European Journal of Law and Economics

, Volume 29, Issue 2, pp 211–237

The London Agreement and the cost of patenting in Europe


  • Bruno van Pottelsberghe de la Potterie
    • Université Libre de Bruxelles (ULB, SBS-EM), ECARES, and Bruegel
    • CEPR
    • CEB and DULBEA
    • Université Libre de Bruxelles (ULB, SBS-EM), ECARES

DOI: 10.1007/s10657-009-9118-6

Cite this article as:
van Pottelsberghe de la Potterie, B. & Mejer, M. Eur J Law Econ (2010) 29: 211. doi:10.1007/s10657-009-9118-6


This paper analyses the consequences for the European patent system of the recently ratified London Agreement, which aims to reduce the translation requirements for patent validation procedures in 15 out of 34 national patent offices. The simulations suggest that the cost of patenting has been reduced by 20–30% since the enforcement of the LA. With an average translation cost saving of €3,600 per patent, the total savings for the business sector amount to about €220 millions. The fee elasticity of patents being about −0.4, one may expect an increase in patent filings of 8–12%. Despite the translation cost savings, the relative cost of a European patent validated in six (thirteen) countries is still at least five (seven) times higher than in the United States.


European patent systemLondon AgreementPatent feesTranslation costsFee elasticity

JEL Classifications


1 Introduction

The European patent convention (EPC) was ratified more than 30 years ago by seven European countries in order to exploit synergies in the search for prior art and the substantive examination process.1 The EPC proved to be an impressive success, especially if the number of member states (signatories) and the number of patent applications are considered success factors. The EPC now includes 34 member states and there have been more than 200,000 applications per year since 2006. Despite this apparent success, the European patent system is heavily criticized by various institutions, including firms, universities, national agencies and the European Commission.2 There are various reasons for these criticisms (see Guellec and van Pottelsberghe de la Potterie (2007)for an in-depth assessment), but one of the most recurrent and pressing relates to the cost and complexity of the European patent system: its high level of fragmentation and its translation requirements make it the most expensive, most complex patent system in the world. Indeed, despite the explicit 1978 objective of creating a Community patent (one patent valid for all member states), the European patent system is still fragmented: once a patent is granted by the European patent office (EPO) it must be enforced (i.e., translated, validated and renewed each year) in each desired national jurisdiction.

An important step forward has recently been taken, however. Out of the 34 member states 14 (as of November 2008) have ratified the so-called “London Agreement” (LA), which drastically reduces the translation requirements when patents are validated at national patent offices. The objective of this paper is to analyse the consequences of this agreement on the cost of patenting in Europe and its potential implications for the behaviour of applicants. First, the paper simulates the cumulated fees and translation costs associated with a patent, before and after the London Agreement. In addition to changes in costs an international comparison is then performed to assess whether important differences still prevail. The paper then investigates to what extent the cost reductions might affect the patenting behaviour of applicants.

The paper is structured as follows. The next section describes the London Agreement and the variations that have been adopted by some member states. Section 3 presents the simulations of the cost of patenting before and after the London Agreement. International comparisons of relative and absolute patenting costs are performed in Sect. 4. Section 5 is devoted to the potential consequences on the number of patent filings and on cost savings. The last section concludes.

The results suggest that the cost of patenting has reduced by 20–30% following the implementation of the London Agreement. Had the London Agreement been ratified by all member states, the cost of patenting would have been reduced by 40–60% (depending on the number of targeted states for protection). With an average translation cost saving of €3,600 per patent, the total savings for the business sector amount to about €220 millions. The relative cost of a European patent is however, still at least five times higher than in the US. The fee elasticity of patents being of about −0.4, one can expect an increase in patent filings of 8–12%.

2 The complexity of the European patent system

The Community patent has been a bone of contention for 30 years and still remains a ‘work in progress’. The current system is based on a multilateral patent treaty—the European patent convention (EPC), signed in 1973. The EPC provides a centralised application and examination procedure for European patents (Fig. 1). A European patent (EP) application has to be filed in one of the European patent office (EPO) official languages (English, German or French), the so called procedural language.3 The examination process begins with a search report that identifies the prior art relevant to the application. Once the search report is published, 18 months after the priority date,4 the applicant has 6 months to request a substantive examination of its patent (the examiner tests whether there is an inventive step). If the invention conforms with the requirements of the EPC (e.g. novelty, inventive step, and industrial applicability) the application is granted.5
Fig. 1

Application and renewal procedure at patent offices. Publication of an application takes place 18 months after the priority date (or the date of first filing, along the Paris Convention for the Protection of Industrial Property of 1883). Renewal, translation and validation occur in all desired countries for protection amongst the 34 EPC contracting states

Once granted, the European patent has to be validated and enforced in each desired member state for an effective protection. This step requires the filing of a translated patent document and the payment of validation fees in the EPC contracting states where the assignee wants to enforce its rights.6 Furthermore, annual renewal fees are payable each year (for up to a total of 20 years from the priority date) on a country-by-country basis. The failure in a particular country to provide a translated document, to pay the validation fees, and to pay the renewal fees each year makes the patent irreversibly lapse and fall into the public domain.

The fact that ‘only’ national patents (i.e., EPO granted patents validated in a national patent office) provide an effective protection mechanism in each EPC member state contributes to the prohibitive costs of the system. Unlike with other patent offices around the world, payments of national validation fees and annual renewal fees, the frequent translation requirements, and the national enforcement practices (e.g., litigation processes and identification of infringers) must be multiplied by the number of countries chosen by the applicant. In case of infringement suspicions in several countries the enforcement mechanism reaches a high level of complexity, as the application and interpretation of the EPC is specific to each national court (which has exclusive rights to judge on validity and infringement cases).7 Existing differences between jurisdictions enhance the risk that in case of multiple litigations relating to a single patent, some national courts may reach ‘opposite decisions’ regarding the same patent, as illustrated by Mejer and van Pottelsberghe de la Potterie (2009) with several case studies on parallel litigations.

Two recent initiatives have been proposed within the EPC legal framework to simplify the existing situation. The first one is the so-called London Agreement (LA) and aims at reducing translation costs. The second initiative is the European Patent Litigation Agreement (EPLA). It tackles the problem of legal uncertainty and aims at implementing a centralised European court for patent-related litigation. The objective of the EPLA is to provide a more homogeneous interpretation of the validity and of the scope of a European patent. The London Agreement and the EPLA are both optional for each of the 34 EPC member states. Whereas the former has been ratified by 14 countries so far (with entry into force as of the 1st of May 2008), the latter is still subject to intense negotiations.

This paper focuses on the consequences that the London Agreement (LA) may have for the cost of patenting in Europe and for the patenting behaviour of applicants. The simplified translation rules set in the LA apply to patent applications published in the European Patent Bulletin after the 1st of May 2008. The newly reduced translation requirements only apply in the 14 following signatory member states: Denmark (DK), France (FR), Germany (DE), Croatia (HR), Iceland (IS), Latvia (LV), Liechtenstein (LI), Luxembourg (LU), Monaco (MC), The Netherlands (NL), Slovenia (SI), Sweden (SE), Switzerland (CH), and the United Kingdom (UK).8 As of May 2008 the LA ratification process was initiated in Belgium (BE), we decided to include it among signatory member states.

The country specificities regarding translation requirements under the LA are illustrated in Fig. 2. The signatory states that have an official language in common with one of the official language of the EPO agree to dispense completely with translation (art 1(1) of the LA). The other signatory states have adopted different facets of the London Agreement, especially for the two main sections of a patent (the description section and the claims section). The states having no official language in common with one of the official languages of the EPO may require a translation of claims into one of their own official languages (art 1(3) of the LA), and optionally they may require the description to be translated into the prescribed language of the EPO (art 1(2) of the LA). In case of litigation, a full translation of the European patent into an official language of the state in which the alleged infringement took place is necessary prior to the suit (art 2 of the LA). The countries that have not signed the London Agreement still require the translation of the whole patent into their official language(s).
Fig. 2

Translation requirements in EPC member states, as of the 1st of May 2008. aThe language of the European Patent at grant is the ‘procedural language’ (one of the EPO official languages i.e. English, French or German). However, the EPO requires a patentee to provide a translation of claims into the two other official languages of the EPO before the publication in the European Patent Bulletin; bThe countries having no official language in common with one of the three EPO official languages may require the translation of the description to be supplied in the official language of the EPO prescribed by that state. cThe London Agreement was still under ratification process as of November 2008

In practice, for the validation of the European patent in, Germany, France, Luxembourg, Monaco, the United Kingdom and Switzerland/Lichtenstein, no translation is required. Belgium will join this group once the LA ratification process is completed. For validation in Latvia and Slovenia only the claims must be translated into their national language, as they accept descriptions written in any of the EPO official languages. Five countries (Croatia, Denmark, Iceland, The Netherlands and Sweden) require claims to be translated into their official languages and the descriptions to be translated into English. Austria and Ireland remain the only countries that have a language in common with the EPO but have not yet ratified the London Agreement.

3 The cost impact of the London Agreement

This section provides simulations of the cost reductions induced by the London Agreement. We begin with a methodological description and then demonstrate the relative cost savings under various scenarios (number of countries and types of expenditures accounted for).

3.1 Methodology and working hypotheses

Expenses associated with the patent granting process and with patent maintenance can be decomposed into four categories (van Pottelsberghe de la Potterie and François 2009):
  1. 1.

    Procedural costs are composed of fees payable to the patent office up to grant, namely: filing fees, search fees, patent publication fees, examination fees, grant fees (for issuing a patent) and validation fees.9 In addition to those fixed fees, patent offices sometimes charge claim-based and page-based fees to limit the filing of large documents. Additionally, the patent office may charge annual fees for maintaining a patent application that is still pending in the examination process.

  2. 2.

    Translation costs occur when the validation procedure requires the applicant to submit a patent translated into the national language of the patent office (Fig. 2 for the translation requirements requested by all EPC member states). The actual translation costs incurred depend on two factors: the patent size and the number of countries that are targeted for protection once the patent is granted by the EPO. Estimating those costs is not straightforward as they have two components: (1) a language specific translation cost10 and (2) a transaction cost which frequently includes intermediation with patent attorneys.

  3. 3.

    Maintenance costs occur after grant and consist of the payment of renewal fees that must be paid in each desired country for a maximum period of 20 years from the priority date. In general, these fees increase over time. Renewal fees are due annually or periodically (North America).

  4. 4.

    External expenses are composed of the service costs of legal advice starting from the very first step of drafting a patent application, up to the representation before the patent office, until the patent falls into public domain. While large firms usually have their own intellectual property departments, with officially accredited attorneys, small firms must always rely on external services provided by legal advisors and accredited patent attorneys.


Estimating patent costs is therefore far from straightforward, as several components are not easy to quantify and depend on the patentee’s filing strategy (e.g. the patent size, the application route, the quality of external services, the desired speed of examination process, and the targeted geographical scope of protection). Patents that have a large number of claims and pages (i.e., biotechnology patents) generate higher costs than smaller patents (i.e. traditional technologies).11 The complexity of the document will influence the number of interactions with the patent office therefore increasing the level of external expenses (Lazaridis and van Pottelsberghe de la Potterie 2007). Furthermore, the cost of patent enforcement will depend on the desired geographical scope and the level of renewal fees.

In order to approximate the cost of patenting, two methods can be applied. The first one is to conduct a survey among the patenting companies about the costs they incur during the granting process and the costs of keeping their patents in force (this first approach has been performed by Roland Berger (2005) for the companies that have filed at least one patent with the EPO). The advantage of a survey is that it provides a direct evaluation of patenting expenses. However, information obtained from the companies will also reflect their filing strategies, the results being dependent on the sample selection (i.e. small versus large companies, different fields of technology, different routes). In addition, there is no standard accounting approach to patenting costs, especially the external (or internal for large firms) services. The second method involves simulating the costs from the available (and complex) fee structures. It is more straightforward than the survey approach, allows calculation of the costs for different scenarios regarding the geographical scope for protection, and makes international comparisons easier.

The adopted simulation methodology is based on the one put forward by van Pottelsberghe de la Potterie and François (2009). Relying on the data presented by Harhoff et al. (2009b) on the share of EPO granted patents that were validated in each EPC contracting states in 2003, we consider four scenarios for the geographical scope of protection:
  • EPO-3: Germany (DE), France (FR) and the United Kingdom (UK)—with at least 75% of the patents granted by the EPO being validated in each of these counties after the grant by the EPO;

  • EPO-6: includes EPO-3 and Switzerland (CH), Italy (IT), and The Netherlands (NL)—the three countries with more than 30% of the patents granted by the EPO being enforced there;

  • EPO-13: EPO-6 and Austria (AT), Belgium (BE), Spain (ES), Denmark (DK), Finland (FI), Ireland (I.E), Sweden (SE)—with more than 12% of the patents granted by the EPO being enforced there;

  • EPO-34: the geographical scope includes all the 34 EPC contracting states;

Due to the existence of different filing routes and differences between patent office fee structures, the following working hypotheses are made:
  1. 1.

    Filing route: direct filing at the EPO. For the sake of simplicity neither the PCT route (and hence the fees requested by the World Intellectual Property Office) nor the national priority filing fees (i.e., the fees paid at the national patent office for the priority filing that precedes the EPO application) are accounted for.

  2. 2.

    No fees for SMEs, and no other special fees. Patent offices frequently allow reductions for individual applicants or small and medium enterprises. In our analysis we do not account for those discounts. All fees that are payable during the granting process are calculated for a standard patent application. Moreover, we assume that all payments are made on time, thus no fines are incurred because of late payment.

  3. 3.

    Patent size. The simulations are based on the average patent filing in terms of the numbers of claims and pages.

  4. 4.
    External expenses. The cost of external services is difficult to measure as it depends on the complexity of the application and the quality of the requested services. Depending on the stage of the patent application, we can distinguish between three types of external expenses: (1) pre-filing expenses: cost of novelty search and of drafting patent application; (2) representation before the patent office: from filing the patent application up to its grant; and (3) post-grant expenses: validation and maintenance in designated states. In order to approximate these arms’ length costs, 11 large patent attorney companies were contacted in five countries. Their fee structures for legal advice, patent drafting and representation before patent offices (for the methodological approach see Table 11 in the Appendix) are summarised in Table 1.
    Table 1

    Value of external (patent attorneys) expenses associated with a patent application, its prosecution and its validation and maintenance in six countries, as of August 2008







        Prior art search and draft





    Procedural (up to grant)






        +10 h of attorney’s work





        +20 h of attorney’s work











        Maintenance (6th–10th year)










    The cost of translation is not accounted for, as it constitutes a separate cost category in our analysis. Validation includes filing translation and taking over representation by the attorney company

    Source Own calculations from raw data provided by 11 patent law firms, for the methodological approach cf. Appendix Table 11

To have the patent application drafted by a patent attorney, the patentee has to spend between €2,500 and €5,000. Additional pre-filing expenses usually include a prior art search (€2,500–€6,500). For intermediation by the patent attorney during the application procedure, the assignee has to pay between €4,000 and €8,500. These costs vary according to the complexity of the patent. Once the patent is granted the external expenses incurred during the validation procedure (i.e. the filing of a translated patent with a national patent office) and further patent maintenance (payment of renewal fees) increase proportionally with the number of countries in which the applicant wants to have its patents enforced. This is due to the fact that most of the national patent offices require a patentee to use a professional representative. Representation by an attorney for the filing of a patent may cost the applicant up to €3,000 when seeking protection in six countries (i.e., about €500 per market).12 The basic maintenance services (i.e. payment of renewal fees) are on average €2,560 (with six states for protection from the 6th up to 10th year from the priority date), or about €110 annually per country.
  1. 5.

    Time spent pending is defined as the lapse between the date of filing and the date of grant of a patent. All patent offices, except the USPTO, require the applicant to officially request the examination. The deadline for filing the examination request varies across patent offices, ranging from 6 months after the date of publication of an application at the EPO to 5 years at the Australian Patent Office (AU-PO).13 We assume that there is no substantial delay due to the strategic behaviour of applicants and that the request for substantive examination is filed once the search report is published (generally 18 months after the priority date).

  2. 6.
    Translation requirements. Since the entry into force of the London Agreement, translation requirements have been drastically reduced in the signatory countries. The new conditions are illustrated in Fig. 2 and summarised in Table 2 for four geographical scopes of protection. The table reports the number of translations required for each of the two main components of a patent, namely the claim section and the description. It is important to keep in mind that the claims still have to be translated into the two EPO official languages other than language of the proceedings, as was the case before the LA. The most important change has occurred for the description section (Sect. 2).
    Table 2

    Required translations before and (after) the London Agreement

    Part of the patent

    Procedural language at the EPO







    2 (0)

    4 (1)

    8 (4)

    24 (16)


    2 (0)

    4 (2)

    8 (4)

    24 (16)


    2 (0)

    4 (2)

    8 (5)

    24 (17)


    EN, DE, FR

    2 (2)

    4 (4)

    8 (8)

    24 (24)

    The parentheses indicate the number of translations required after the entry into force of the London Agreement in the 15 states. EPO-3 includes Germany, Great Britain and France, which have all ratified the London Agreement. EPO-6 (or −13, or −34) stands for the patents validated in the 6 (or 13 or 34) most frequently targeted countries

    Table 3

    Relative cost savings due to the ratification of the London Agreement, 2008


    EPO-3 (LA15)

    EPO-6 (LA15)

    EPO-13 (LA15)

    EPO-34 (LA15)

    Translation (absolute cost savings in EUR)





    Translation (%)





    Procedural and translation (%)





    Procedural, translation and external services (%)


    10Y excl. external services (%)





    10Y incl. external services (%)


    Cost saving with 15 countries ratifying the London Agreement. Procedural fees include the validation fees. EPO-3 includes Germany, Great Britain and France, which have all ratified the London Agreement. EPO-6 (or −13, or −34) stands for the patents validated in the 6 (or 13 or 34) most frequently targeted countries. External expenses include the average cost of pre-filing, processing an application, including 20 h of patent attorney’s work and validation; novelty and search cost are not taken into account (Table 2)

    Source Based on own calculations: cf. Tables 8, and 10 in the Appendix

    Table 4

    Procedural and translation costs relative to the US, 2008


    Per patent

    Per claim

    Per capita

















    EPO-6 (LA15)





    JPO, Japan





    KIPO, South Korea





    SIPO, China





    CIPO, Canada





    IN-PO, India





    BR-PO, Brazil





    AU-PO, Australia





    Source cf. Tables 7, 8, and 9 in the Appendix, own calculations

    a3C-index refers to the cost per claim per million capita

    Table 5

    Estimated fee elasticities of the demand for patents in the literature


    Data and method

    Estimated elasticity

    Archontopoulos et al. (2007)

    Event analysis of the claim-based fees increase in the USA implemented in December 2004


    de Rassenfosse and van Pottelsberghe de la Potterie (2007)

    Cross-country analysis of the determinants of priority filings, including 29 EPC Member States

    −0.5 to −0.3

    de Rassenfosse and van Pottelsberghe de la Potterie (2009)

    Cross-country analysis of the demand for priority filings, including 34 patent offices

    −0.5 to −0.3

    de Rassenfosse and van Pottelsberghe de la Potterie (2008)

    Time series analysis of 25 years of patenting in the USA, Japan and Europe


    Harhoff e t al. (2009a)

    Cross section analysis of bilateral validation behaviour between EU countries. Negative impact of high translation costs dummy, and negative impact of cumulated validation and renewal fees



It is worth noticing that these six working hypotheses imply that the simulated costs displayed in the present analysis are lower bounds. Neither the fees for national priority filings, nor the fees induced by the optional PCT route, nor the fees due to numerous claims and delays in responses, are taken into account in the simulations that follow.

3.2 Cost simulations, before and after the London Agreement

The detailed cost structures of a patent filing at the EPO, for the four main scenarios, are displayed in Appendix Tables 8, and 9 and graphically summarised in Fig. 3. The relative cost reductions are reported in Table 3. In Fig. 3 the white area represents the cumulated fees (from filing and examination fees at the EPO, to validation fees at national patent offices) and the dark area represents the translation costs. The cost reductions that would be induced if the 34 member states ratified the London Agreement (LA34) are also illustrated. The London Agreement appears to have a clear and substantial impact, ranging from a cost reduction of 21% if all member states are targeted to a reduction of 29% if the patent is validated in ‘only’ six countries.
Fig. 3

Cost of patents and relative savings due to London Agreement, May 2008 (in EUR). The cost savings are simulated for three configurations: before the LA, after the LA in its current format, with 15 member states (LA15); and (LA34), with all EPC contracting states having supposedly ratified the London Agreement. Procedural fees include the validation fees. EPO-3 includes Germany, Great Britain and France, which have all ratified the London Agreement. EPO-6 (or −13, or −34) stands for the patents validated in the 6 (or 13 or 34) most frequently targeted countries. Cf. Fig 6 in the Appendix for similar figures including renewal fees. Source: Based on own calculations: cf. Table 8 in the Appendix

Had all the member states ratified the London Agreement, we would have seen a decline of 40–60% in the cumulated procedural and translation costs, twice as much as the current reduction induced by the 14+1 signatory countries.

The relative cost reduction varies according to the cost components included in the total patenting costs. This is illustrated in Table 3. For instance, if the focus is essentially on translation costs, the savings induced by the London Agreement ranges from 26% if all member states are targeted by the applicant, to 78% if the geographical scope of protection is limited to three countries. For protection in six countries the translation cost decline is of a substantial 60% (€3,648). If the cumulated costs include fees, translation and external legal advice and intermediation services, the cost decrease due to the London Agreement falls to 16% only (for protection in six countries). When the renewal fees for 10 years of effective protection are added to translation and examination fees, the relative decrease is also smaller, around 11% when six countries are targeted.

In a nutshell, the relative cost reduction varies according to the desired geographical scope of protection and the type of cost considered. If companies continue to limit their validation strategies to about six countries, the London Agreement leads to a drop of about 30% in the cumulated cost of patenting. This is a significant drop, and will probably substantially affect the patenting behaviour of applicants.

The level of cost reduction is substantial. Before the London Agreement, a firm having a patent granted by the EPO and validated in six countries (the observed average) would bear translation costs of about €6,300. Since the implementation of the LA, the costs are reduced to €2,600, resulting in total translation savings of about €3,700 per patent. Multiplying this cost-cutting by the number of patents granted each year by the EPO, around 60,000, yields a total saving for the business sector of about €220 million. Before analysing the behavioural consequences of the London Agreement (Sect. 5), it is worth analysing how Europe compares with the rest of the world in terms of patenting costs.

4 International comparison

The detailed cost structures of the most important patent offices in the world are described in Appendix Tables 8, and 9. Figure 4 displays the major components of these costs. Despite the substantial step forward induced by the London Agreement, a European patent remains much more expensive, in absolute terms, than anywhere else in the world. The cumulated translation and procedural costs are of about 17,000 US PPPs if 13 countries are targeted and 10,000 US PPPs with six countries. In all other large patent offices the cost is about five times smaller, and fluctuates around 2,000 US PPPs. If renewal fees for a 10 year protection period are included in the cumulated costs, a European patent fluctuates between 17,000 and 35,000 US PPPs, according to the geographical scope of protection. This is to be compared with about 5,000 US PPPs or less in the USPTO and all other patent offices. In other words, 10 years of protection in the US and anywhere else in the world is at least three times less expensive than having a patent granted and translated in the European patent system.
Fig. 4

International comparison of cumulated patent costs, 2008 (in US PPPs). Source: Own calculations based on the fee structure provided by the national patent offices. For the methodological approach see Appendix Tables 8, and 9. KIPO is the national patent office of South Korea, SIPO is for China, JPO for Japan, BR-PO for Brazil, IN-PO for India, AU-PO for Australia, CIPO for Canada. Cf. Appendix Fig 7 for similar data in EUR

This ranking may however, be biased in two main respects (van Pottelsberghe de la Potterie and François 2009). First, patent size (in term of the average number of claims included in a patent) varies substantially across regions. For instance, Japanese patents include far fewer claims per patent (about 9.5) than the average US patent (about 23). It seems that taking the number of claims as an indicator of demand for intellectual property might be more appropriate than the number of patents, because Japan and, to a lower extent, the US and Europe have claim-based fees, which underline the importance of claims. In addition, Japanese applicants are known to merge several of their priority filings into one large application filed before the USPTO or the EPO, leading to some extent to a substitution effect between the number of patents and their size (see Dernis et al. 2001; Archontopoulos et al. 2007; van Zeebroeck et al. 2009 for an in-depth analysis of the number of claims included in patent applications at the EPO). The second source of bias is related to the size of the market covered by the patent office. If two countries have a similar cost of patenting, but different size, the relative cost (per market unit) will logically be lower in the larger country. Table 4 provides an evaluation of the patenting cost, i.e. procedural fees (including validation fees) and translation costs, relative to the US for all countries, before and after the London Agreement (LA15).

The first potential bias is illustrated in the differences between the first two columns. Without the London Agreement a patent (claim) is six to nine (seven to 11) times more expensive in Europe than in the US. After the London Agreement the relative difference varies from four to seven (five to nine). In other words one patent (or claim) is still at the very least five times more expensive in Europe than in the US. All other countries or regions seem to match the US costs at the patent level. At the claim level, as patents in Asia are generally smaller than in the US the costs are slightly higher than in the US, but are far from the prohibitive costs of Europe.

The second correction, accounting for market size, is shown in the third column. Despite the implementation of the London Agreement the patent cost per capita in Europe is 4–6 times higher than in the US but is comparable with Australia, Canada and South Korea. In China and India, the markets with the largest population, the cost per capita is the lowest (10 times lower than in the US).

The simultaneous correction for the two sources of bias is illustrated in the last column of Table 4, with the 3C-index (the cost per claim per capita) of all offices with respect to the US index. After the implementation of the LA, the 3C-index for the European patent validated in six countries seems to match Australia, Canada and Japan and is five to eight times higher than in the US. In China and India the index is three times lower than in the US.

In a nutshell, despite the savings on translation costs induced by the London Agreement, the relative cost of a European patent validated in six (thirteen) counties is still five (seven) times higher than in the US.

5 Potential impact on the demand for patents

The likely impact of the London Agreement on the patenting behaviour of applicants is twofold. On the one hand the cost reduction of about 30% may induce firms to file more applications at the EPO, while keeping an average target for validations of about six countries on average per patent. On the other hand, as the cost reduction concerns the translation requirements, one may also expect companies to validate their patent in more countries. The actual effect will likely fall between these two opposite impacts, with an increase in both the number of patents and the number of countries of validation each year. The remainder of this section subsequently investigates the two potential effects in the light of recent quantitative findings.

Let us first assume that, due to the high level of renewal fees in each EPC member state, applicants will continue to limit their geographical scope of protection to six countries on average. In this case, the cost reduction induced by the London Agreement will essentially affect the number of patent applications at the EPO. Figure 5 plots the number of claims filed at patent offices with respect to their relative costs. The horizontal axis represents the cost per claim per million capita (in US PPPs) and the vertical axis represents the number of claims (in millions) filed at the most important patent offices in the world in 2006. Relative costs are logically smaller in the largest geographical areas, such as China or the United States. If one focuses on the largest patent offices in countries with a similar level of economic development (the US, Japan, South Korea and Europe), a demand curve appears, with a high demand for patents in the least expensive patent offices (e.g., in the United States) and a lower demand in the regions where the relative patenting costs are higher (i.e., in Europe and South Korea).
Fig. 5

The 3C-index and the demand for patents. The x-axis shows the cost per claim per million capita, expressed in US PPPs for the year 2006. The cost corresponds to the process fees up to the grant, validation fees and translation costs. The y-axis shows the total number of claims filed in 2006 in each patent office (with at least one million claims filed in 2006). Cf. Table 4 for the abbreviations. Source: cf. Tables 7, 8, 9 and 10 in the Appendix, own calculations

This graphical illustration suggests that the fee elasticity of patents is negative and substantial. In order to have a more precise evaluation of the fee elasticity of patents, a quantitative analysis must be performed. But to the best of our knowledge few or no estimates have been produced so far for EPO applications. A few studies, summarised in Table 5, have estimated fees elasticities of the demand for patents. They have been derived either from an event analysis (the US increase in claim-based fees in December 2004) or from cross country econometric studies. The estimates actually range from −0.5 to −0.2, witnessing an ‘inelastic’ demand curve (the relative impact on the demand for patents is smaller than the relative change in fees). Assuming a fee elasticity of patents of −0.4, the impact of the 30 (or 20) percent drop in the cost of patenting would lead to a 12 (or eight) percent increase in the demand for patents, everything else being equal. In other words, as the EPO receives more than 100,000 patents each year (including EP direct application and PCT filings in regional phase), one would expect a net increase of more than 10,000 patents.14 Such a claim on the future should, however, be considered with a high level of caution. Indeed, the London Agreement induces an important structural change in the European Patent System, which may actually influence the fee elasticity of the demand for patents, as suggested by Ericsson et al. (1998).

The second likely effect of the London Agreement is to increase the average number of targeted countries per patent. Harhoff et al. (2007) have analysed this issue with a gravity model that aims to explain patent flows between applicant and target countries within the European patent system. The results show that the size of countries, their wealth and the distance between their capital cities are significant determinants of patent flows. Validation fees and renewal fees further affect the validation behaviour of applicants. Translation costs seem to have an impact as well, especially in countries with high translation costs. The marked effect of validation and renewal fees (which are not affected by the LA) suggest that the aggregate effect of the London Agreement will be on both the number of patents filed at the EPO and on the number of countries of validation per patent. In their companion paper, Harhoff et al. (2008) rely on a probabilistic model to understand the probability of validating a patent in each member state. The authors’ simulations suggest that the number of patent validations induced by the London Agreement will increase by about 20%, i.e., more than 50,000 additional patents.

6 Concluding remarks

This paper has analysed the potential consequences of the recently ratified London Agreement, which reduces the patent translation requirements in 15 out of 34 EPC Member States. The simulations suggest that the cumulated costs of patenting have been reduced by 20–30% thanks to the coming into force of the London Agreement, in May 2008. The relative drop depends on the number of countries targeted for protection and on the type of costs considered. If external expenses and renewal fees for protection for 10 years are considered, the drop is only about 11% of total costs.

In nominal terms, for the average patent which targets six countries for protection, the cost savings are about 3,600 EUR. Given the fact that the EPO grants about 60,000 patents each year, the total saving for the business sector is about €220 million. As well as these substantial cost savings, one may expect an increase in the demand for patents of eight to 12%, everything else being equal.

It is important to keep in mind that translation costs still have to be supported for the claims’ section of a patent, and that cumulated (over the number of countries targeted for protection) national validation and renewal fees outperform by far the renewal fees observed elsewhere. Despite the substantial reduction induced by the London Agreement, the relative cost of a European patent is still at least four times higher than in the US (and any other large national patent office). With a larger geographical scope for protection the relative cost of a patent in Europe could increase up to ten times the cost of a US patent.

The question that directly pops up when these results are considered is whether these costs affect the behaviour of applicants, and if yes, whether it is good or bad for patent systems. On the first question, the evidence suggests that the propensity to patent is affected, although through an inelastic price elasticity of −0.4. Is this then good for the patent system? A more expensive patent is not particularly bad for an economy, provided it correlates with the quality of the examination process, and provided the costs are not too prohibitive. In Europe the high costs actually constitute a prohibitive barrier to patenting, as witnessed by the relatively small attractiveness of the European market (which has only half the attractiveness of the US patent system in terms of total applications, including PCT international applications). Given the current backlog issues raised by the heads of the US, Japanese and European patent offices, one would be tempted to conclude that it is actually good for the economy, as it limits somewhat the number of applications. However, it must be kept in mind that the simulations presented in the present paper all too amply demonstrate that what makes a European patent relatively expensive is to a large extent due to the remaining translation costs and national renewal fees that must be paid in the countries targeted for protection. In other words, the high cost are no reflection on the quality of the patent system, but the managerial complexity and financial burden induced by a highly fragmented patent system.


The seven signatory states in 1978 were Belgium, France, Germany, Luxembourg, The Netherlands, Switzerland and The United Kingdom.


For instance, Single Market Commissioner Charlie McCreevy in his statement on 3rd April 2007 said: “Patents are a driving force for promoting innovation, growth and competitiveness but the single market for patents is still incomplete.In today’s increasingly competitive global economy, Europe cannot afford to lose ground in an area as crucial as patent policy.”


Patent applications can also be filed in the national language of the EPC contracting state. However, the translation into one of the three EPO official languages has to be submitted within 3 months of the filing date.


i.e., the first time an application is filed, generally at a national patent office; cf. Stevnsborg and van Pottelsberghe de la Potterie (2007) for a detailed description of the various routes that can be used to reach the EPO.


Novelty, non-prejudicial disclosures, inventive step, and industrial applicability requirements are set according to art. 54, art. 55, art. 56 and art. 57 of the EPC, respectively.


Art 137 of the EPC.


Mejer and van Pottelsberghe de la Potterie (2009), who provide several case studies of litigations that yielded opposite results in various countries. One of the patent infringements lawsuit concerns Document Security Systems (DSS) v. European Central Bank (ECB). The US company DSS sued the ECB in August 2005 at the European Court of First Instance, alleging that the euro banknotes produced by the ECB infringe its European patent EP 0455750 relating to anti-counterfeiting technology (DSS accused the ECB of infringing its technology in the production process of banknotes). Differences across jurisdictions led to uphold the patent at the court of first instance in Germany and the Netherlands but it was invalidated in France.


Note that the entry into force of the London Agreement was conditional on ratification by at least eight countries, including France, Germany, and United Kingdom (London Agreement 2000).


Once the European patent is granted, it must be validated in the desired member states to be effectively enforced. Validation fees are paid to national patent offices for the publication of the translated patent.


Harhoff et al. (2009b) classify languages according to the level of costs of translations incurred by the patent holder. In general, translations into languages spoken in central and Southeastern Europe are less expensive than translations into the Nordic languages. Ginsburgh (2005) and Fidrmuc and Ginsburgh (2007) provide theoretical and empirical evidence supporting the idea that some languages are more difficult than others, implicitly inducing higher translation costs.


Stevnsborg and van Pottelsberghe de la Potterie (2007) present a typology of filing strategies and discuss their impact on the examination process.


This cost can be lower as the countries that ratified the London Agreement may not require an official representative to be appointed.


At the Brazilian and Chinese patent offices the request must be performed within 3 years from the application date; in India, within 4 years from the application date and in Australia, Canada and South Korea 5 years are allowed. Since 2005 the Japanese patent office allows for a 3 years period to file the request of examination (previously it was 7 years). Cf. Lazaridis and van Pottelsberghe de la Potterie (2007) for the relationship between patent size and the length of examination, and van Zeebroeck (2007) for an in-depth analysis of examination duration at the European Patent Office.


Such an increase will probably not occur in 2008 and 2009, due to the current inflationary pressures and economic slowdown, which have a more important impact on the propensity to patent than fees (de Rassenfosse and van Pottelsberghe de la Potterie 2008, 2009).



The authors are grateful to Marco Connor, Andrew Fielding, Stephen Gardner and Gaétan de Rassenfosse for their useful comments as well as to the participants of the CESifo Venice Summer Institute—Workshop Reforming Rules and Regulations, 18–19 July 2008.

Copyright information

© Springer Science+Business Media, LLC 2009