Department of Foreign Affairs and Trade

• Ministers
• International Issues
• Trade Policy
• Country information
• About Australia
• Embassies
• Visas
• Home

• Travel advice
• Passports

Australia-China FTA Conference in Shenzhen

28-29 June 2006

Day 2: Rules of Origin

Rules of Origin

Associate Professor Donald McLaren, University of Melbourne

Peter Lloyd and Donald MacLaren, University of Melbourne

Address for Correspondence:
Department of Economics,
University of Melbourne,
Parkville, Vic. 3010,
Australia

Phone: 61 3 83445291
Fax: 61 3 83446899
Email: pjlloyd@unimelb.edu.au

• Introduction

Rules of Origin (ROOs) are a required feature of a China-Australia Free Trade Agreement (CAFTA), as they are for all free trade areas in which the member countries retain their own different systems of trade regulation.  Yet, little attention has been paid in the past to rules of origin in Australia and, indeed, in most parts of the world.  In a recent paper, Augier, Gasiorek and Tong (2005) state that
"Rules of origin are usually ignored for two reasons: they are dauntingly complex and at first sight appear mind-numbingly dull."
The technicality and complexity of the rules disguise their importance.

In fact, the tariff rate applied to any good entering a country is determined by

• the tariff classification of the good
• the rule of origin determining its origin
• the tariff rate for the good from this origin
• the value for duty of the good (unless duty is based on volume)

The first of these is settled by the Harmonised System of tariff classification that is now used by all members of the WTO, giving a common classification down to the 6-digit level.  The value for duty system is used for all imports and is governed by Article VII of GATT 1994 and the Agreement on Customs Valuation concluded in the Uruguay Round. 

Hence, the application of tariffs depends, practically speaking, on the tariff rates and the rules of origin.  Since the tariff rates on goods entering from a preferential source under a FTA will be lower for many goods than those applying to goods entering from a non-preferential source, the determination of the origin of an import may substantially affect the tariff rate actually applied and the competitiveness of these imports.   

Similarly, the application of measures other than tariffs that restrict imports of goods depends on the rule of origin jointly with other rules that determine the conditions of entry for goods coming from a particular origin.   This applies, for example, to quotas and tariff quotas. 

Thus, in a free trade area, ROOs together with tariff rates and the conditions of entry for non-tariff measures jointly determine whether goods enter on preferential or non-preferential terms.  Stringent ROOs take away some of the potential gains in an RTA resulting from the tariff preferences and improved market access that have been negotiated.  ROOs are therefore a most important part of the total set of rules.

A corollary of this proposition is that ROOs matter only for goods which have a strictly positive MFN tariff rate and goods where non-tariff measures restrict imports from MFN sources.  If goods enter free of tariffs and other restrictions, the entry of goods from all sources is unaffected by ROOs.  This has important consequences, as we see later.

Origin of goods is becoming increasingly difficult to determine in a globalised world as the processes of production are becoming more fragmented, particularly for manufactures and most particularly for product groups such as Clothing, Textiles and Footwear (CTF) and telecommunications equipment.  That is, the production processes that produce a traded good have been carried out in more than one country, upstream stages being carried out in one country and later downstream stages in one or more other countries.  Typically, in making manufactured goods, countries now specialise not in the production of final goods but in the production of stages of goods, using materials or components or other intermediate inputs that have been produced at an earlier stage in another country.  It is this fragmentation or dis-integration of production processes that makes the determination of origin difficult in many cases.  (For general discussions of this phenomenon, see Feenstra, 1998; Arndt and Kierzkowski, 2001; and Cheng and Kierzkowski, 2001.)  In some cases, the value added within the country from which a good is sourced may be small, certainly less than 50 per cent of the value at the point of exportation.

Australia fits this pattern.  Australian exporters of manufacturers have always used many imported inputs.  One feature of the Australian Customs Tariff has encouraged this, namely the system of by-law imports, or concessional imports as they are known today.  This system allows imported intermediate inputs to enter free of duty or at a low concessional rate if the imported good does not compete with an Australian-produced substitute good.  This has been a feature of the Australian tariff system since the first Tariff was introduced in 1902.  In the last two decades or so, Australian production of manufactures has become more limited both in terms of the range of goods produced and in terms of the stages carried out within the country for those goods it does produce.   One part of this trend is the increasing use of "offshore" production whereby an Australian manufacturer sets up an operation in another country as a foreign investor or joint venturer, or as a contractor.  The semi-finished article or part is then shipped to Australia where it is finished off for sale in domestic or export markets.  China is one of the countries where Australian "offshore" production takes place.

Chinese production processes for manufactures tend to be more integrated than Australian processes and indeed than the processes of most countries in the global economy.  This is because its manufacturing sector is large and diversified.  Yet, there has been a similar increase there in the proportion of imported intermediate inputs as the Chinese economy has opened up. 

The rules of origin in a CAFTA agreement must deal with these realities of fragmented and offshore production.   

• The basic problem

There is at present no set of rules or standard in the WTO which lays down preferential rules of origin.  The original GATT 1947 and the GATT 1994 which was embedded in the rules of the WTO contain no rules of origin for free trade areas or customs unions.  The Uruguay Round concluded an Agreement on Rules of Origin but this is confined to non-preferential rules.  It contains a work programme for negotiations to harmonise non-preferential rules of origin.  These negotiations have been going on in the WTO Committee on Rules of Origin for more than a decade but are a long way from complete (see Imagawa and Vermulst, 2005). 

Annex II of the Uruguay Round Agreement contains a common declaration on preferential rules but this is confined to a declaration of basic principles of transparency and administrative standards and procedures such as judicial review and non-retrospectivity, and it is non-binding.  Thus members of the WTO are free to adopt what rules of origin they wish in negotiating free trade areas. 

At present non-preferential ROOs are guided by the 1973 Kyoto Convention.  This was made under the auspices of the Customs Cooperation Council, the predecessor of the World Customs Organisation, but it too is non-binding.  The Kyoto Convention lays down several methods of assessing origin.  Under the convention, the first question is whether the product is wholly obtained or produced in one country, or whether two or more countries have been involved in the manufacture of the product. 

If a product is wholly obtained or produced in one country, it clearly has the origin of that country.  Origin has not proven controversial in this context. 

If, however, more than one country is involved in the manufacture of the product, the product admitted into a country from an FTA partner country will incorporate some material originating in a non-partner country, or non-originating material as it is called.  Here, the general concept is that the product will have the origin of the country where the last substantial transformation took place.  Beyond this principle, the rules are very general. 

There are three main methods that can be utilised to ascertain where the last substantial transformation took place:

• A percentage test
• A change in tariff classification (CTC) test
• A technical test

The CTC test is sometimes known, inaccurately, as a Change of Tariff Heading or CTH test.  The third test is used for a few product groups only or as a supplement to either of the first two tests.  The primary choice is, therefore, between a percentage test and a CTC test, or some combination of them.  Within each of these tests there are many choices. 

The percentage test comes in three forms:

• The import content method
• Domestic content (or regional value content) method
• Value of parts method

The first two methods should, in principle, lead to the same result as the import content and the domestic content are complements, summing to 100 per cent of the value.  One can, therefore, pose a test in terms of the maximum permissible import content or the minimum required domestic content.  In practice, however, there are many differences.  In particular, the domestic content method requires an analysis of production costs.  Production costs can be broken down into costs of manufacture and overhead costs, both manufacturing overheads and general overheads such as sales, outward freight and insurance.  One must include profits in the domestic content for equivalence of the two methods.  For the valuation of materials used in manufacture under the domestic content method, there is a question of at what stage materials and other expenditures should be valued, i.e. in ascending order: ex works, fob, cif or delivered (into factory). 

Similarly, for the CTC test, there are many choices. These relate to the choice of aggregation in the classification of goods which is used to determine the change in classification and whether the test is used alone or supplemented by a percentage criterion or a technical test.

In fact, many different ROO systems are used in RTAs around the world.  WTO (2002) and Estevadeordal and Suominen (2003) provide excellent surveys of the overall picture.  The WTO surveyed 93 RTAs that were in force in March 2001.  Of the three tests, the CTC was the most common.  89 of the 93 RTAs used a CTC test, compared to 77 that employed a percentage test and 74 that employed a Technical Test.  These three numbers add up to more than twice 93 because most RTAs use more than one test, and some employ all three for different product groups; eg PANEURO and NAFTA.  Of those RTAs which use a percentage test, 70 use an import content method, 67 use a value of parts method and only 9 (including CER) use a domestic content method.  

 In addition, all RTAs have general rules and most allow exceptions to some rules.  Those using a percentage test may have general rules relating to, for example, tolerance or cumulation or outward processing or drawback.  Those using a CTC may have de minimis or cumulation or absorption or drawback rules for example.  Some of these make the rules less stringent (e.g. tolerance or de minimis, outward processing and cumulation) and some make them more stringent (e.g. disallowing customs drawback) than they would otherwise be.

The outcome is a great variety of rules with little uniformity among RTAs.  Two ROO systems are widely used.  One is the PANEURO system used by the countries of the EU, the EFTA States, the Central and East European States and the Mediterranean States with which the EU has separate agreements.  This system applies only to trade between the areas as the EU itself is borderless.  It is a completely harmonised system; all of the rules are identical across the areas.  The other is the NAFTA system.  The US has used NAFTA ROOs as a template and imposed NAFTA-style rules of origin in most RTAs which it has negotiated post-NAFTA.  (There is, however, one notable exception; the US-Jordan Agreements use a percentage test. The US-Israel agreement also uses a percentage test but this was concluded before NAFTA.)   Canada and Mexico have done the same with their bilaterals.  (Estevadeordal and Suominen, 2003, pp. 11-12).  This has given a NAFTA family of ROO systems, with similar features though they are not harmonised. 

The NAFTA and PANEURO families are based on CTC but use other tests for many product groups.  The negotiations of a harmonised system of non-preferential ROOs are also based on CTC; the draft of the Agreement allows a percentage or technical test as a "supplementary criterion" only where the exclusive use of the HS nomenclature does not allow for the expression of substantial transformation (Imagawa and Vermulst, 2005, p. 619).   These features led Palmeter (1993, p. 329) to declare "CTH appears to be the wave of the future".  Nevertheless, a number of well-known RTAs use some form of percentage test as the primary test; these include the European Free Trade Area, AFTA, CER (currently) and the Andean Community.  Moreover, the description of NAFTA as CTC-based is misleading.  As noted above, ROOs do not matter when the applied MFN tariff rate is zero. When zero tariff rate items are removed from the list of products in the US Tariff, around 70 per cent of tariff items are subject to a percentage rule (Productivity Commission, 2004a, p. 33), and these include most of the heavily protected products in the CTF and automobile sectors.  The PANEURO system determines the country in which the "last working or processing" is carried out by product-specific rules which are a mixture of CTCs and percentage tests.  In the event of insufficient work and processing in one country, origin is allocated according to country with the highest percentage of value added.  Thus, percentage tests are still a major part of ROO systems around the world.

In addition to choosing a test or tests, there is a second problem faced today by countries in negotiating rules of origin.   Many countries now are members of more than one RTA.   In the case of the EU and the US this does not pose a problem for ROOs as they are big enough in terms of markets to impose their primary EU and NAFTA rules respectively on other countries seeking to negotiate an RTA with them.  Smaller countries, however, are not in this position.  For example, Australia has currently a sole percentage test in the CER Agreement with New Zealand and in the Singapore Australia Free Trade Agreement but when negotiating the Australia-US FTA it had to accept a variant of NAFTA ROOs.

• Australian experience with ROO systems.

 

Australia has negotiated ROOs as a part of the negotiations of each of the FTAs of which it is a member; the Closer Economic Relations Agreement with New Zealand (CER), the Singapore Australia Free Trade Agreement (SAFTA), the Thailand Australia Free Trade Agreement (TAFTA) and the Australia US Free Trade Agreement (AUSFTA).  All have different ROOs.  This section summarises the main features only of each of these systems.   For the full details of the method used, the Agreements and any related documents and rulings must be consulted.  The main sources are: for CER, the Joint Australian/New Zealand Information Booklet; for SAFTA, the Agreement itself; for TAFTA, Australian Customs Service Origin Manual, volume 8D; for AUSFTA, Australian Customs Service Origin Manual, volume 8C.

Table 1 sets out the main features of the ROOs in the four Agreements.  The features listed are the choice of primary test, the presence or otherwise of a supplementary test, and four other features which makes the rules less or more stringent.

In all cases the tests outlined in the Table relate to those products that are subject to a test because of the presence of non-originating materials.  For example, in CER, there are three categories of goods:

• Goods wholly the produce of the country (unmanufactured raw products)
• Goods wholly manufactured in the country
• Goods partly manufactured in the country

Goods in categories 1 and 2 enter without further conditions.  Goods in category 3 are subject to the percentage test.  That is, the features in Table 1 apply to category 3 goods only.  Very similar categories apply in the other Agreements.

For goods partly manufactured in one of the CER countries, the CER rules seem straightforward at first sight but they are not.  CER uses a percentage test as the sole test.1  These require

• The "last process of manufacture" be performed by a manufacturer in either Australia or new Zealand
• Not less than 50 per cent of the "factory cost" (including materials, labour and overhead) associated with the process must be "qualifying expenditures", i.e. incurred in Australia or New Zealand.

Both parts are troublesome.  The first requires a definition of manufacture and a determination of the last place of manufacture.  This poses problems when the production system uses contracting out or commission work.  The second requirement leaves out costs included in the country of manufacture but outside the factory.  There are additional problems with materials supplied free of charge or at reduced cost and materials and containers of mixed origin.  The Productivity Commission (2004a) reviewed this system in detail.

The ROOs in SAFTA are a variant of those in CER.  The Governments took the opportunity to refine the percentage test.  SAFTA provides a definition of "manufacture" whereas one is lacking in CER.   It introduced the concept of the "principal manufacturer" as
"…the person in the territory of a party who performs, or has performed on its behalf, the last process of manufacture of the goods."
This concept allows costs incurred by the principal manufacturer for work done outside the factory and on commission to be taken into consideration in assessing the percentage whereas they are not allowable under the CER definitions.  This extends to materials processed outside the territories of the two parties to the Agreement.  These features allow outsourcing expenditures to be counted as domestic content.  Outsourcing is a feature of many modern manufacturing operations.  As a high-wage country, Singapore outsources parts of its manufacturing processes to plants in neighbouring Indonesia and Malaysia with much lower wage costs.  All of the RTAs signed by Singapore contain outward processing provisions.

By contrast, TAFTA uses CTC as the primary method.    All lines require a change of classification, except for a few in which a technical test is substituted.  This CTC is, for some products, chapter-to-chapter (specified at the 2-digit level), or parts thereof.  For other products, heading-to-heading (specified at the 4-digit level) or sub-heading-to-sub-heading (the 6-digit level) or parts thereof are used or item-to-item.    In less than 20 per cent of the total lines, there is a regional value content test, with the required minimum percentages being 40 or 45 or 55 per cent.  In most cases, this is a supplementary test that makes the rule more difficult to satisfy.  The most important area of application of a supplementary regional value content test is in the Clothing, Footwear and Textiles and Base Metals product groups.  In a few cases the regional value content test is an alternative to a CTC test.  This makes the test less stringent.  There is a de minimis rule that allows a specified maximum (10 per cent) percentage of non-originating materials to be used without losing origin.  This also makes the test less stringent.

AUSFTA is the most recent of the RTAs that Australia has concluded.  It is another agreement which is a NAFTA derivative.  Most of the product-specific rules of origin require a CTC from the non-originating material to the traded good.  This CTC is chapter-to-chapter or heading-to-heading or sub-heading-to-sub-heading or item-to-item.   For a minority of goods, a specific rule of origin requires the good to meet an additional regional value content.  Either the build-up or the build-down method may be used in most cases, but some are restricted to the build-down method.  Where there is a choice of method, the percentage required for the build-up method is lower than that for the build-down method; for example, 35/45 per cent.  The percentage for build-up and build-down varies across tariff lines.  For a few goods, a regional value content test or a technical test is substituted for a CTC test.  There are special rules for clothing and textile products (including a yarn forward rule), and for automotive products.  There is a de minimis rule that allows a specified maximum (10 per cent) percentage of non-originating materials to be used without losing origin, though this is not available for some products.   Where applicable, this makes the test less stringent. 

Thus Australia currently has four different ROO systems.  Two use a percentage test as the primary test and the last two use a CTC test as the primary test.  There are differences between the two systems that are based on a percentage test and between the two that are based on a CTC test.  Comparing TAFTA and AUSFTA, the tests specified in many lines are the same but some are different.  All four agreements use some kind of percentage test as a primary or secondary test but the method of calculating the percentage, and the required percentage are different in all four cases.  Across all four systems, there are differences in other general features.  These differences among the four systems make it more difficult and costly for the Customs officials to administer the systems and increase the costs of compliance for the private traders. 

Since the negotiation of SAFTA, the Australian Government has been searching for an improved ROO system.  In December 2004, the Trade and Economic Ministers of the two CER countries announced a decision to replace the present percentage test by a CTC test.  On 3 February 2006, the Ministers announced that final agreement had been reached to adopt a CTC approach for the CER rules and it is anticipated that the changes will come into effect on 1 January 2007.  Currently Australia is negotiating four trade agreements - bilateral agreements with the United Arab Emirates, Malaysia and China, and an agreement covering ASEAN and Australia and New Zealand.  As a part of its consultation with the community, the Department of Foreign Affairs and Trade recently issued a Background Paper seeking views on the form of ROOs to be used in these agreements (DFAT, 2006). 

4.         The problem of Choice of ROOs

In this situation of non-uniform sets of rules, the obvious question is - which is the best set of rules?  Unfortunately, it is generally agreed that no one system is best.  For example, WTO (2002, para 12) concludes "In sum, there is no fully satisfactory methodology for origin determination, applicable to all products and serving all purposes."2  

Advantages and disadvantages of a ROO system

Each test is commonly seen as having advantages and disadvantages (see, for example, La Nasa, 1995; WTO, 2002, para 9-12; Lloyd, 2003; and Imagawa and Vermulst, 2005).  We list the advantages and disadvantages of the percentage and CTC tests.

When it is used as the primary test, the percentage test is not usually supplemented by another test.  That is, it is the only test.  Most ROO systems using the percentage test as the primary test use only one percentage rate.  This has the advantage that the test is consistent across tariff items in that it sets out a single minimum percentage of domestic value added that applies uniformly to all imports.  (This percentage varies among RTAs using a percentage test as the primary test from 30 to 70 per cent, the mode being 50 per cent; see Estevadeordal and Suominen, 2003, Table 3.)  As a sole test with a sole percentage rate, it cannot be manipulated to give a higher hurdle for some products than for others.  It is also the only one of the three main methods that easily allows cumulation across trading partners.  This advantage is cited by Augier, Gasiorek and Tong (2005) as the reason for preferring this test. 

Offsetting these advantages, a percentage test is more complex than the other two tests and, therefore, has higher compliance costs.  It has greater uncertainty for traders, partly because of the complexity and partly because, for a given production process, the domestic content or the import content varies with the exchange rate applied to imported non-originating materials and parts.  It provides a disincentive for greater efficiency in value adding in the domestic economy because greater efficiency lowers the domestic content and may, therefore, result in the loss of origin.

A CTC test has the advantages of being much simpler and easier to determine, provided there is no supplementary test.  It tends, therefore, to have lower compliance costs and greater certainty for traders. 

Offsetting these advantages, it has disadvantages.  First, "…the HS [Harmonised System of tariff classification] was not designed to serve the purpose of conferring origin; its basic aim is commodity classification and statistics.  A simple change in tariff heading may therefore not be an adequate measure for fulfilment of the substantial transformation requirement: conversely, certain substantial transformations may not entail a change in tariff heading." (WTO, 2002, para 9).  As a result, the extent of transformation or value added will vary from tariff item to tariff item, even if the rules uniformly use the same digit levels to determine a change in tariff heading.  In its Review of the CER rules, the Productivity Commission (2004a, p. xxiii) found
"Unlike the CER RoO, the CTC method does not treat industries or products uniformly.  This is because the extent of transformation involved in a change in tariff classification varies greatly between headings.  As a result, it provides inconsistent origin determinations across industry sectors and can produce distortions in trade."

Second, it has the disadvantage, from an economic point of view, that it is easily manipulated to give more protection to local producers.  By its nature, CTC is a product-specific rule.  In an RTA that uses CTC as the primary test, every line requiring its own CTC specification or some alternative test.  While a specified change of tariff classification by itself leaves no room for administrative discretion, the trade-restrictiveness of the rule may be increased by using a lower level of digit to determine a CTC for a particular product and/or combining a CTC with a tight supplementary percentage or technical test.  In this way it may protect either the domestic producer of a final product or producers of intermediate inputs used as originating inputs.

We can seek to quantify the effects of different ROO systems.  For example, a number of studies overseas have been done of the compliance costs of ROOs.  Compliance costs are important because they absorb real resources and because, as noted below, they restrict trade by reducing the incentive to apply for preferential access.  These studies indicate that compliance costs range from 1.5 per cent to 6 per cent of the value of the products traded (see Productivity Commission, 2004a, chapter 6).  The preference margin provides an upper bound estimate of the compliance cost as a percentage of the value of an import, as Goldfarb (2003, pp. 8-9) noted.  For an import that is eligible for a preference, an importer always has the option of avoiding the compliance cost by paying the MFN rate. 

The main focus has been on measuring the extent to which ROOs restrict the trade of the member countries.  Unfortunately, it has not been possible to test directly the effects of ROOs on trade between the member countries at the digit level of the tariff classification or the calculations of the percentages at which the rules actually operate The ROO systems are too complex to do this. 

A number of studies have adopted an index approach (see particularly, Estevadeordal and Suominen, 2003 and Productivity Commission, 2004a)3.  This consists of choosing the basic features of ROO systems, giving a qualitative assessment of each feature and then combining them in an index of the trade-restrictiveness of different ROO systems.  The most refined of these indices is that compiled by the Productivity Commission (2004a, Table A.2 and Figure 4.1).  The Productivity Commission compiled the index score for 18 ROOS, including CER and SAFTA as well as NAFTA, PANEURO and other important systems. 

The results show a considerable variation in the level of trade-restrictiveness across RTAs.  NAFTA has the highest index and SAFTA the lowest.  Another feature is that the lowest 7 scores are all achieved by systems which use a percentage test as the primary test; SAFTA, US-Jordan, Singapore-NZ, AFTA and US-Israel, the Andean Community and CER in ascending order.

Another form of testing trade-restrictiveness is the study of preference utilisation.  As noted above, an importer will choose the MFN rate rather than the preferential rate if the compliance costs are greater than the preference margin.  This indicates too that utilisation rates should be calculated only for tariff items with non-zero tariff rates as the rules of origin are irrelevant to items with zero tariffs.  For these items with non-zero tariff rates, non-utilisation will be higher the lower the MFN rate and the preference margins.

A number of studies have been done of utilisation rates in different RTAs and also in non-reciprocal preference schemes; for example, Goldfarb (2003) and Anson et al (2003) studied NAFTA; Augier, Gasiorek and Tong, 2005 studied a range of Mediterranean countries and three Central and Eastern European countries which have agreements with the EU; Manchin, 2005 studied the ACP countries’ trade with the EU.   Typically, these studies have found that preference utilisation rates are generally low.  For example, Goldfarb (2003, Table 1) finds that they are 55 per cent for Canadian exports to the US. As an extreme case, McKinsey (2003, p. 39), in a survey of intra-ASEAN imports in 2000, found that less than 5 per cent of intra-ASEAN trade entered under Common Effective Preferential Tariff rates.  We need also to know if there is any correlation or statistical relationship between "sector" utilisation rates and the margin of preference for the sectors.  (The margins equal the nominal MFN rates of protection if the preferential rates are zero.)  In an impressive study, Manchin (2005) shows that, for trade between ACP countries and the EU, the utilisation rate is positively and significantly related to the margin of preference, as predicted.  Her study covered both tariff and preferential quota arrangements.  We are currently trying to estimate the utilisation rate for imports into Australia under the four agreements Australia has signed but this work is not completed.

Manchin also shows that there exists a minimum or "threshold" value of tariff preference needed for traders to request preferences.  This rate is 4 percentage points.  If the preference margin is less than 4 percentage points, there is no incentive for traders to seek the preferential rate as the costs of obtaining the preferences are expected to be higher than the value of the preferences.  This shows that compliance costs have an important trade-deterrent effect, as well as absorbing resources.  For example, in the study of Mexican exports to the US, Anson et al (2003) found that average preference margin available to Mexican exporters were 4 per cent while the  compliance costs were 5 per cent : "As a result compliance costs have largely eroded the preferential access afforded by the PTA."  

One may note that the decrease in average levels of tariffs in Australia and most countries around the world in the last decade as a result of the Uruguay Round and of unilateral  changes in some countries has meant that compliance costs has risen relative to the margin of preference in RTAs generally.  Consequently, ROO are becoming more trade-restrictive for this reason alone.

These empirical studies produce strong evidence of the trade-restriction effect of ROO systems and, as a consequence, of the importance of ROOs for determining the real improvements in market access coming from the formation of RTAs.  

To get a more precise appreciation of how we can devise better ROO systems, we need to consider principles of design of these systems. We should ask what are the purposes or objectives of the ROO system?  We should device a ROO system for an RTA in order to advance these purpose(s) or objective(s).  Unfortunately, this does not lead far.  Rarely is a purpose declared; for example, none of the four ROO systems examined in Table 1 declare a purpose or objective.4  However, accompanying documents make it clear that ROOs are intended to prevent trade deflection, that is, the deliberate routing of traded goods through a member country in order to qualify for a lower preferential tariff or better preferential access.  Hence, they are an anti-circumvention device and nothing more.

Accepting ROOs as an anti-circumvention measure, we can take as our objective that a ROO system within some RTA should be operated so that, while preventing trade deflection, it should maximise the benefits to the member countries of the formation of the RTA.  The economics literature on the benefits of a discriminatory regional liberalisation of trade in goods is ambiguous in one respect.  There is a category of trade within discriminatory trade agreements called "trade diversion".  This is the diversion of imports into one member country that previously were sourced from a country outside the agreement to another member country within the agreement as a result of the preferences now given to this country.  However, such diversion is not necessarily harmful to the country whose import trade is diverted because production and consumption effects may offset the higher cost of the imports.  This ambiguity is an example of the economics of the Second Best.  A second best world is one in which some distortions remain after reform.  The removal of one distortion does not necessarily improve welfare of the parties to the free trade are agreement because it creates new distortions. 

There is a presumption that regional trade is beneficial to the member countries. (For a recent discussion of this theme, see Lloyd and MacLaren, 2004).  This does not deny that there may still be some products were trade is diverted and this diversion is harmful to the importing country.  Trade diversion is, however, a smaller risk today as a result of the Australian Government steadily lowering tariff rates in the last two decades.  The average tariff rate on all imports in 2003-04 was 3. 5 per cent and, of more relevance to ROO analysis, that on dutiable goods only (i.e. excluding imports where there are no duties paid) was 9.5 per cent.5  In the case of China, there is little risk of trade diversion as China is the least-cost supplier of a very wide range of goods.    Moreover, in agreeing to a preference during the negotiations of an agreement, the member countries’ Governments have declared that they wish to promote trade in these items.  Hence, one can interpret the objective of ROOs as one of maximising trade between the members, subject to the prevention of trade deflection.  The best ROO system is, therefore, that which is least trade-restrictive, provided it prevents trade deflection.

Recently a literature has arisen which looks at ROOs as an instrument of protectionist policy rather than as a technical problem of determining origin.  Pressure from protectionist groups has, in some RTAs, led to particularly trade-restrictive rules for some product groups (see Productivity Commission 2004a, Box 8.3 for examples).  Estevadeordal and Suominen (2003) find that generally around the world the product-specific ROOs are most restrictive for agricultural goods and for the CTF group.  As an example, the case of NAFTA rules is often cited.   The more restrictive CTC and selective percentage tests in NAFTA apply to tariff items protected by higher MFN rates in the US. (See Estevadeordal and Suominen, 2003 and Productivity Commission, 2004b, chapter 4.1).6  Very stringent supplementary percentage or technical tests are used for some products.  Textiles and automobiles, among the most heavily protected of product groups, have especially highly trade-restrictive product-specific rules.  These trade-restrictive rules maintain some of the protection for these groups.  This is form of hidden protection. This pattern is not surprising as the gains to domestic producers from this form of protection are the greater the greater the margin of protection.

ROOs may divert FDI flows as well as trade flows.  FDI may be diverted to a country which is a member of an RTA in order to take advantage of preferential market access for exports from the FDI host and member country to another member; for example, foreign investors have chosen to relocate some of their international production to Mexico in order to gain preferential access to the lucrative US market.  Jensen-Moran (1995) provides examples from the rules in force in the EU and NAFTA, mainly from high-tech industries.  In the case of FDI, diversion is generally beneficial to the member countries as they gain additional FDI. 

Looking at a ROO system as a design problem

To pursue this objective of maximising the welfare gains from a free trade area agreement, we need to consider aspects of the design of a ROO system.  How can we design a ROO system that has minimal trade-restriction effects and other desirable properties such as low compliance costs?   Unfortunately, there is little literature to guide us here.

The main concern should be with the trade-restrictiveness of a potential ROO system.  The trade restriction effect is the most important effect of a ROO system.  Substantial trade restriction denies the purpose of offering preferential access and the benefits of this access.

One design possibility is the establishment of common external tariffs.  For example, there is a substantial movement in Canada in support of a common NAFTA external tariff and common measures for other trade barriers against third country countries as a way of avoiding the costs of NAFTA ROOs (see Goldfarb, 2003 and references therein.)  But this is a big step that the members of a Free Trade Area have already chosen to avoid in negotiating a free trade area agreement rather than a customs union and they are, therefore, unlikely to revise this decision.

There are a number of features of the design of a ROO system which may mitigate the trade-restriction effects.

Stephenson (1997) suggested that an RTA be required to opt for a single test only, except where necessary for technical reasons.  This would rule out the use of supplementary tests in CTC-based systems that account for much of the trade-restrictiveness of these systems.

Another design possibility is that of a waiver when the tariff rates on a tariff item are the same or close between the members of a Free Trade Agreement.  This is, in effect, a partial customs union type solution.  A waiver has been proposed for NAFTA by Hufbauer and Vega-Canovas (2003) and independently for CER by the Productivity Commission (2004a, chapter 8.5).  Hufbauer and Vega-Canovas proposed a waiver when the difference was no greater than 1 per cent whereas the Productivity Commission proposed a maximum difference of 5 per cent.  The proposal is made on the presumption that the costs of transhipment would make it most unlikely that goods would be deflected when the MFN tariff difference is small.  This proposal can be applied to any ROO system: NAFTA is a system with CTC as the primary test and CER is a system with a percentage as the sole test.  The difference tolerated should be somewhere between 0 and 5 per cent.

This proposal would, at a stroke, simplify any ROO system.  The extent of the saving of compliance costs and the liberalisation of trade will depend on the number of items with identical or near-identical positive tariff rates.  (There is no saving if the rate is zero.)   In the case of CER, the Productivity Commission (2004a, chapter 6) estimated that 44 per cent of tariff items have a rate which is strictly positive in at least one country and the rates differ by 5 percentage points or less.  This percentage is particularly high in the CER case because both countries have low tariffs and admit many goods free of duty.

Another design question relates to drawback of customs tariffs.   In a free trade area goods exported from one member country to another will often contain non-originating materials and parts that have been imported from a third country and on which duty would ordinarily be payable.  The question arises whether drawback of such duties should be allowed for trade within a free trade area.  Some ROO systems allow drawback on intra-area trade and some do not (e.g. PANEURO).  Allowing drawback makes the ROO system less trade-restricting.  However, it is sometimes argued that drawback is not efficient in these circumstances because it gives a producer exporting to another member country and using drawback on inputs an advantage over two groups of producers of the same product: the competing producers in the same exporting country who sell on the home market rather than export the product, and the producers in the importing country who similarly sell on their home market, neither of whom can use drawback.  This argument is true as far as it goes but it is incomplete.   Disallowing drawback for intra-area trade would correct this anomaly but it would create new anomalies.  For example, the exporter who exports to the member country would then be at a disadvantage vis-à-vis the exporter who exported to a third country where drawback is allowed.  It is a strange outcome for a free trade area to treat exports within the area less advantageously than exports to destinations outside the area!  (Some of these arguments are presented in Productivity Commission, 2004a, pp. 95-98 and Driessen and Graafsma, 1999).

The essence of the problem is that we are again in a Second Best world.  Hence, one cannot state that in general the removal of drawback is necessarily better or worse for the countries.  However, there is a presumption in favour of drawback.  It moves prices in these economies into line with those faced by their competitors in the world as a whole.  Area suppliers are, therefore, able to compete equally within the area and in third country markets.

Another general feature is that dealing with cumulation.  Cumulation adds areas where location qualifies for origin.  "Non-originating" now means not originating in the cumulation area.  Cumulation is a feature of a ROO system which, in designated circumstances, allows exports to qualify as originating when they otherwise would not under the rules.  It is, therefore, another feature which reduces the trade-restrictiveness of an RTA.  We note that cumulation provisions can apply both to goods that are wholly produced in the free trade area and those that are partly manufactured in the area.   They can apply both to ROO systems that use CTC as the primary test and those that use a percentage test as the primary test, though the modification of the test is more straightforward in the case of ROO systems that use a percentage test as the primary test (see WTO, 2002, paras 14-16, 31-34 and Annex 1).

Bilateral cumulation treats inputs originating in an FTA partner country as domestic content for the exporting member country.  This is not controversial.  ROOs today routinely allow bilateral accumulation. 

This raises the question as to how far cumulation provisions can be extended, that is, the definition of the cumulation area.  Diagonal cumulation extends the area in which production may originate to a country or countries outside the RTA.  Many countries now are a member of more than one RTA.  There is little debate in extending the area when the two (or more) countries that are members of one RTA both (or all) have free trade area agreements with a third country (or countries).  In this case, the logic of free trade suggests that any product that is traded within the RTA and originates in the area which is the union of these sets of countries should be counted as originating in the exporting member country.  The most important example by far of diagonal cumulation is the PANEURO agreement among the countries of the EU and other States with which the EU has separate agreements.  (Some of the countries in the original 1997 Agreement are now full members of EU-25).  A similar case could arise in a free trade area with the PRC as it might argue that the relevant territory for the purposes of defining origin is "All China" (the PRC plus Hong Kong).  The case for more extended cumulation is less clear cut when only one (or a subset) of the countries has a link to an outside country.  Full cumulation occurs when a group of countries are all linked to each other by a network of agreements and the cumulation area is the whole area covered by the countries in two (or more) separate RTAs.  (A full cumulation system may differ from diagonal cumulation in other respects too.)  The problem of defining the extent of the cumulation area will become much more important and much more complex as the links between free trade areas are constantly expanding with the formation of new areas.

Beyond these general features, we have to look at the design of the primary test and other features of a particular ROO system. 

Consider first, an RTA which has a percentage test as the primary test.  To start with, it should have a single percentage in order to avoid arbitrary differences in the percentage required across tariff items.  Although economic theory does not indicate any percentage rate that results in a "substantial" transformation, this rate should be less than 50 per cent and as low as possible.  Then the most important feature is the choice of imported or domestic content.  It is generally agreed that determination of origin is much simpler if an import content approach is used.  "In general, the import content seems preferable over the domestic content test because it is easier to apply and leaves the importing country administrators with less discretion.  This is because it is simply based on the process paid by the producer, which can be easily checked through invoices, thereby obviating the need for complete cost of production calculations and allocations." (Imagawa and Vermulst, 2005, p. 607).

Any ROO system with a percentage test will still, however, have an incentive at the margin, for some goods, to substitute domestic materials and parts for non-originating materials and parts.  And it is not possible to eliminate the uncertainties associated with changes in exchange rates and foreign prices.  These features are inherent in a ROO system based on a percentage test, though they may be mitigated by a lower percentage rates and a tolerance provision.   

Consider now, an RTA which has a CTC test as the primary test.   It should avoid a supplementary or replacement test, except where it is necessary to use one because of the nature of the transformation in the tariff item.   Apart from this, it should avoid a selection of a low level of tariff classification change in order to protect domestic manufacturers.  Unfortunately, it is not possible to insist on a single level, say the 4-digit Heading level, because the HS system was not devised for this purpose and the choice of any single level would leave anomalies. 

Thus, product specificity and the danger of manipulation in a CTC-based system in order to restrict trade will remain.  Any ROO system based on CTC will involve arbitrary choices and a variable transformation rule.  It must also remain open to manipulation to restrict trade, especially when the MFN and, consequently the preferences margins, are high.  These features are inherent in a system based on a CTC test.

5.         Conclusion

ROO systems vary greatly among RTAs in their principal features.  As a consequence, they vary greatly in their trade-restrictiveness and in their effects on compliance costs, trade uncertainty and trade patterns. 

The most important of these effects is trade-restrictiveness.  All ROO systems are trade-restrictive to some extent.  They have to be in order to prevent trade deflection.  However, most ROO systems are much more trade-restrictive then they need to be.

Any ROO system with a single percentage test will have a uniform transformation rule.  The lower the percentage rate the less trade-restrictive the system and the lower the compliance costs.  The system will still, however, have an incentive at the margin, for some goods, to substitute domestic materials and parts for non-originating materials and parts.  And it is not possible to eliminate the uncertainties associated with changes in exchange rates and foreign prices.  These features are inherent in a ROO system based on a percentage test.  Product specificity and the danger of manipulation are inherent in a system based on a CTC test.

It is possible to design ROO systems with minimal trade-restrictiveness.  This holds both for those with a CTC as the primary test and those with a percentage test as the primary test.  In this sense, the choice of CTC or a percentage test as the primary test is not the most crucial choice. 

What is most crucial is that the details of the ROO system for an RTA be chosen in order to minimise trade-restrictiveness and, as far as possible, other untoward effects on the costs of compliance, uncertainty and the pattern of trade.   These details include the details of the primary test.  They also include general features.  The selection of features such as tolerance and de minimis provisions, cumulation and a waiver along the lines recommended by the Productivity Commission and Hufbauer and Vega-Canovas can greatly reduce trade-restrictiveness and compliance costs.

FOOTNOTES

1.  The architects of the Agreement chose to follow the criteria that had been used in the previous New Zealand Australia Free Trade Agreement.  The two main features of these rules - the "last process of manufacture" rule and the percentage of factory cost - can in turn be traced back to earlier preferential trade agreements with Great Britain and Canada (see Steele and Moulis, 1994).

2.  Lloyd (1993) provides a solution.  The basic problem arises because origin in these rules is treated as an all-or-nothing property, as if a good originates in one country, whereas the production is actually spread over two or more countries.  His solution is a tariff that can be levied differentially on the value added by each country in the production chain, depending on whether it is in or out of the area qualifying for preferential treatment.  In the context of free trade areas, this solution would require all member countries to adopt this form of tariff.  See also Augier, Gasiorek and Tong (2005).   

3.   Another approach is to use a gravity model, e.g. Estevadeordal and Suominen (2003) and Augier, Gasiorek and Tong (2005).  However, this too must proceed at a level of aggregation well above that at which the actual rules apply.  Tests based on gravity models can give only an indication of the trade-restriction effect.

4.   For TAFTA, the Australian customs Service Manual (current, Division 4) states "ROO are necessary to provide objective criteria for determining whether or not imported goods are eligible for the preferential rates of duty available under TAFTA."  In almost the same language, the Guide to determining the origin of goods for AUSFTA states "We need rules to provide objective criteria for determining whether or not goods are eligible for the benefit of preferential rates of duty that are provided under the Free Trade Agreement".  These statements describe what ROOs do, not what objective they serve.

5.   From the Australian Bureau of Statistics time series of customs duty collected supplied to the authors.  This average combines duties paid at both MFN and Preferential rates.   

6.   According to Palmeter (1993, p. 329), "no ascertainable rule or principle was employed in determining what level of change would be required for any particular product".
REFERENCES

Anson, J., O. Cadot, J. de Melo, A. Estevadeordal, A. Suwa-Eisenmann and B, Tumurchudur (2003), "Rules of Origin in North-South Preferential Trading Arrangements with an Application to NAFTA", Centre for Economic Policy Research Discussion Paper No. 4166.

Arndt, S. W. and H. Kierzkowski (eds.) (2001), Fragmentation: New Production Patterns in the World Economy, Oxford University Press, Oxford.

Augier, P., M. Gasiorek and C. L. Tong (2005), "The Impact of Rules of Origin on Trade Flows", Economic Policy, July, 567-624.

Australian Customs Service (current), Manual - Origin, Australian Customs Service available on line at www.customs.gov.au.

Cheng, L. K. and H. Kierzkowski (eds.) (2001), Global Production and Trade in East Asia, Oxford University Press, Oxford.

Department of Foreign Affairs and Trade (DFAT) (2006), Rules of Origin and Australia’s Free Trade Agreement Negotiations. A Background Paper on Issues for Consideration, DFAT, available on line at www.dfat.gov.au.

Driessen, B. and F. Graafsma (1999), "The EC’s Wonderland: An Overview of the Pan-European Origin Protocols", Journal of World Trade, 33, 19-45.

Estevadeordal, A.  and K. Suominen (2003), "Rules of Origin: A World Map" (Preliminary draft),  Inter-American Development Bank, Washington, D. C.

Feenstra, R. C. (1998), "Integration of Trade and Disintegration of Production in the Global Economy", Journal of Economic Perspectives, 12, 31-50.

Golfarb, D. (2003), "The Road to a Canada-U.S. Customs Union: Step-by-step or in a Single Bound", C. D. Howe Institute Border Papers No. 184, June, C. D. How Institute, Toronto.

Hufbauer, G. and G. Vega-Canovas (2003), "Whither NAFTA: A Common Frontier?" in P. Andreas and T. J. Bierstecker (eds.), Integration and Exclusion in a New Security Context, Routledge.

Imagawa, H. and E. Vermulst (2005), "The Agreement on Rules of Origin" in P. F. J. Macrory, A. E. Appleton and M. G. Plummer (eds.), The World Trade Organization: Legal, Economic and Political Analysis, volume I, 601-78.

Jensen-Moran, J.  (1996), "Trade Barriers as Investment Wars: The Coming Rules of Origin Debate", The Washington Quarterly, 19, 239-53.

LaNasa, J. A. (1995), An Evaluation of the Uses and Importance of Rules of Origin, and the Effectiveness of the Uruguay Round’s Agreement on Rules of Origin in Harmonizing and Regulating Them, Working Paper 1/96, Jean Monnet Center, New York School of Law.

Lloyd, P. J. (1993), "A Tariff Substitute for Rules of Origin in Free Trade Areas", The World Economy, 16, 699-712.

Lloyd, P. J. (2003), "Rules of Origin and Fragmentation of Trade" in L. K. Cheng and H. Kierzkowski (eds.) (2001), Global Production and Trade in East Asia, Oxford University Press, Oxford.

Lloyd, P. J. and D. MacLaren (2004), "Gains and Losses from Regional Trade Agreements: A Survey", Economic Record, 80, 445-67.

McKinsey and Company (2003), ASEAN Competitiveness Study, ASEAN Secretariat, Jakarta.  (A summary of this report is provided by A. Schwarz and R. Villinger, "Integrating Southeast Asian Economies", The McKinsey Quarterly, Issue No. 1, 1-8.)

Manchin, M. (2005), "Preference Utilization and Tariff Reduction in European Union Imports from African, Caribbean and Pacific Countries", World Bank WPS3688, World Bank, Washington, D. C.

Palmeter, D. (1993), "Rules of Origin in Customs Unions and Free Trade Areas" in K. Anderson and R. Blackhurst (eds.), Regional Integration and the Global Trading System, Harvester Wheatsheaf, New York.

Productivity Commission (2004a), Rules of Origin under the Australia-New Zealand Closer Economic Relations Agreement, Productivity Commission Research Report, Productivity Commission, Melbourne.

Productivity Commission (2004b), Rules of Origin under the Australia-New Zealand Closer Economic Relations Agreement: Restrictiveness Index for Preferential Rules of Origin, Supplement to Productivity Commission Research Report, Productivity Commission, Melbourne.

Steele, K. and D. Moulis (1994), "Country of Origin: The Australian Experience" in E. Vermulst, P. Waer and J. Bourgeois (eds.), Rules of Origin in International Trade: A Comparative Study, University of Michigan Press, Ann Arbor, 195-256.  

Stephenson, S. (1997), "The Economic Impact of Rules of Origin in the Asia-Pacific Region", Asia-Pacific Economic Review, 3, August, 14-27.

World Trade Organisation (WTO) (2002), "Rules of Origin Regimes in Regional Trading Agreements", WTO Committee on Regional Trade Agreements WT/REG/W/45, World Trade Organisation, Geneva.

Table 1.  Main features of ROO Systems
Test	CER	SAFTA	TAFTA	AUSFTA
Primary Test	Percentage test (Last process of manufacture+ Domestic content method Factory cost basis 50 per cent for all goods)	Percentage test (Domestic content method Total cost of ‘principal manufacturer’ 50 per cent /30 per cent for goods specified in Annex 2D)	CTC	CTC
Secondary Test	No	No	Regional value content for goods specified in Annex 4.1 (Build-down method) Or Technical test in small number of lines	Regional value content for goods specified in Annexes 4-A and 5-A (Build-down method or Build-up method) Or Technical test in small number of lines
General Features
Tolerance or de minimis rules	2 per cent tolerance in unforeseen circumstances	2 per cent tolerance in unforeseen circumstances	De minimis (up to 10 per cent) for all goods	De minimis (up to 10 per cent) for most goods
Cumulation	Bilateral +Diagonal for materials originating in Forum Island countries	Bilateral	Bilateral	Bilateral
Outward processing	No	Yes (except for goods in Annex 2C)	No	No
Customs drawback	Yes	Yes