An Assessment of United Kingdom’s Trade With Developing Countries Under the Generalized System of Preferences

The European Union (EU) generalized system of preferences (GSP Scheme) grants preferential treatment to 88 eligible countries. There are, however, concerns that the restrictive features (such as rules of origin, low preference margin and low coverage) of the existing scheme indicate gravitation towards commercial trade agenda to which efficiency imperatives appear subordinated. Whether these concerns are genuine is an empirical question whose answer largely determines whether, after Brexit, the United Kingdom continues with the existing specifics of the EU scheme or develops a more inclusive United Kingdom-specific GSP framework. This study quantitatively examines the efficiency of the EU GSP as it relates to United Kingdom beneficiaries from 2014 to 2017. We draw on the descriptive efficiency estimation (the utilization rate, potential coverage rate and the utility rate) using import data across 88 beneficiary countries and agricultural products of the Harmonized System Code Chapter 1 to 24. Asides the Rules of Origin that, generally, harm the uptake of GSP, low preference margin is found to cause low utilization rates in a non-linear manner. Essentially, a more robust option (such that allows “global Cumulation” or broader product coverage) could, substantially, lower the existing barriers to trade and upsurge the efficiency of the GSP scheme.

The European Union (EU) generalised system of preferences (GSP Scheme) grants preferential treatment to 88 eligible countries. There are, however, concerns that the restrictive features (such as rules of origin, low preference margin and low coverage) of the existing scheme indicate gravitation towards commercial trade agenda to which efficiency imperatives appear subordinated. Whether these concerns are genuine is an empirical question whose answer largely determines whether, after Brexit, the United Kingdom continues with the existing specifics of the EU scheme or develops a more inclusive United Kingdom-specific GSP framework. This study quantitatively examines the efficiency of the EU GSP as it relates to United Kingdom beneficiaries from 2014 to 2017. We draw on the descriptive efficiency estimation (the utilisation rate, potential coverage rate and the utility rate) using import data across 88 beneficiary countries and agricultural products of the Harmonised System Code Chapter 1 to 24. Asides the Rules of Origin that, generally, harm the uptake of GSP, low preference margin is found to cause low utilisation rates in a non-linear manner.
Essentially, a more robust option (such that allows "global Cumulation" or broader product coverage) could, substantially, lower the existing barriers to trade and upsurge the efficiency of the GSP scheme.

Membership of the European Union (EU) Customs union allows the
United Kingdom to apply the EU tariff schedule to its imports (Holmes, Rollo, Dawar, & Mathis, 2016). This includes the application of preferential terms to exports of developing countries within the special and differential treatment (SDT) framework (Mendez-Parra, te Velde, & Kennan, 2016). Such "special and differential treatment," offered based on "non-reciprocity," specifically models the assertion that "treating unequals equally simply exacerbated inequalities" (UNCTAD, 2015). This was not the intention of the Generalised System of Preferences (GSP hereafter), which was an essential aspect of the SDT framework, and designed to integrate developing countries into the world trading system (Brenton, 2003). Brexit has meant that the United Kingdom designed its GSP scheme and given the paucity of time it is agreed that the existing EU GSP scheme will be rolled over to ensure continuity in the trading arrangement with developing and least developed countries.
There appears to be a consensus that unless GSP arrangements are put in place, the trade regime of the EU will no longer apply to imports into the United Kingdom immediately after Brexit (Jones, 2016;Stevens & Kennan, 2016). The easiest way to overcome the resultant market access challenge involves replicating the EU's GSP scheme in the short term (Hoekman, Rollo, Wilkinson, & Winters, 2016;Jones, 2016;Marx, 2018;Molinuevo, 2017). Such a route would allow the United Kingdom to stabilise its trade relationships with developing countries and situate effective transitional market access arrangements in the immediate post-Brexit period (Herman, 2016;Mendez-Parra et al., 2016;Ungphakorn, 2016). However, the 2014 EU's revision to the GSP scheme indicates an inclination towards commercial intents to which efficiency imperatives appear subordinated (Siles-Brügge, 2014). The reform made all high-income and upper-middle-income countries ineligible for GSP treatment while also made it easier for a beneficiary's competitive product sections to be graduated (De Ville and Orbie 2014;Siles-Brügge, 2014). The basis for reducing GSP beneficiaries from 176 to 88 and for introducing the so-called "graduation" mechanism is to keep with the EU's claim of being "the world's major development actor". Rather than representing a move towards improving the value of trade preferences for the "economies most in need," there are however reasons to argue that the revision theoretically provides the EU with more leverage to pursue its commercial interests (Damro, 2015;De Bièvre & Poletti, 2013;De Gucht & Barnier, 2012;De Ville, Orbie, & Relations, 2014;Langan & Price, 2016;Larik, 2015). Whether these concerns are genuine is an empirical question whose answer largely determines whether, after Brexit, the United Kingdom continues with the existing specifics of the EU scheme or develops a more inclusive United Kingdom-specific GSP framework.
This article, therefore, evaluates the United Kingdom's trade under the current EU GSP scheme to ascertain the utilisation of preferences by beneficiary developing countries. Secondly, identifies thematic recommendations applicable to the United Kingdom GSP stemming from the evaluation of the current system and relative efficiency of the different regimes-GSP, GSP+ and Everything But Arms (EBA). Broadly speaking, the recommendations are targeted at facilitating policy options towards designing a new United Kingdom GSP regime that is better at supporting development than the status quo.
The efficiency evaluation of the existing GSP scheme is based on estimations of the utilisation rates, utility rates and potential coverage.
The outlook of the efficiency indicators is matched to the four core features of the existing GSP scheme-preference margins, rules of origin, non-tariff barriers and product coverage. A dataset of 88 GSP beneficiary countries is used for the period 2014-2017 and exporter-products estimation is carried out at 6-digit level across HS Chapter 1 to 24.
The section that follows presents a review of the literature featuring efficiency, utilisation of the GSP scheme, the rules of origin, value of preference as well as preference utilisation. This is followed by data and methods in Section 3, while Section 4 discusses the main findings of the study. Section 5 makes valuable post-Brexit trade policy recommendations.

| LITERATURE REVIEW
The current study reviews past studies that assess the linkage between existing EU GSP scheme and the strategies adopted by the United Kingdom. This matters for the growth and strength of the economy, alongside other studies that have either assessed macroeconomic policy, the role of energy in the growth-emissions nexus (Adedoyin, Alola, & Bekun, 2020;Adedoyin & Zakari, 2020;Kirikkaleli, Adedoyin, & Bekun, 2020) as well as the importance of trade and FDI in an economy (Adedoyin, Bello, Isah, & Agabo, 2020;Udi, Adedoyin, & Sarkodie, 2020;Udi, Bekun, & Adedoyin, 2020).

| The efficiency and the value of existing EU GSP
Several studies highlight the development relevance of non-reciprocal preferences for developing countries and express concerns about preference erosion that could arise from the coexistence of "reciprocity." Theoretically, non-reciprocal trade preferences (such as the GSP scheme) are a vital element of protection patterns across the globe and play a crucial role in shaping trade opportunities for beneficiary developing countries. This is particularly the case for the 88 beneficiaries of EU generalised system of preferences (GSP). It is not surprising that the gravitation towards reciprocity raises concerns about the erosion of preferences and betrayal of "trade-development" nexus of the GSP scheme. Whether these concerns are genuine, and how they might be addressed is largely dependent on the "interpretation of reality" of these arrangements. Within this context, this study aims to estimate how much the existing EU preferences are utilised by exporters, how the utilisation rates diverge across eligible products and countries and situating the resultant divergence within the existing features of the scheme.
So far, the majority of empirical studies have assessed the efficiency of EU preferences based on the entire preferential regimes that exporters are eligible for (DeMaria, Drogue, & Matthews, 2008;Keck & Lendle, 2012;Wijayasiri, 2007). Keck and Lendle (2012) opine that this is more realistic than estimating the efficiency of a specific regime alone. However, this does not necessarily reflect the actual picture of each regime especially with the likelihood of putting both reciprocal and non-reciprocal arrangements in the same basket. This, indeed, overstates the efficiency rates even in the face of eroded preference margins or restrictive rules of origins (RoOs). The efficiency estimation must, therefore, be adapted specifically to each regime, given that the administrative requirements and the rules of origin vary from one preference regime to the other.
This study follows Zhou and Cuyvers (2012) which specifically examines the efficiency of the EU GSP scheme as it relates to ASEAN beneficiary countries, for the period 1990-2007. However, the study of Zhou and Cuyvers (2012) is dated especially that the 2011 Rules of Origin revision and the 2014 GSP reform remain unaccounted for. By adapting the efficiency estimation of United Kingdom GSP to this specific context and by covering all the beneficiary countries, this study aims to derive more general conclusions about the efficiency estimates and fill the gap in the literature.

| The utilisation of GSP scheme
A first approach to examining the economic value of a preferential regime is to assess the degree of utilisation (Cirera, Foliano, & Gasiorek, 2016). The GSP utilisation rate is the proportion of goods eligible for GSP treatment that use it. This is given as the ratio of imports that enter via a preferential arrangement and the eligible imports that could have benefitted from preferential treatment. (Hakobyan, 2015) suggests that low utilisation rates are evidence that trade preferences confer little or no economic value. It is posited by some others that the utilisation rate may not give the best perspective of the value of a preferential regime. According to Persson and Wilhelmsson (2016), the utilisation rate may be informative in some ways. For instance, low utilisation rates could reveal certain hidden costs (related to rules of origin requirements, administrative cost and other compliance costs) that come with requesting preferential treatment. These unobserved trade costs constitute Non-Tariff Barriers (NTBs) and unambiguously lowered export volumes while having ambiguous effects on the value of trade preferences.
Differently, low utilisation rates could indicate an insufficient difference between the preferential tariff and non-preferential tariff rates-preference margin (Nilsson, 2016) or it could indicate inadequate product coverage (Cirera & Cooke, 2015). On the basis thereof, the vast majority of the assessment of ineffectiveness of preferential arrangement emanates from the evidence of low utilisation rates. Table A1 shows empirical studies on the utilisation of trade preferences.

| Rules of origin
At a time when an increasing number of exporters are globalising their parts procurement and production networks, it becomes imperative to determine where goods originate (Donner Abreu, 2013). The economic nationality of imported goods must be ascertained and should be suitably linked to trade policy measures (Tsirekidze, 2017).
To achieve this, certain criteria such as rule of origin requirements are applied. The rules of origin require that products are deemed to have originated from a particular country if they are either wholly obtained in that country or sufficiently worked or processed there (Tsirekidze, 2017).
Products are "wholly" obtained in a particular beneficiary country if only the entire production is carried out in the same country (Krishna, 2005). In such a case, the smallest addition or input from any other country disqualifies such product from being "wholly obtained." This categorisation applies mainly to things that have natural occurrence and to products that are made entirely from them (Bombarda & Gamberoni, 2013). Nevertheless, originating status can still be obtained on the condition that the non-originating materials used have gone through "sufficient working or processing" (Tsirekidze, 2017).

| The value of preference margin
The most widely used measure of the value of trade preferences is the Value of Preference Margin (VPM). In most cases, the preference margin reflects a static welfare gain derived for using trade preferences. Against this backdrop, the value of preference margin is computed as the quantity exported multiply by the absolute difference between the "most favoured nation (MFN)" tariff and the preferential tariff given to preference beneficiaries.
One of the earliest evidence of the value of preference margin was by Yamazaki (1996). Yamazaki evaluated the effect of the "Uruguay Round Agreement on Agriculture" on the value of preferences.
As a percentage of imports, the author reported 9% preferential imports for the EU, 2% for Japan and 6% for the United States. The aggregate value of preferences across the three donor countries amounted to $1,853 million, in which the EU provided roughly 73% with Japan and the United States providing 14% with each. In the case of the EU, the benefits accrued to sugar was about 46%.
Recent studies have made a further attempt to measure the value of EU preferences. For instance, Candau and Jean (2009)  The estimation in Candau and Jean (2009) covered all products exported into the EU but not extended to those covered under preferential tariff rate quotas (such as bananas, beef and sugar) which accounted for a substantial share of rents.

| Preference utilisation and preference margin nexus
The probability of preference usage is positively related to preference margin (Keck & Lendle, 2012). There is an easy explanation to this; given that certain costs, such as rules of origin requirement and administrative costs, are linked to preference usage, trade preferences may only be used when exports are enough to generate significant duty savings. Using different methodologies, studies reveal that utilisation rates adjust in response to preference margins and volume of exports across different regimes (Alexandraki, 2004;Bureau, Chakir, & Gallezot, 2007;Candau, Fontagne, & Jean, 2004;DeMaria & Drogue, 2008;Inama, 2003;Manchin, 2006;Wijayasiri, 2007). A number of these studies specifically relate to either the United States or the EU market. For instance,   A similar study by Candau and Jean (2009) reports a preference utilisation rate well over 82% for 2001 product exports into the EU and higher for products with sufficient preferential margins. However, Hakobyan (2011) indirectly recognises the significance of fixed costs by illustrating a possible non-linear linkage between the preference margin and the utilisation rate. Specifically, Hakobyan examined a ceiling beyond which the positive linkage fades out. This finding is insightful and particularly corroborate Manchin (2006) which establishes that preference margin has a little or no effect on the size of preferential trade following the decision to request preferential treatments. The analysis of the scheme is carried out using the most current and detailed trade and tariff data obtained from Eurostat of the European Commission. Tabulation is used in presenting the extent to which eligible countries and products have utilised the United Kingdom GSP at the post-reform period. Throughout this analysis, it is acknowledged that the extent to which GSP preferences could be offered to developing countries is primarily dependent on the utilisation level. Additionally, it is identified that GSP utilisation level depends on other factors like preference margin, product coverage, rules of origin requirements and non-tariff barriers.

| Data description
An in-depth estimation of the GSP efficiency is conducted, bearing in mind the aforementioned factors which could hinder effective utilisation. Trade data by country and by products were matched with tariff data at 6-digit level. Precisely, the analysis is based on export flows towards the United Kingdom market of roughly 472 agricultural products of HS Chapter 1-24. The rationale for focusing on agricultural products is that substantial trade preferences are granted to developing countries for agricultural exports, while the EU applies a modest trade restriction to its nonagricultural imports. The trade data of the 88 beneficiaries are segmented into three in line with the variants of GSP scheme: 27 countries benefiting from the basic GSP, 13 countries for GSP+ and 48 beneficiaries of the EBA program.
To allow for simplicity in presentation, some of the results are aggregated across products and countries at the 2-digit level. The analysis used trade data for the 4 years after the scheme's reform.
Any analysis extended beyond this period could result in a significant data distortion because of changes in policy over the years. Hence, this article allows for an accurate evaluation of the influence of the modified regulation.
To achieve a good level of data consistency, the United Kingdom import data was used against the beneficiaries' export data. This is because the reliability and accessibility of the United Kingdom import data are far better than that of all the beneficiary countries under investigation. The overall consistency of the findings in this article is thus enhanced. Import data for the United Kingdom is taken from Eurostat which allows downloading data by preference eligibility and by import regime. There is no such information as to whether product imports entered via a preferential quota regime. It must be noted that exporting countries are faced with different preferential regimes (such as EPA, SADC, PAC and ESA) and offered by the United Kingdom, and as such preference regimes are found to overlap for many products.
The EU data only shows regimes for MFN = 0, MFN > 0, GSP = 0, GSP > 0, Preference = 0, Preference > 0 and Unknown. It then becomes difficult to identify how many preferential arrangements fall under the "Preference" category. Taking all preference regimes together could give the wrong impression that overall preference utilisation is high. As the focus of this article is on GSP scheme and its variants, this article used data that applies to only MFN = 0, MFN > 0, GSP = 0 and GSP > 0 by import regime and data for "only MFN" and "only GSP" by preference eligibility. Imports for which the regime used is "unknown" are disregarded.

| Estimation technique
This article provides tables in which the efficiency indicators are summarised, hence ignoring all imports that enter via non-dutiable MFN rate (MFN = 0) or imports that are not eligible for GSP preference. The presented tables show a summary of the obtained data based on the estimation of the utilisation rate, coverage rate and the utility rate for different subsets of the data. Table 1 shows utilisation rates for different agricultural products at the 2-digit level. As expected, we find that the utilisation rate increases with the coverage rate for most of the products. For some other products, utilisation rates remain high even for coverage rates below average. While the latter indicates the existence of huge potentials for coverage expansion, one cannot possibly observe any clear "threshold" beyond which additional coverage either stagnates or diminishes the utilisation rate.
We also aggregated import flows by HS section. Table 2 shows utilisation rates for each section. A noticeable pattern in all countries is that utilisation rates are high across agricultural products of section 4 (Prepared Foodstuffs, Beverages, etc.-HS Chapter 16-24).
However, one may expect that the utilisation rate is relatively lower for cocoa and cocoa preparations (HS Chapter 18) despite that it falls in the section 4 category. Partly, this is because the MFN tariffs on cocoa beans (raw or roasted) are currently set at 0% which reflects a low level of competition at the primary production stage. Also, developing countries which are moving up to the next stage of the production process (cocoa preparations) are faced with tariff escalation and restrictive rules of origin. Table 3 shows utilisation rates, coverage rates and utility rates for the largest exporters into the United Kingdom (by imports value), while  regimes. This is an aggregation across both country and product groups. For the estimation in

| Estimating GSP efficiency via descriptive efficiency indicators
A primary approach to estimating the effectiveness of the GSP is to assess the extent of utilisation. Until preferences are utilised, it is difficult to specify the definite value derived by beneficiaries. Largely, the empirical literature has employed descriptive indicators like the utilisation rate (UR), potential coverage rate (PCR) and the utility rate (U y R).

| Preference utilisation rate
Preference utilisation rate indicates the usage level of the existing preferential Scheme. A preference regime offers a lower tariff rate, relative to the MFN rate, to a beneficiary country for a specific product. This could be granted within a reciprocal arrangement (such as the Economic Partnership Agreements-EPAs) or a non-reciprocal preferential scheme (such as the GSP scheme). Imports are described as dutiable (MFN 6 ¼ 0) if they originate from a non-preference-receiving country. In some instances, dutiable imports could come from a preference-receiving country especially if certain products are not eligible for preferential treatment. As a rule, preferential regimes have exclusion lists and as such, do not accommodate all products from a preference-receiving country in all cases. A few times, the MFN rates for certain products are non-dutiable (MFN = 0) even in a no-preference situation. Because of this, this study classifies importer-exporterproduct level import flows into four categories as shown in Figure 1.
The utilisation rate of the GSP regime μ GSP is given as "the ratio of imports that enter via the GSP regime and the imports eligible for GSP treatments." Simply put, Import flows A and B are disregarded, as they are not within the "covered by GSP scheme" classification. Essentially, imports are described as eligible for GSP treatment if products from a beneficiary country are covered by the GSP scheme based on the tariff schedule.
In effect, MFN tariff T MFNz 6 ¼ 0. It, therefore, follows those products with MFN tariff T MFNz = zero are ignored. In most cases, preferential tariff T* xyz is mostly non-dutiable (T* xyz = 0) and if dutiable often takes a lower value than the MFN tariff T MFNz (that is T MFNz > T* xyz ). In such a case, the exporting country derives a welfare gain from a preferential arrangement because of the positive difference between T MFNz and T* xyz (known as Preference Margin).
Based on the foregoing, Equation (1) is modified as: where, M GSP (z,x) indicates imports that enter via the GSP regime.
Е GSP (z, x) represents total imports eligible for GSP treatments.
Equation (2) above defines preference utilisation for product z from exporting country x. The value of μ GSP (z, x) for an individual transaction is either 0 or 1. However, since data at the exporter-product level is an aggregation of various transactions, then μ GSP (z, x) takes a value between 0 and 1. To present a descriptive analysis, this study aggregates data at the importer-product level across import regime, GSP eligibility, HS chapters and exporters. Keck and Lendle (2012)  the same result for transaction-level data, there is a noticeable variation in the results for aggregated data. Nonetheless, the most suitable measure of preference utilisation for aggregated data (country group or product group) is the trade-weighted approach (Keck & Lendle, 2012).
The simple average approach shows a downswing biasedness as a result of modest usage of preferences in several small transactions. Similarly, certain product imports do not attract duties especially if the preferential regime is used and this tends to understate preference utilisation rate in the case of duty-weighted average approach. In this light, this study estimates product-exporter GSP utilisation using the Trade-weighted Average approach.

| Utilisation rate by trade-weighted average
This is given as the ratio of total preferential imports (in import value term) and the total preference eligible imports. By aggregating across the exporter-product level x and z, utilisation rate by trade-weighted average (URTWA) is illustrated thus; Equation (3) above represents the aggregated utilisation rate at the exporter-product level. The utilisation rate of GSP based on Equation (3) illustrates the share of eligible imports used by beneficiaries.
In a strict sense, this may be too restrictive, given that it reflects only actual exports. A wider approach takes account of product exports of GSP beneficiaries that are not covered or that are excluded by the scheme (often referred to as potential coverage). A higher percentage of such estimate indicates how completely generalised the GSP scheme appears for products.

| Potential coverage rate
The potential coverage rate is the ratio of imports eligible for GSP treatment and the total dutiable imports. At the individual transaction level, the potential coverage rate is calculated based on the following formula: where, Е GSP (z, x) represents GSP eligible imports.
M DUTY(z,x) is the dutiable imports.
By aggregating across the exporter-product level x and z, Equation (4) becomes: Abbreviation: GSP, generalised system of preferences.
Source: Author's computation based on Eurostat data.

| Utility rate
The utility rate offers and indicates how much of imports pay preferential rates compared to MFN rates. This is the ratio of imports receiving preferential treatment and total dutiable imports. At the product level, the Utility Rate is estimated using the formula below: where, M GSP (z, x) represents imports via the GSP scheme.

M DUTY(z,x) is the dutiable imports.
When Equation (6) is aggregated at the exporter-product level, it becomes:   Figure 2, and this provides a basis for low utilisation rates.
F I G U R E 1 Types of import flows. Source: Adapted from Keck and Lendle (2012) The absolute difference between the MFN rates and the GSP rates is termed "GSP Margins." In most cases, MFN tariffs are higher than the GSP tariffs, making the two graphs to deviate and reflect a gap. In the graphical analysis above, the two graphs are found to overlap each other across most of the imported HS1 and HS2 products (at 6-digit level). This status of no gap between the MFN tariff graph and the GSP tariff graph indicates a condition of insufficient GSP margins and explains the rationale behind low GSP utilisation rate. This finding is consistent with Keck and Lendle (2012) in the case of Canada (where the utilisation rate is only 17% for preference margins below 1 and 75% for margins above 1%). Although, the authors reported high utilisation rates for the United States and EU even with low preference margins.
Given that the MFN rates for products of HS1, HS2 and HS5 are set at zero (non-dutiable) in 2017, exporters are more inclined to export under MFN conditions without having to comply with rules of origin requirements and utilisation rate would naturally tend towards zero (as in Nilsson & Dotter, 2011).

Figure 3 ranks the United Kingdom's imports across HS chapter
1-24 based on top 10 products with highest utilisation rate. Similar to Dean and Wainio (2006), the 100% utilisation rate of HS 4 product goes with an exceptionally low potential coverage rate of 23.69%. This is informative and suggests that the efficiency of the GSP scheme should not be assessed by estimating only the utilisation rate in isolation of the coverage rate and the utility rate. Otherwise, certain products, for which there exist potentials to expand products coverage, would be ignored.
For instance, even with the utilisation rates above 97% for products of HS4, HS8, HS10, HS11 and HS17, the coverage rates are still below average. What could be responsible for these significant coverage gaps? In recent time, some EU GSP beneficiaries have been faced with a tariff increase in higher-value products as a consequence of their decision to increase processing capacity. As revealed by the 6-digit level analysis of HS 9 products (coffee, tea and spices) where unroasted green coffee beans attract no tariff under the United Kingdom GSP. However, roasted coffee beans attract a 2.6% ad valorem tariff and unroasted decaffeinated coffee attracts a tariff of 4.8%.
Even in the face of increased tariff for higher-value products, GSP tariff for same products remains below the MFN duties, but such tariff increase places a huge restriction on product coverage. The low utility rates of section 2 and 3 confirm the stringency of nontariff barriers discussed in Siles-Brügge (2014). to 58.47%. The exclusion of several "sensitive" products, as noted in Siles-Brügge (2014), further undermined preferential treatment and resulted in lower product coverage.
It is worth noting that, other factors, such as preference margin, rules of origin requirements, administrative costs (non-tariff measures) also influence the utilisation of preferences. Many of these are related to the existing conditions and the operational structure of the GSP scheme. So, even in the face of extensive product coverage, beneficiary countries may still not utilise the GSP scheme adequately (Gasiorek et al., 2010). This is reflected in the case of Vietnam whose GSP covered imports (£4,883,500,180) far exceeded that of Pakistan (£3,659,508,794). Yet, Pakistan recorded a greater value of "imports via GSP" (Vietnam-£ 2,934,445,526, Pakistan-£3,583,056,027). This is not to undermine the ability of products coverage extension in driving GSP utilisation but a direct call to suggest a policy mix around product coverage, eligibility criteria, rules of origin and supply-side initiative. Table 4 contains aggregate imports data for the United Kingdom from 88 beneficiary developing countries.
During the period 2014-2017, aggregate imports from 88 beneficiary countries were £252.37 billion. Of this figure, only £39.73 billion were eligible for GSP preferences. This suggests that only 6.35% of the total imports from developing countries were covered under the GSP scheme. By implication, the remaining 93.65% of the total products was imported under the MFN tariffs. The paltry 6.35% of the imports covered between 2014 and 2017, maybe traceable to several factors but largely to the revision of the scheme that became effective in 2014. As pointed out earlier in this article, such revision resulted in a downward review of GSP beneficiary list from 176 to 88 and graduation of certain sectors. As a result, the scope for preferential imports considerably reduced.
It may be insufficient to assess the effectiveness of the GSP scheme by merely looking at the share of total imports (MFN + GSP) covered by the scheme (Zhou & Cuyvers, 2012). While it seems logical to assume a direct relationship between product coverage and preference utilisation, it is much more important to assess the success of a preferential scheme by looking at how much available preferences are utilised. The utilisation rate indicates how well the inherent benefits of the GSP scheme are tapped by beneficiaries. It follows that more attention should be paid to actual imports via the GSP scheme. A look at Table 4 shows that a total £34.12 billion benefitted from the scheme out of the £39.73 billion imports covered. This figure represents a significant 85.88% utilisation rate of the United Kingdom GSP.
However, with 6.35% of imports covered by the GSP scheme of the total imports (MFN + GSP), there exists a tremendous potential to improve products coverage. Especially, considering that the potential in Persson (2015) and evidenced the direct relationship between preference margins and preference utilisation. As a result, fewer beneficiaries utilised the United Kingdom GSP, during the period of low preference margins.

The relationship between preference margins and the United
Kingdom GSP utilisation is further illustrated using a frequency distribution table as shown in Table 5.
The analysis in Table 5 shows the frequency distribution of 340 agricultural products eligible under each variant of the United Kingdom GSP. Of the 340 products analysed under the basic GSP, about 304 products come under preference margin of less than 5% (PM < 5%), 26 products under preference margin 5-10% (5% < PM ≤10%) and 10 products under preference margin 10-20% (10% < PM ≤20%). Out of the 304 products of "PM < 5%" category, about 138 (45.39%) products recorded zero rates of GSP utilisation, 92 products (30.26%) with a utilisation rate of less than 100% but greater than 0% and 74 products (24.34%) with 100% utilisation rate. This suggests that GSP beneficiaries took less advantage of the United Kingdom GSP scheme when preference margin was low (PM < 5%) by not using preferences at all for roughly 45.39% and utilising only 24.34% of the 304 products eligible under the scheme.
On the flip side, in the scenario of a relatively higher preference margin (10% < PM ≤20%), about 50% of the eligible products recorded 100% utilisation. This is higher than the 24.34% obtained in a low preference margin scenario. And directly refute the generalisation of Manchin (2006) that preferential margin does not affect the amount of preferential trade, once the decision to use a preferential arrangement has been made.
However, the analysis of preference margin-GSP utilisation relationship for EBA and GSP+ shows a widely divergent route. For instance, despite 10% < PM ≤20%, higher percentages ( and/or the bureaucratic requirements of claiming GSP preference. Table 6 shows the frequency distribution for rules of origin classification and how it relates to the utilisation rate. The frequency distribution table above shows the frequency of occurrence for the four "rules of origin" classifications, along with four ranges of the United Kingdom's GSP utilisation rate. The analysis illustrates the utilisation rates of 340 agricultural products based on RoOs classification. About 175 products come under the "Wholly" rule and this represents 51.47% of the total products analysed. It is not surprising that such percentage of agricultural products was imported under the "wholly" rule as this categorisation applies mainly to things that have natural occurrence and to products that are made entirely from them. We also find that roughly 109 (62.28%) of the total "wholly" imports showed significant utilisation rates of between 76 and 100%.
Understandably, the figures are less in the other categories. For instance, fewer products were imported under the "Specific" rule (115), "Any Heading" rule (41) and "% Value" (9). However, the "Specific" criterion recorded a higher percentage (73.04%), of the total "specific rule" imports, having utilisation rates of between 76 and 100%. This simply suggests that despite agricultural products having natural occurrence which adequately qualifies them as "wholly obtained." Developing countries are still attempting to increase their processing capacity and have become enthusiastic in processing agricultural products beyond their natural form before exporting. Quite frankly, the United Kingdom's GSP rules of origin make some provisions for higher-value products, but developing countries are required to satisfy the "sufficient working or processing" before originating can be obtained.
Essentially, RoOs could act as a "push factor" for developing countries to consider further processing of agricultural products both for local consumption and for exports, having obtained originating status via "sufficient working or processing" (Hoekman et al., 2016).
Meanwhile, stringent RoOs depress higher-value agricultural exports and subsequent utilisation. This finding specifically points out to one issue identified in Khanal (2011) that RoOs discourage the utilisation of GSP preferences. This article, therefore, presents an entirely different lens through which RoOs could be seen as posing a challenge to GSP utilisation.

| POST-BREXIT PROSPECTS AND CONCLUSION
The existing GSP arrangements have the potential to benefit preference recipients through diverse channels -Export growth, Preference margin, cumulation rules (Helpman, Melitz, and Rubinstein, 2008;Melitz, 2003;Panagariya, 2002). The potential benefits, however, are dependent on certain features of the existing scheme; First, RoO is found to impact on product exports of developing countries by increasing the cost of using the scheme (Bombarda & Gamberoni, 2013;Georges, 2010;Grinols & Silva, 2008;Jakob & Fiebiger, 2003). Second, the magnitude of product coverage in the beneficiary countries incentivizes preference utilisation especially for products that are classified as country's main exports Chaplin & Matthews, 2005;Manchin, 2006). Third, the extent of NTBs faced by developing countries specifically restricts the utilisation of preferential arrangements (Mohan, Khorana, & Choudhury, 2013).
The idea that the above-mentioned features may moderate the likelihood to utilise the GSP scheme is not entirely new; however, to the best of our knowledge, the United Kingdom's GSP has not been examined in a unified context of these features. We fill this gap.
The results in this article could lead us to an intuitive conclusion that GSP beneficiaries failed to utilise the preferences to the full In what follows, the present United Kingdom GSP has only favoured a relatively small cluster of developing countries that appear to be well integrated into the multilateral trading system. This is evidenced by the result in Table 3

| Streamline the existing rules of origin
It is important to point out that, rules of origin provide a good platform to subject product imports to checks to ascertain compliance with regulatory standards and to consider for appropriate tariffs.
However, excessively stringent rules of origin could be harmful to regional cooperation and restrain the scope to enhance developing countries' share of value addition. A handful number of developing countries depend on components and products that are not produced locally. When such parts are imported and added to the list of raw materials in production, products of developing countries might be disqualified as "wholly obtained" under rules of origin criteria. This, essentially, limit the ability to develop countries to process and export higher-value products, and in effect, restrict their exports of primary products. This description reflects the complex approach of the current EU scheme in which the regulation on GSP rules of origin includes different thresholds depending on the relevant HS chapter.
By streamlining the United Kingdom rules of origin criteria to allow for a robust Cumulation, materials from other countries can be combined, processed, and exported by developing countries without necessarily facing tariff escalation or penalty. The post-Brexit GSP scheme could encompass the proposals of developing countries in the Nairobi WTO ministerial declaration by adopting flexible and simple rules of origin that allow the Cumulation of products from any eligible country. The United Kingdom could even pursue a more robust option of allowing the LDCs to effectively abolish rules of origin by creating a "global Cumulation" classification. This would mainly relax the origin criterion to a maximum of 100% foreign content, lower the existing barriers to trade and increase the utilisation rate of the GSP scheme.

| Enhance the eligibility criteria for preferential market access under the United Kingdom GSP
The WTO rules specifically condition GSP eligibility on objective developmental criteria. However, there is no universally agreed-upon criterion for what makes a country developing, making it contentious to identify certain developing countries as most in need of GSP treatment. The World Bank classification of countries based on income groups has attracted huge criticisms on account that it does not adequately capture certain developmental metrics. This article does not aim to recommend a one-size-fits-all approach but rather suggests a flexible framework that can address a variety of needs. The United Kingdom can essentially design inclusive criteria that cover noneconomic indicators, whether it be social, socioeconomic, or environmental, to determine the eligibility of developing countries for GSP treatment.
It is recommended that such methodology captures a large set of indicators such as educational attainment, the tendency of natural disaster, level of gender equality and carbon embodiment. This way, the United Kingdom would come up with a future-oriented GSP scheme that addresses the needs of structurally weak, vulnerable, and small economies (SWVSEs). The adoption of such criteria would also enable a wider range of "countries most in need" to be granted GSP treatment than may have been excluded based on income classification alone. This offers an opportunity for the United Kingdom to pioneer a forward-thinking assessment for identifying countries most in need under the GSP scheme.

| Address supply chain constraints, regulatory institutions and economic governance in developing countries
The effectiveness of the GSP scheme can be strengthened by aid for trade, regulatory cooperation, trade facilitation and investment pro-