utilisation can be expressed through preference utilisation rates (PURs). PURs are defined as the proportion of the value of imports that used preferential tariffs out of the total preference-eligible imports for a given agreement, product or other factors.

PURs can differ for several reasons, including:

• time since a PTA entered into force

• transaction value

• compliance and administration costs to the trader

• awareness of the PTA

• product-specific rules of origin

For more information, see the Department for Business and Trade's (DBT's) Preference utilisation of UK trade in goods publications.

Our analysis focuses on imports of goods by UK businesses and excludes imports by the public sector and households. We found that UK businesses do not always use available lower preferential tariffs when importing goods.

We used microdata to explore how the uptake of preferential tariffs differs across businesses characteristics and other factors, such as the length of time since the agreement entered into force and the transportation route.

We derived a detailed business import tariff dataset from:

• the International Trade Centre's Market Access Map database to extract tariffs information

• imports customs declarations from HM Revenue and Customs's (HMRC) International Trade in Goods data to identify UK businesses' imports 

• the Inter-Departmental Business Register (IDBR) to derive business characteristics

We identified preference-eligible imports as those where a preferential rate is available that is lower than the default MFN rate and excluded those where the MFN is equal to zero. Other preferences such as Generalised System of Preferences (GSP) and Developing Countries Trading Scheme (DCTS) are excluded. This means we only consider Free Trade Agreements (FTAs) in our analysis.

We conducted our analysis based on the country of origin of imports. The country of origin (COO) is the country in which the goods were originally produced or manufactured.

For imports from COOs outside of the EU, we analysed data for 25 trading partner countries from 2012 to 2023 (for the list of non-EU countries included in our analysis, see Section 6: Glossary). For a full list of trade agreements, see DBT's UK trade agreements in effect guidance.

Following Brexit, the EU-UK Trade and Cooperation agreement applies from 1 January 2021. This agreement sets out preferential agreements in several areas, including trade in goods and services. Custom declarations were introduced for imports from the EU into Great Britain on 1 January 2022. We used data for 2022 and 2023 for EU COOs.

We used information at IDBR enterprise or enterprise group level, depending on the business structure, to measure firm-level characteristics, such as employment and turnover. If an enterprise is part of an enterprise group, we assume the enterprise group makes the decision to trade. Otherwise, it is the enterprise that decides to trade.

We applied a logistic regression model to estimate the relationship between a set of explanatory variables and the probability that a preferential tariff is used on an eligible import, while controlling for other factors (like product and country of origin). This helps isolate the association of each variable to the outcome.

Our analysis builds on previous Office for National Statistics (ONS) Data Science Campus analysis on Employing data science to analyse the use of preferential tariffs in free trade agreements.

We are updating this previous work by increasing the number of countries in our analysis (including Australia, New Zealand and EU countries) and focusing on data from 2012 to 2023. We also added new variables, such as: 

• the transportation route

• business experience in using preferential tariffs

• the length of time since the agreement entered into force 

• use of an intermediary

Results are subject to change while we refine and improve our methodology and regression model and should be interpreted with caution.

All values are calculated using value terms (current prices) and are not adjusted for inflation.

We apply statistical disclosure control to published outputs.

Preference utilisation rate by UK businesses for imports of goods

The two-year average for imports of goods by businesses from the countries we analysed was £271.1 billion for 2022 to 2023. Of this, total preference-eligible imports were £138.3 billion.

The preference-eligible import two-year average for 2022 to 2023:

• from EU countries was £118.7 billion, with £107.4 billion in preferential imports and a preference utilisation rate (PUR) of 90%

• from non-EU countries was £19.6 billion, with £15.6 billion in preferential imports and a PUR of 80%

Preference-eligible trade from non-EU countries increased from a period average of £9.3 billion for 2018 to 2019 to a period average of £19.6 billion for 2022 to 2023. This increase might be influenced by multiple factors, such as:

• general increase in the value of imports from non-EU countries by UK businesses

• the addition of more countries in our data sets, as new agreements entered into force

• the introduction of custom declarations for imports from the EU in 2022.

Total UK businesses' imports from non-EU countries increased by around £90 billion between 2019 and 2022, according to our UK trade by business characteristics dataset. Free Trade Agreements (FTAs) were introduced with New Zealand and Australia on 31 May 2023. Imports sent from the EU but with a non-EU country of origin became observable in our dataset from 2022.

Figure 2 shows lower PURs for imports from non-EU COOs from 2020. Uncertainty around the coronavirus (COVID-19) pandemic and Brexit may have affected the rates. Businesses cited the end of the Brexit transition period and the pandemic as the main causes of importing challenges, according to data for 29December 2020 to 9 January 2022 from Waves 21 to 47 of our Business insights and impact on the UK economy dataset.

Preference utilisation rate by importer business size

We define four categories to explore the uptake of preference tariff by importer business size:

• micro – 0 to 9 employees

• small – 10 to 49 employees

• medium – 50 to 249 employees

• large – over 249 employees

PURs vary across business sizes. They are higher for EU countries for all business sizes and show smaller variation than non-EU countries.

We observed lower non-EU PURs for all business sizes from 2020, compared with 2019, with micro businesses especially affected (Figure 3). The two-year average for preference-eligible trade from non-EU COOs increased by around £10.3 billion between 2018 to 2019 and 2022 to 2023. During the same period, microbusinesses' eligible imports increased by around £1.4 billion, while their preferential imports increased by £0.9 billion.

Figure 3: Preference utilisation rates for imports from non-EU countries of origin were lower in 2020 than 2019 for all business sizes

Preference utilisation rates (PURs) for imports of goods from the EU and a set of non-EU countries of origin (COOs) by importer business size, 2012 to 2023

Notes:

1. It is not possible to assign all trade by businesses to a business size category because the number of employees cannot be determined from the Inter-Departmental Business Register (IDBR) in some instances.

Preference utilisation rate by transportation route

FTAs include provisions that generally allow goods to move through or be stored in countries not in the agreement, if they stay under customs supervision. However, the importer may be required to submit evidence of non-manipulation and to satisfy other requirements to claim preferential tariffs.

We explored whether the uptake of preferential tariffs varies across direct and indirect transportation routes.

For non-EU countries, direct transportation is where the country of origin (COO) is the same as the country of dispatch (COD). Indirect transportation is where COO and COD differ. For imports from the EU, indirect routes are where the COD is one of the non-EU countries in our analysis, given free movement of goods within the EU.

For both EU and non-EU COOs, indirect transportation had lower PURs than direct transportation in 2023 (Figure 4). PURs for indirect transportation were 53% for imports from non-EU countries and 47% from the EU. PURs for direct transportation were 95% for imports from non-EU countries and 91% from the EU.

On average, 69% of non-EU eligible trade was imported by direct transportation between 2022 and 2023. The value of eligible trade by indirect transportation increased substantially for non-EU countries in 2022 to 2023, compared with earlier available years. This is likely associated with the introduction of custom declarations for imports from the EU, as COO and COD became available in our dataset for imports sent from the EU.

Preference utilisation rate by country of origin and time since agreement entered into force

The top six non-EU countries in our analysis from which the UK imported goods in 2023, by value of imports (Figure 5), according to our UK trade: goods and services dataset, were:

• Norway

• Turkey

• Switzerland

• Japan 

• Canada

• Republic of Korea

Total import of goods from these countries was 25.1% of total UK goods imports in 2023.  

Continuity agreements with existing non-EU partner countries were adopted after Brexit, alongside new trade agreements. Continuity agreements were developed as new bilateral agreements that replicated, as far as possible, the effects of pre-Brexit existing trade agreements with partners. For example, when the UK was an EU Member State, its trading relationship with Turkey was governed by the EU-Turkey Customs Union, and agreements on agriculture, and coal and steel. This arrangement transitioned into an FTA post-Brexit (as described in the Department for Business and Trade's Trade with Turkey guidance).

PURs for UK businesses' imports of goods from Turkey, Norway and Switzerland are lower in 2021, compared with 2019 rates, while they slightly increased for Republic of Korea. PURs were 90% for Norway and 91% for Turkey in 2023. However, we saw a general increase of PURs from Japan and Canada, for which FTAs were available from 2019 and 2017, respectively. PURs were 72% for imports from Japan and 70% for imports from Canada in 2023.

We further investigated differences in uptake by time since agreement entered into force (Figure 6). An FTA "enters into force" (EIF) when it becomes legally binding and can be used by traders to import and export with signatory countries. Time since agreement is defined as the number of years between the date the agreement entered into force and the date of imports' acceptance by custom.

We defined five categories:

• agreement in force for less than a year

• agreement in force for more than a year, but less than two years

• agreement in force for more than two years, but less than three years

• agreement in force for more than three years, but less than five years

• agreement in force for at least five years

The category with the highest PUR is "at least five years" since EIF (89%), which is similar to what is shown in Figure 5. This category represents 88% of preference-eligible trade with non-EU countries in our dataset.

These estimates suggest a potential "learning by doing" effect, as businesses might use an agreement multiple times if they are in force for longer. We also looked at experience of using preferential tariffs to explore this mechanism further.

Preference utilisation rate by business experience in using preferential tariffs with non-EU countries of origin

To investigate whether PURs differ across experience of importing under preference, we defined three categories:

• no experience – UK business had not previously imported under preferential tariffs 

• indirect experience – UK business had imported under preferential tariffs but from a different COO 

• direct experience – UK business had imported under preferential tariffs from the same COO

The category with the highest PUR is direct experience (94%), which represents 58% of preference-eligible trade with non-EU countries in our dataset (Figure 6). This further suggests a potential "learning by doing" effect.

We cannot account for any experience before 2012, as we use data for 2012 to 2023.

We applied a logistic regression model to estimate the relationship between a set of explanatory variables (for example, business size) and the probability that a preferential tariff is used on an eligible import, while controlling for other factors (for example product, or country of origin).  

We ran the non-EU model on a set of different time periods:

• 2012 to 2023

• 2012 to 2019 

• 2022 to 2023

This was to explore potential differences in results influenced by the coronavirus (COVID-19) pandemic, the introduction of custom declarations, and Brexit. We found similar results for most variables included in the analysis. We found some differences in the estimated direction of the correlation of intermediary types and the interaction variable formed from indirect experience and business size.

Results discussed in this section are based on 2012 to 2023 regression results for non-EU countries and 2022 to 2023 regression results for EU countries.

The variables included in our model are:

• logarithm of potential duty savings

• transportation route

• business size

• logarithm of turnover

• ultimate control owner status

• custom procedure code (custom warehouse or direct imports)

• agent representation

• business experience using preferential tariffs

• length of time since the agreement entered into force (non-EU model)

• Harmonized System (HS) product section

• country of origin

• business regional location

• years

We use odds ratios (OR) to interpret our results. An OR is a measure of association between a characteristic and an outcome.

The models' performance statistics indicate that our regression model performs better for non-EU data compared with EU data. Further work is required to improve the performance of the EU model, such as accounting for previous trading experience with EU countries.

Our modelling approach treats observations as independent, while transactions can be clustered within businesses. This means that some businesses might be more likely to use preference tariffs for their imports.

Overall, our estimates suggest that important predictors of preferential tariffs use by businesses are:

• potential duty savings

• direct experience in using preference tariffs, compared with no experience

• transportation route

• length of time since the agreements entered into force

We discuss the results for some of the factors investigated in the following sections.

These results should be read with caution, as we are unable to account for other important factors that might relate to variables in our model and use of preferential tariffs.

Potential duty savings

Potential duty savings is the estimated value that could be saved on the import by using the preferential tariff instead of the Most Favoured Nation (MFN) rate. This represents the economic incentive for traders to use the FTA.

Potential duty savings is calculated as:

After controlling for other characteristics, we estimated that an increase in potential duty savings is positively correlated with the odds of an import using a preferential tariff.

Transportation route

We used a binary variable to indicate whether a good has been dispatched from a country that is not the country of origin (COO). In the case of imports with the EU, this is where the country of dispatch (COD) was outside the EU.

We estimated a strong negative association between the import entering the UK by an indirect route and the odds of using preference, compared with direct route (after controlling for other characteristics).

FTAs can allow goods to move through or be stored in countries not in the agreement, if they stay under customs supervision. However, failure to meet non-alteration or direct transport clauses in the agreements may influence the ability of the businesses to claim preferential tariffs.  

Experience and business size

To explore the combined influence of business size and experience, we defined an interaction variable between business size and business experience in importing under preference for the non-EU model.

We estimated a positive strong association between the UK business having imported under preference before with the same country for all business sizes and the odds of using a preference, compared with micro businesses that did not use preference previously (holding other factors constant). This suggests a potential link between experience of using an FTA and their later use.

For the EU model, we estimated a weak association between previous experience of using FTAs for trade with non-EU countries and use of preferential tariffs for imports from the EU, compared with not having imported under preference before.

Figure 8: Direct experience led to the highest odds of using preferential tariffs for all business sizes, compared with microbusinesses without experience

Logistic regression model results from non-EU countries of origin (COOs), 2012 to 2023

Notes:

1. Table 1 shows the results of our logistic regression analysis for non-EU COOs.
2. Decimal rounding can influence the distance between the point estimate and confidence interval bounds.

Time that the agreement has been in force

There is a positive correlation between later time brackets and the use of preferential tariffs, compared with agreements in place for less than a year (holding other factors constant). This might suggest a "learning by doing" effect, as businesses may be more likely to have previously used the agreement if it has been in force for longer.

Figure 9: Imports using agreements in force for at least five years had the highest odds of using preferential tariffs, compared with agreements in force for less than one year

Logistic regression model results from non-EU countries of origin (COOs), 2012 to 2023

Notes:

1. Table 1 shows the results of our logistic regression analysis for non-EU COOs.
2. Decimal rounding can influence the distance between the point estimate and confidence interval bounds.

Intermediary use and representation

When importing and exporting, businesses might appoint an agent to act on their behalf. Agents normally complete customs declarations on behalf of traders, sometimes alongside other services, such as paying tariffs or import duties. These roles are split differently by intermediaries. Intermediaries include: 

• customs brokers and agents 

• fast parcel operators (FPOs) 

• freight forwarders 

• warehouse operators 

• hauliers 

Importers must indicate if they are using an intermediary on the customs declaration and, if so, the type of intermediary used.

Types of representation include:

• direct agent representation – when the agent acts in the importers name, but the trader is fully liable

• indirect agent representation – when the agent acts in its own name but on behalf of the importer, with joint liability

To explore how agent representation might influence uptake of preferential tariffs, we defined a categorical variable with three categories:

• trader declaring – the importer submitted the custom declaration form themselves

• direct representation – the importer used a direct agent intermediary  

• indirect representation – the importer used an indirect agent intermediary

Our regression results suggest a positive association between the odds of using preference tariffs and direct representation, compared with trader declaring. However, we estimated a negative association between the odds of using preferential tariffs and indirect representation, compared with trader declaring (holding all other factors constant).

This result could be influenced by potential intermediary hesitation to be liable for the customs debt. However, importers must have a physical presence in the importing country to use direct representation, whereas indirect representation can be used by a business with no presence in the importing country.

Figure 10: Imports using indirect representation had the lowest odds of using preferential tariffs, compared with trader declaring

Logistic regression model results from non-EU countries of origin (COOs), 2012 to 2023

Notes:

1. Table 1 shows the results of our logistic regression analysis for non-EU COOs.
2. Decimal rounding can influence the distance between the point estimate and confidence interval bounds.

Country of origin and country of dispatch

The country of origin (COO) is the country where the goods were originally produced.

The country of dispatch (COD) is the country that sent the goods to the UK. This may not be the country of origin or manufacture. COD is the country where the last commercial transaction took place, where a change of ownership occurs on the way or where the sale occurs via an intermediary country.

Logistic regression

Logistic regression is a statistical model used to measure the correlation between explanatory variables in a set and the probability of an event occurring, in each case controlling for the effects of the other explanatory variables. In our analysis, the event is a business using preferential tariffs for import of goods in the UK when goods are eligible for preferential tariffs. 

Most Favoured Nation rates

Most Favoured Nation (MFN) rates are normal non-discriminatory tariffs charged on imports between World Trade Organisation (WTO) members. This excludes preferential tariffs under Free Trade Agreements and other schemes or tariffs charged inside quotas.

Non-EU countries of origin

The non-EU countries of origin included in our analysis are Australia, Botswana, Canada, Chile, Colombia, Ecuador, Egypt, Eswatini, Iceland, Israel, Japan, Kenya, Republic of Korea, Liechtenstein, Mexico, Morocco, Mozambique, Namibia, New Zealand, Norway, Peru, South Africa, Switzerland, Tunisia, and Turkey.

Odds ratio

An odds ratio (OR) is a measure of association between a characteristic and an outcome. It is calculated by exponentiating the log odds:

For categorical variables, odds ratios above 1 indicate that being in the category increases the odds of preference utilisation, compared with the reference category. Odds ratios below 1 indicate that being in the category decreases the odds of preference utilisation, compared with the reference category.

For continuous variables, the odds ratio gives the estimated partial correlation associated with a one-unit increase in the continuous variable on the odds of preference utilisation, while holding all other factors constant.

Preferential tariffs and preference-eligible imports

Preferential tariffs are reduced rates of customs duty offered under preferential trade agreements (PTA) between the UK and partner countries that are signatories to the agreements.

Preference-eligible imports are defined here as imports where a preferential rate is available and lower than the Most Favoured Nations rate.

Preference Utilisation Rates

Preference Utilisation Rates (PURs) are defined as the percentage value of goods imported under preferential tariffs by UK businesses out of the total value of preferential tariffs eligible imported goods by UK businesses.

We constructed a detailed business transaction tariff dataset linking UK import data with business and tariff information. This involved:

• linking customs declaration data to the Inter-Departmental Business Register (IDBR) to enrich the dataset with business-level characteristics

• identifying transactions eligible for preferential tariffs (excluding those under the Generalised System of Preferences and Developing Countries Trading Scheme) by linking to tariff data from the International Trade Centre's Market Access Map

We define preferential tariff-eligible imports as those where a preferential rate is available that is lower than the default Most Favoured Nation (MFN) rate and exclude those where MFN is equal to zero.

We also exclude:

• transactions with values of less than £135, as they are generally exempt from customs duties

• transactions not associated with relevant customs procedure codes; we retain only direct imports and customs warehousing for goods entering free circulation

• imports of monetary and non-monetary gold and diamonds

Business-level data 

To derive business characteristics, we reconstructed business structures using multiple IDBR table snapshots. This process involved: 

• re-establishing identity links across business units within the same business

• assigning characteristics (for example, turnover, ownership and employee count) at either the enterprise or enterprise group-level

• assigning a regional location based on the reporting unit with the largest number of employees

Business-level data limitations

The IDBR is not updated in real time, which can introduce time lags.

Pooling data across business units is complex and may introduce uncertainty, especially in regional analysis. 

Product code matching and tariff data 

Because of data availability: 

• before 2022, we used 10-digit product codes

• from 2022 onwards, we matched trade and tariff data at the 8-digit product code level and derived average ad valorem equivalents

Tariff data limitations

Tariff rates are estimated and may not reflect the exact duty due on a given transaction. 

We use ad valorem equivalents, which may not capture specific duties based on weight, volume, or quotas. 

We are unable to account for quota-based tariffs. 

Trade data

As we use custom declarations, we exclude imports of goods from the EU into Northern Ireland, as these are not subject to custom declarations under the Northern Ireland Protocol. 

For imports of goods by UK businesses where the country of origin (COO) is outside of the EU, we investigate a set number of countries. 

Methodological caveats 

Our method for identifying preferential tariff eligibility is less sophisticated than methods used by HM Revenue and Customs (HMRC) in their Import data by preference dataset. This is because of differences in data sources and matching methods. As a result, some transactions may be incorrectly classified as eligible or ineligible for preferential tariffs within a given year. 

More detail on data sources, linkage processes, and quality assurance is available in our The use of microdata for firm-level analysis of preference tariff utilisation in the UK: technical report.

Acknowledgements

This research was externally funded by the Department for Business and Trade (DBT). DBT colleagues helped draft this article.