Ownership of intellectual property and corporate taxation

Rachel Griffith; Helen Miller; Martin O'Connell

doi:10.1016/j.jpubeco.2014.01.009

Griffith, Rachel;Miller, Helen;O'Connell, Martin;

2014-04-01 00:00:00

Journal of Public Economics 112 (2014) 12–23 Contents lists available at ScienceDirect Journal of Public Economics journal homepage: www.elsevier.com/locate/jpube a,b, a, a,c, Rachel Grifﬁth ⁎, Helen Miller ⁎, Martin O'Connell ⁎ Institute for Fiscal Studies, United Kingdom University of Manchester, United Kingdom University College London, United Kingdom article i nfo abstract Article history: Intellectual property accounts for a growing share of ﬁrms' assets. It is more mobile than other forms of capital, Received 31 May 2011 and could be used by ﬁrms to shift income offshore and to reduce their corporate income tax liability. We Received in revised form 24 January 2014 consider how inﬂuential corporate income taxes are in determining where ﬁrms choose to legally own Accepted 25 January 2014 intellectual property. We estimate a mixed (or random coefﬁcients) logit model that incorporates important Available online 6 February 2014 observed and unobserved heterogeneity in ﬁrms' location choices. We obtain estimates of the full set of location speciﬁc tax elasticities and conduct ex ante analysis of how the location of ownership of intellectual property will JEL classiﬁcation: respond to changes in tax policy. We ﬁnd that recent reforms that give preferential tax treatment to income F21 arising from patents are likely to have signiﬁcant effects on the location of ownership of new intellectual F23 O3 property, and could lead to substantial reductions in tax revenue. H3 © 2014 The Authors. Published by Elsevier B.V. Open access under CC BY license . Keywords: Corporate tax Intellectual property Multinational ﬁrms Patent Boxes 1. Introduction heterogeneity in ﬁrms' location choices (see inter alia, Berry et al. (1995, 2004), Nevo (2001) and Train (2003)). A key strength of this approach The growing importance of intellectual property as a factor in is that it allows us to compute own and cross tax elasticities across loca- production, and concern that it is easier for ﬁrms to shift income tions that reﬂect patterns of correlation in observed choices in the data, from this source than it is from others, presents challenges for tax and therefore to capture more realistic substitution patterns than stan- design. Firms can and do position their intellectual property with a dard logit models. Our estimates allow us to conduct ex ante analysis of view to reducing tax liabilities. However, despite these concerns, ﬁrms how the location of ownership of intellectual property will respond to do not by and large locate the legal ownership of intellectual property changes in policy. We use our estimates to simulate responses to recent in the lowest tax countries, and corporate income taxes still raise policy reforms that provide preferential tax treatment to income arising considerable amounts of revenue in most developed countries. In from patents. We ﬁnd that these reforms are likely to have signiﬁcant ef- this paper we address the question of how inﬂuential corporate income fects on the location of ownership of new intellectual property, and could taxes are in determining where ﬁrms choose to legally register lead to substantial reductions in tax revenue. Our estimates could be used ownership of an important form of intangible assets, patents. to simulate a wide range of other counterfactual situations. Our contribution is to extend the empirical literature on public We use comprehensive panel data on all patent applications made to policy and ﬁrm location choice by introducing new methods to this the European Patent Ofﬁce (EPO) by a large number of innovative area of public economics. We estimate a mixed (or random coefﬁcients) European ﬁrms over 1985–2005. A patent is a legal document that logit model that incorporates both observed and unobserved grants a ﬁrm the exclusive rights to use or licence a novel technology for a speciﬁed period of time. A ﬁrm can register legal ownership of a patent in a subsidiary that is located in a country different to the ﬁrm's ⁎ Corresponding authors. E-mail addresses: rgrifﬁ[email protected] (R. Grifﬁth), [email protected] (H. Miller), headquarters, different to the location where the underlying technology [email protected] (M. O'Connell). was created and different to the location where the intellectual property OECD (2006, p34) describes the growing signiﬁcance of intellectual property and its will be applied. Lipsey (2010) notes that, in multinational ﬁrms, intangi- simultaneous use by many different parts of a ﬁrm as “one of the most important commercial ble assets “have no clear geographical location, but only a nominal location developments in recent decades”. NESTA (2009) estimates that in the UK knowledge investment overtook ﬁxed capital investment in the mid-1990s and is now about 50% higher. determined by the parent company's tax or legal strategies.” For example, http://dx.doi.org/10.1016/j.jpubeco.2014.01.009 0047-2727 © 2014 The Authors. Published by Elsevier B.V. Open access under CC BY license . R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 13 Fig. 1. Share of patent applications made by subsidiaries of UK parent ﬁrms that are located offshore and separately from innovative activity. Notes: The bars show the share of total patent applications made by subsidiaries of UK parent ﬁrms where the subsidiary is located outside of the UK, and is not in a country where associated innovative activity was carried out. Low (high) tax countries are deﬁned as those locations that have a statutory tax rate less (greater) than the UK. Fig. 1 shows the share of patent applications made by UK parent ﬁrms Tax rules limit a ﬁrm's ability to manipulate where income arises for where the legal ownership is registered outside of the UK and in a tax purposes. Shifting income typically requires that payments made to separate place to where the underlying innovative activity occurred. compensate the company that conducts the R&D, or royalties made for This share has increased six-fold over the past two decades. The largest the use of a technology, are at preferential prices. There are transfer proportion has gone to countries that have a lower tax rate than the UK, pricing rules that aim to enforce the principle that the prices of intra- but the amount going to countries with a higher tax rate has also ﬁrm transactions are set as if they had occurred between unrelated increased. parties — this is the arm's length principle. However, these transactions We model the impact of tax on where ﬁrms choose to locate the often do not have market counterparts, which means that ﬁrms may legal ownership of patents. Tax could inﬂuence this decision because have opportunities to set the prices of related transactions in such a the legal ownership of the patent will be one of the determinants of way as to reduce tax liability. Tax rules, including those that dictate where the income derived from the patent is taxed. The proﬁts earned how a ﬁrm can allocate the returns to innovative activities, differ across from the exploitation of intellectual property will be the result of a European countries and are different to those faced by US multina- number of activities, including the research and development (R&D) tionals. For example, countries differ on the acceptable methods used investment undertaken to create the new idea, the ﬁnancing of this to calculate payments for contracted R&D services, and where there investment and the subsequent commercialisation. When these activi- are cost sharing agreements, countries differ in the requirements over ties take place in multiple countries, as is often the case for multination- whether all subsidiaries involved in the agreement need be engaged al ﬁrms, the returns must be allocated to individual jurisdictions for tax in R&D (in contrast to the US, not all European countries allow holding purposes. companies in low tax locations to be part of cost sharing agreements). Firms have an incentive to arrange their activities in such a way that, The corporate tax rate is likely to be an important determinant of the all else equal, proﬁts accrue in the country in which they would pay location in which a ﬁrm chooses to hold legal ownership of intellectual the lowest tax. There are a number of strategies that can be used to property. However, it is unlikely to be the only factor; we would not achieve this. Such strategies commonly require that the income earned expect all intellectual property to be legally registered in the lowest from exploiting intellectual property accrues outside of the country in tax countries. Indeed, legal ownership of patents is rarely in the set of which the underlying R&D took place. One way to achieve this is small countries that are often considered to be tax havens. The patents through contract R&D. For example, a subsidiary in a relatively low tax that are legally owned in such countries accounted for fewer than 0.5% country may ﬁnance (and bear the risk for) R&D activities that are of all patent applications made to the European Patent Ofﬁce over the contracted to a related subsidiary in a higher tax country (possibly period 2001–2005, and many of those are unrelated to European with the beneﬁt of R&D tax incentives and access to high skills levels). ﬁrms. This could be due, at least in part, to the operation of Controlled The contract will specify the payment to be made for the R&D activities Foreign Company (CFC) regimes, which effectively seeks to tax income (commonly equal to the costs incurred plus an arm's length mark-up). at the higher home country tax rate if it is deemed to be located in a low Returns above this payment, either from using the technology directly tax country for tax purposes. More generally, there may be characteris- or from licensing it, will accrue to the subsidiary that bore the ﬁnancial tics of a location over and above its corporate tax rate that ﬁrms value. risk. There is a tax advantage to this strategy if the true value of the R&D For example, the strength of intellectual property rights protection activities is less than the price paid for the contract R&D. A similar result and market size might play a role, and, all else equal, ﬁrms may may be achieved through the use of a cost sharing agreement that speciﬁes how subsidiaries will share the costs, risks and returns When determining the correct transfer price there are both conceptual difﬁculties — it can be hard to separately determine the value that arises from integrated activities that associated with an R&D project. Such agreements may be designed take place across countries, and practical difﬁculties — ﬁrms have more information than such that the right to exploit and capture the returns from a technology tax authorities and an incentive to minimise their tax liability. accrues to a subsidiary in a low tax country. The strategies available to a 3 Figure based on patent applications made by applicants located in Bahamas, Barbados, ﬁrm depend on how the ﬁrm is organised and on the precise tax rules Bermuda, Cayman Islands, Gibraltar, Hong Kong, Liechtenstein, Malta, Monaco, they are subject to (Finnerty et al. (2007)). Netherlands Antilles, Panama or Singapore. 14 R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 be more likely to co-locate ownership of intellectual property with The structure of the paper is as follows. In Section 2 we outline a associated real innovative activity due to externalities from co-location. model of a ﬁrm's decision over where to locate the legal ownership of There is likely to be a large degree of heterogeneity in how respon- apatent. In Section 3 we describe the data we use to estimate the sive ﬁrms are to tax when deciding where to locate the legal ownership model. Section 4 presents the estimated coefﬁcients and the tax of their intellectual property; a number of papers have emphasised elasticities between locations. An example of how the model can be the importance of incorporating heterogeneity in ﬁrms' decisions used to conduct policy simulations is given in Section 5,where we (Melitz (2003), Bernard et al. (2007a, 2007b), Krautheim and consider the impact of recent reforms that reduce the tax rate for Schmidt-Eisenlohr (2011)). This heterogeneity will arise for a number income derived from patents. A ﬁnal section summarises and concludes. of reasons, some of which relate to observable factors, and others that relate to factors unobserved by the econometrician. For example, ﬁrms 2. Firm behaviour are likely to be more sensitive to tax when choosing the location in which to legally own patents with a relatively high expected value When a ﬁrm generates a new idea, it expects to earn a stream of (Becker and Fuest (2007), Bohm et al. (2012)). Firms are also likely to income on the application of that idea in the future. Ideas will vary be differentially responsive to tax due to differences in their both in their expected values and in the number of patents they give organisational structures. Their existing network of subsidiaries, the rise to (some will lead to one patent and some will lead to many). A proﬁciency of their tax department and the tax strategies they are ﬁrm faces the decision over where to initially locate the legal ownership able to employ for managing income from intellectual property will of each patent. It will make this decision based, in part, on the rate of tax play a role. Firms with headquarters in different countries might that it expects to face on income generated by the use of the patent in respond differently if countries differ in the stringency of their the future. Unobserved attributes of ideas are likely to be crucial, and tax rules and in the effectiveness with which they are applied. Firms potentially could generate correlations in patent location decisions. operating in some markets or using certain technologies might respond The ﬁrm will also take account of other characteristics of locations differently, because, for example, transfer pricing rules may be easier to that it may value, for example, whether the real innovative activity circumvent for ﬁrms operating in markets where a high share of associated with that intellectual property is also located there, the transactions are intra-ﬁrm meaning it is difﬁcult for tax authorities to potential size of the market (if it also expects to commercialise the accurately assess what is a fair market price. Both ﬁrm size and industry idea in that location), intellectual property rights' protection, technolog- have been highlighted as important in the context of ﬁrm decision ical condition, and many other location speciﬁc factors, at least some making over how to organise offshore activities (Graham and Tucker of which are likely to be unobserved by the econometrician. The (2006) and Desai et al. (2006)). Indeed, the value of a patent, the importance of these location characteristic are likely to vary across relative attractiveness of a location and a ﬁrm's strategies and ideas. For example, high value ideas may be more tax sensitive and organisational structures are likely to vary across industries and, the importance of intellectual property protection may differ across within industries, across ﬁrms. industries. Our work relates to several papers in the literature. Most closely We develop a tractable empirical model that captures these related, Dischinger and Riedel (2011) and Karkinsky and Riedel (2012) determinants of location choice. estimate the relationship between corporate tax and, respectively, the quantity of intangible assets and the number of patent applications 2.1. Firm payoffs made by subsidiaries located in each of a number of European countries. Also related is Ernst and Spengel (2011) who estimate the impact of We specify a model in which a parent ﬁrm decides where to locate R&D tax incentives and corporate tax on patenting. In common with the legal ownership of each of its patents. Firms, indexed f =1,…,F, these papers, we are interested in the relationship between corporate realise ideas, indexed i =1,…,I. Ideas are assumed to arise exogenously tax and where ﬁrms choose to locate intellectual property. We extend over time, indexed t. Each idea can yield a single patent or a group of this literature by estimating a choice model that allows us to compute related patents; patents are indexed p =1,…,P. We model the country, the full set of own tax and cross tax elasticities and which allows us to indexed j =1,…,J, in which the parent ﬁrm decides to locate the legal carry out ex ante analysis of how location decisions will respond to ownership of each patent, allowing for correlation in decisions between potential policy changes. Our work is also related to Cohen (2012), related patents (those that are part of the same idea). We consider all which uses a discrete choice framework to study how the design of US patents taken out by a parent ﬁrm that are technologically related in a state tax rules inﬂuence US ﬁrms' decisions over in which state to quarter as part of the same idea; the precise deﬁnition of an idea is incorporate. given in Section 1. There is a considerable literature in the Hall and Jorgenson (1967) For each patent, the parent ﬁrm chooses the location that yields the tradition that considers the impact of taxes on production activity and highest payoff. The payoff the parent ﬁrm gets from choosing a location on the location of R&D. Hines (1996, 1999) and Devereux (2006) depends on the tax rate it expects to face, τ , the quality of the idea, q , fjt i provide surveys of the empirical literature. This literature ﬁnds that, whether any research activity that gave rise to the idea is located there, despite the many factors that will inﬂuence a ﬁrm's location decision, a , the strength of the country's intellectual property rights protection, ijt tax exerts a signiﬁcant effect on location choices. Hines and Jaffe the size of the local market (measured as GDP), and the level of (2001) show that tax affects the location of ﬁrms' innovative activities technological innovativeness (measured as total annual business R&D within US multinational groups. Most relevant for our analysis, previous expenditure as a share of GDP), captured in the vector x . Crucially, jt work has highlighted the role that intangible assets play in allowing we also allow location choice to depend on unobserved characteristics ﬁrms to organise their activities with a view to reducing their tax of both the idea and the location. We allow the impact of all observed burden (Altshuler and Grubert (2006)). Empirical studies provide and unobserved factors to vary across medium and large ﬁrms and for indirect evidence of tax avoidance by, for example showing that ﬁrms technologies in different industries; the subscript r =1,…,R denotes have relatively high proﬁtability in low tax countries (Grubert and the industry–ﬁrm size category an idea belongs to. Mutti (1991), Hines and Rice (1994)) and that the share of royalty We assume the payoff that ﬁrm f obtains from placing legal payments associated with low tax countries is higher than expected ownership of patent p (belonging to idea i)inlocation j takes the form, (Grubert and Mutti (2009)). Grubert (2003) formalises how intangible assets can be used to shift income and ﬁnds that about half of the in- come shifted from high-tax to low-tax countries by US manufacturing π ¼ α τ þ β a þ γ x þ ξ þ ϵ : ð1Þ pjt i fjt i ijt r jt rj pjt ﬁrms can be accounted for by income from R&D linked intangibles. R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 15 The parameters α and β vary across ideas and are functions of both across location options for a given patent, and across patent locating i i observable and unobservable idea characteristics, decisions for all patents in a given idea (Train, 2003); the covariance between the payoff of locating patent p in location j and locating α ¼ α þ α q þ σ η ; η ∼NðÞ 0; 1 ; ð2Þ another patent p′ associated with the same idea in location k is given i r r i r i i τ a by: Cov π ; π ′ ¼ σ τ τ þ σ a a . Therefore, although the pjt p kt fjt fkt ijt ikt r r model still includes an iid type I extreme value term, it also contains other unobservable components that allow for rich correlations, the β ¼ β þ σ ν ; ν ∼NðÞ 0; 1 : ð3Þ i r r i i nature of which is in large part determined by the data (along with We assume that η and ν are uncorrelated with each other and with the the distributional assumptions we make). An important consequence i i other covariates, and that the additive shock ϵ is distributed iid type I of allowing for such correlations is that the IIA property is not present pjt extreme value. ξ is a location–industry–ﬁrm size ﬁxed effect. The ﬁrm in the mixed logit model, allowing the model to capture much more rj chooses option j if, ﬂexible substitution patterns (see Berry et al. (1995, 2004), Nevo (2001) and Train (2003)). π Nπ ∀ j≠j : ð4Þ We include in x a number of time varying location characteristics pj t pjt jt that ﬁrms are likely to value when choosing where to locate the legal The resulting choice model is a mixed logit with unknown parameters ownership of intellectual property. However, there are likely to be τ a α ; α ; σ ; β ; σ ; γ ; ξ Þ. other location characteristics that ﬁrms value that we do not observe. r r r r r r rj The tax rate τ varies across ﬁrms, because the tax system in a ﬁrm's To capture these we include location–industry–ﬁrm size ﬁxed effects, fjt residence jurisdiction may interact with the rules of the countries in ξ . These will control for location speciﬁc costs, such as the legal costs rj which it is considering locating ownership of a patent through the associated with setting up a subsidiary, or location speciﬁcbeneﬁts, operation of Controlled Foreign Company (CFC) rules. We use the tax such as government provided public goods, the relevance of which rate dated t, making the assumption that when a ﬁrm chooses the might differ for ﬁrms of different size or industry. location of a patent it expects that the current tax regime will apply in Note that there may be many other patent, idea or ﬁrm speciﬁc the future. factors that do not vary across location but that do inﬂuence the The tax parameter, α , is at the idea level. We allow it to vary with an costs or beneﬁts of a location. However, because these do not vary observed measure of idea quality, q . Patents that are part of a high across location they will not enter the location choice decision, and quality idea are likely to have a higher expected value, and thus their therefore are not explicitly entered into Eq. (1); including them location may be more sensitive to tax. We also allow the idea level tax would lead to an observationally equivalent choice model, because parameter to include a random term η . This captures all components they would drop out when payoff comparisons are made across of ideas that determine the responsiveness of location choice to tax locations. and are observed by the parent ﬁrm but not by the econometrician. For instance, the quality variable is likely to be an imperfect measure 2.2. Choice probabilities of the expected value of the idea. There could be other factors that are correlated with the idea's expected values, unobserved by us, but The choice model described above implies that the probability that available to ﬁrms, which will be captured by η . legal ownership of patent p is located in location j, conditional on The parameter α captures the mean marginal effect of tax on realisations of the idea speciﬁcvariables η and ν , takes the form, r i i the payoff, α tells us how this varies with the observed quality of τ a α þ α q þ σ η τ þ β þ σ ν a þ γ x þ ξ the idea, and σ tells us the standard deviation in the effect of tax on r r i r i fjt r r i ijt r jt rj ρ η ; ν ¼ ð5Þ pjt i i τ a the payoff. These parameters all have an r subscript, indicating that we X α þ α q þ σ η τ þ β þ σ ν a þ γ x þ ξ r r i r i fkt r r i ikt r kt rk e : allow both the mean impact of tax on the payoff, and how this varies with both observed quality and unobservables, to vary across industries The unconditional probability is obtained by integrating out the unob- and across ﬁrm size. servable idea speciﬁc random terms, Similarly, we model the parameter on real innovative activity, β,as an idea level random coefﬁcient; the impact of real innovative activity P ¼ ∫ ρ η ; ν dF η ; ν : ð6Þ pjt pjt i i i i on the payoff function varies across patents with the random term ν . Firms may value locating the legal ownership of intellectual property Eq. (6) can be used to compute the impact of marginal changes in in the same country as it was created, and the strength of this motive tax on location choice probabilities. For instance, the elasticity of is likely to vary across ideas. the probability that legal ownership of patent p is in location j A central assumption of the standard multinomial logit model is that with respect to a marginal change in the tax rate in location k is the stochastic error term associated with the payoff from a particular given by, option (in our case the decision to locate ownership of a patent in a particular location) has an iid type I extreme value distribution. This ∂P τ ∂ρ η ; ν τ pjt ikt pjt i i ikt rules out correlation in latent payoffs. This assumption leads to a closed e ¼ ¼ ∫ dF η ; ν : ð7Þ pjkt i i ∂τ ∂τ pjt ∫ ρ η ; ν dF η ; ν form solution for the location choice probabilities, which is empirically ikt ikt pjt i i i i convenient. However, it is restrictive, leading the multinomial logit model to imply restrictive substitution patterns. In particular, the lack We compute the elasticity of the share of patents with legal owner- of correlation in payoffs endows the model with the independence of ship in location j with respect to the statutory tax rate in location k irrelevant alternatives (IIA) property. in year t (τ ), by aggregating across Eq. (7) for all the patents that kt Random coefﬁcients allow us to relax the assumption of zero arose in year t (denote this set of patents ϒ ). We explicitly account correlation in payoffs inherent in the standard multinomial logit for the operation of CFC regimes (see Section 2 for a description of model. In particular, we group patents into ideas and allow some how CFC regimes work). For a patent owned by a ﬁrm resident in a of the parameters in the payoff function (those on tax and on real country that has a CFC regime, and that deems country k as a low innovative activity) to be idea speciﬁc. We model these idea speciﬁc tax jurisdiction, changes in the statutory tax rate of location k preference parameters as random coefﬁcients — which leads to the should have no bearing on their location probabilities. Deﬁne an in- mixed logit model. An implication of including random coefﬁcients at dicator variable D ,where D = 0 if patent p at time t is subject to pjt pjt the level of an idea is that it allows for correlation in the payoffs both CFC rules which bind in location j,and D = 1 otherwise. We can pjt 16 R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 then write the elasticity of the share of patents located in country j by the same amount, the marginal effect of tax on the payoffs would with respect to tax in location k in year t as, not be identiﬁed. Such variation arises in our framework for two reasons. First, there is variation over time in statutory tax rates; as outlined below in Section 2, over the period for which we have data (1985–2005) there has been a ∂ρ η ; ν pjt i i kt E ¼ ∫ D dF η ; ν X : ð8Þ general pattern of declining statutory tax rates. The size of this decline jkt pkt i i ∂τ ikt ∫ ρ η ; ν dF η ; ν p∈ϒ t pjt i i i i p∈ϒ has varied across countries, and the changes have occurred at different times, meaning that tax reforms have given rise to variation in the set of tax rates across locations. Second, in addition to this time series 2.3. Identiﬁcation variation, CFC regimes lead to another source of variation; two parent ﬁrms taking a decision at the same point in time, but resident in Our primary interest lies in pinning down the ceteris paribus impact different countries, can face a different set of tax rates due to cross of a change in the corporate tax rate set by any one country on the country differences in CFC rules. The variation in location choices, shares of patent applications made by subsidiaries in both that and in conditional on other factors, in response to this variation in tax rates alternative locations. To do this we must consistently estimate the pins down the impact of tax on location choice. parameters of the payoff function outlined in Eqs. (1)–(4), and in partic- ular the parameters governing the marginal impact of tax on the payoff 3. Data associated with selecting a location, which are modelled as random coefﬁcients. Train (2003) shows that the random coefﬁcient model To estimate the model we need information on where ﬁrms have has a duel interpretation as an error component model. Under this chosen to locate the legal ownership of their patents, the corporate representation, the mean of the random coefﬁcient can be interpreted tax regime and other conditioning variables. as a ﬁxed coefﬁcient — it is pinned down by variation in location choices in response to variation in taxes faced by ﬁrms, conditional on other 3.1. Patents data observables included in the model. The standard deviation of the random coefﬁcient is interpreted as a component in the error We use data on patent applications ﬁled at the European Patent term — it is pinned down by correlation in payoffs across locations Ofﬁce (EPO) by the European and US subsidiaries of parent ﬁrms located (both in a given choice set and across location decisions within a in fourteen European countries. We exclude from our analysis ﬁrms that given idea). patent infrequently. The number of patent applications by location of Berry and Haile (2010) have established that in random utility the subsidiary that ﬁled the patent application is shown in Table 3.1. multinomial logit models the distribution of unobserved preference Our data include 1083 parent ﬁrms that collectively have 4,823 parameters is non-parametrically identiﬁed given sufﬁciently rich patenting subsidiaries, which ﬁle 379,849 patent applications over the micro data. However, non-parametric estimation of this model is period 1985–2005. These account for a 70% of all corporate applications computationally burdensome, and we therefore follow most papers in ﬁled at the EPO by ﬁrms parented in these fourteen European countries the literature by assuming payoffs are linear with independent additive during this period. shocks and that the distribution of unobserved parameters is normally Each patent application lists the ﬁrm that ﬁles the application distributed. These assumptions give us a convenient approximation. (the applicant), this is the legal owner of the patent. We identify the The standard identiﬁcation concerns still apply here; to consistently parent ﬁrm using information from company accounts (from Amadeus), estimate the parameters we require that the additive shock (ϵ ) and pjt company websites, business directories and other sources (see the idea speciﬁc random terms (η and ν ) are independent from each i i Abramovsky et al. (2008) for details). We use ownership information other and from the other explanatory variables. Speciﬁcally, if there at a ﬁxed point in time (2004), and we do not observe changes in are factors which inﬂuence location choice, and that are not captured ownership after an application has been ﬁled. by the observed and unobserved controls that we include, this would For each patent application this gives us a mapping between the lead to inconsistent estimates. To mitigate this concern, we include a location of the parent ﬁrm and the location of the subsidiary that number of controls in the model. We include location–industry–ﬁrm legally owns the patent. We also observe the location of the inventors size ﬁxed effects; these control for all country characteristics that affect (individuals) that created the technology underlying the patent applica- a ﬁrm's payoff and that do not vary through time, but that potentially do tion. There are often inventors located in multiple countries, and often vary across ﬁrms in different industries and different parent ﬁrm sizes. in different countries to that of the applicant. Thelocationofboththe We include time varying (non-tax) location characteristics. These applicant and the inventors are distinct from the patent ofﬁce to include a measure of the presence of real innovative activity associated which the ﬁrm is applying for protection. For patent applications ﬁled with the intellectual property. This controls for the fact that some ﬁrms, at the EPO, each application also designates the individual countries for reasons other than the tax rate they will face in a particular location, in which ﬁnal patent protection will be sought; it is these individual may wish to co-locate legal ownership of intellectual property with real countries, not the EPO, that grants patent protection. innovative activity. If decisions over the location of real innovative We use Thomson's Derwent database to classify patent applica- activity are inﬂuenced by corporate tax rates, then failure to control tions based on the technology embodied in the patent and the for this would result in an inconsistent estimate of the impact of tax markets in which the technology is used. We use three broad indus- on patent location choice. We also control for the strength of intellectual try groups — Chemical, Electrical and Engineering. A patent applica- property rights protection in the location, market size and technological tion can be relevant for more than one industry group if it has innovativeness — all factors that vary over time and location, and may applicability in more than one of these industries. Where this is be expected to impact on intellectual property location choices. the case we include the patent application in each of the industry Identiﬁcation of the tax coefﬁcients also relies on the presence of informative variation in taxes in the data; speciﬁcally we need to In most EPO applications there is one applicant. For the handful of patent applications observe variation in the set of taxes in potential locations across patent that are ﬁled by multiple applicants in multiple locations we randomly select one of them. choice situations, conditional on all other factors that inﬂuence location A small number of patent applications (0.07% of those ﬁled by all applicants in the 15 choice. Crucially, it is necessary that there is variation in differences in countries we consider) have missing inventor data. We exclude these applications. taxes between locations across choice sets. So, for instance, if the only This is different to data on patents ﬁled at the US Patent and Trademark Ofﬁce for source of tax variation was that all tax rates changed simultaneously which the inventors are the patent applicants. R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 17 Table 3.1 Patent applications and tax rates by applicant country. Applicant country Number of patent % patent applications: Number of applications rate changes By industry By ﬁrm size High Statutory tax rate Chemical Electrical Engineering Large Medium quality min max mean (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Belgium 6935 47.8 22.9 29.3 74.2 25.8 29.0 34.0 45.0 40.4 5 Denmark 4975 56.9 15.2 27.9 65.7 34.3 31.2 28.0 50.0 36.0 7 Finland 10151 13.9 72.6 13.5 83.1 16.9 33.8 25.0 60.2 36.8 7 France 50017 26.6 45.2 28.2 87.5 12.5 37.1 33.3 50.0 38.6 13 Germany 166606 29.1 34.3 36.6 78.5 21.5 35.8 38.3 61.7 52.5 15 Ireland 387 70.6 7.7 21.7 52.5 47.5 26.0 10.0 12.5 11.3 1 Italy 11155 34.1 26.3 39.5 47.9 52.1 29.9 37.3 53.2 44.6 7 Luxembourg 792 52.9 25.6 21.4 32.8 67.2 41.4 30.4 39.4 37.5 6 Netherlands 34918 29.3 51.4 19.3 90.2 9.8 36.1 31.5 43.0 37.2 4 Norway 1083 35.3 29.4 35.3 47.7 52.3 36.8 28.0 50.8 39.4 1 Spain 994 44.7 19.8 35.5 23.9 76.1 33.5 35.0 35.0 35.0 1 Sweden 17300 21.1 49.0 29.9 82.7 17.3 43.3 28.0 52.0 36.7 2 Switzerland 30221 45.6 23.7 30.7 73.8 26.2 39.6 21.3 28.5 25.1 6 United Kingdom 27955 45.0 30.0 25.0 67.3 32.7 38.2 30.0 40.0 33.8 5 United States 16360 48.7 29.9 21.4 88.6 11.4 37.9 34.0 49.5 40.1 5 Total 379849 32.3 36.8 30.8 78.8 21.2 36.5 10.0 61.7 43.2 6 Notes: Applicant country refers to the location of the subsidiary that is the applicant on a patent application. The ﬁnal row gives means across all patent applications in our data. The mean tax rate in the ﬁnal row is averaged across all choice options, so it differs from Table 3.3. sub-samples, and when we calculate the market level elasticities we their invention to other countries if they deem it worthwhile.” (OECD, weight each patent application so that the sum of the weights equals 2012). one (so if a patent application is in two industries it will get a weight We expect triadic patent applications to be of a higher value since of 0.5 in each). Table 3.1 columns (2)–(4) shows the industry split of there is a cost to ﬁling patent applications at each of these patent ofﬁces, patent applications, 32.3% are in Chemical, 36.8% in Electrical and and the main incentive to do this is if ﬁrms expect the technology to 30.8% in Engineering, but this varies across countries. We restrict have a wide application. Each idea (group of patent applications) is our analysis to ﬁrms that are above the 20th percentile in terms of classiﬁed as high quality if over half of the associated patent applications the number of patent applications per ﬁrm in their industry. We dis- are triadic. As seen in Table 3.1 (column (7)) on average 36.5% of patent tinguish large ﬁrmsasthose with alevel of patentingabove the80th applications are classiﬁed as being part of a high quality idea. percentile in their industry. 78.8% of patent applications in our sam- ple are held by large ﬁrms. 3.1.2. Patent ownership and income from patents Firms often take out a number of related patent applications at the We model the impact of tax on where ﬁrms choose to locate the same time; we allow for correlation in these decisions. We group legal ownership of patents. In the introduction we discuss the reasons together related patent applications that can be considered to be part that we might expect tax to affect a ﬁrm's decision of where to hold of the same idea. We identify patent applications as part of the same legal ownership of intellectual property. idea if they are made by the same parent ﬁrm, are ﬁled in the same The extent to which ﬁrms have arranged their activities in such a quarter (i.e. three month period), are classiﬁed in the same industry way that income can reasonably be deemed to be attributable to the and share a network of common inventors. The number of patent subsidiary that legally owns the intellectual property will differ; ﬁrms applications in an idea varies: on average an idea contains one patent will differ in how aggressively they seek to manage their tax liabilities. application; ideas containing more than one application account for For some ﬁrms, the choice of where to earn income may be a choice 26% of all patent applications. between those countries in which real innovative activities already The importance of ideas in our empirical strategy is that we allow takes place; others may employ strategies that allow them to earn correlation across patent applications at the level of the idea; the income in a separate country. decisions over where to locate ownership of these related patent There are many factors that affect the costs and beneﬁts of choosing applications is unlikely to be independent, and the inclusion of random a particular location. For tax havens these costs may be particularly coefﬁcients at this level allows for them to be correlated. high: CFC rules are more likely to bind; the transfer of proﬁts to locations where there is little real activity will be more difﬁcult; tax 3.1.1. Patent quality havens are likely to be less attractive locations along non-tax dimen- There is a large literature that highlights the skewness of patent sions such as intellectual property rights protection. We would not value and quality (Pakes (1985, 1986), Blundell et al. (1999), Lanjouw expect legal ownership of all patents to be located in such countries. and Schankerman (2004) and Hall et al. (2007)). Firms ﬁle patent However, it is possible that some ﬁrms are particularly aggressive in applications for a variety of reasons; some are ﬁled to protect valuable their tax planning and organise their activities in such a way that new ideas, others are ﬁled strategically to provide option values or to income is earned in a location that is not where legal ownership is block competitors. Patent value also varies because some ideas are located. We do not observe income ﬂows, so we do not explicitly commercially more valuable than others. We identify high quality model this behaviour; to the extent that it makes the decision over patent applications as those that are part of a triadic patent family, i.e. the location of legal ownership less related to tax we would be less likely a related patent application has been ﬁled at each of the EPO, the to ﬁnd an impact of tax. In our model we aim to capture this variation in US Patent and Trademark ofﬁce and the Japan Patent Ofﬁce. The behaviour across ﬁrms, and within ﬁrmsacrossideas,bytheinclusionof OECD uses triadic patent families “… to improve the international observed and unobserved heterogeneity. comparability and quality of patent-based indicators … patents included An additional complicating factor is that a ﬁrm might ﬁle a patent in the triadic family are typically of higher economic value: patentees application from one subsidiary, but later transfer ownership of that only take on the additional costs and delays of extending the protection of patent to another related ﬁrm. However, ﬁrms have an incentive to 18 R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 consider tax when making the initial location decision, because in many taxation in their home country. CFC rules set out criteria for identifying situations there are tax costs to transferring the ownership of intangible subsidiaries that are located in a country deemed to be ‘low-tax’ and assets. For there to be a tax beneﬁt to the sale or transfer of an asset it earning a signiﬁcant amount of ‘passive income’ (income that is not must be the case that this can happen at a value below the true market associated with real activity). When a CFC regime is in place in a parent value. The transfer of intellectual property will be subject to transfer ﬁrm's country of residence, and a subsidiary is located in a country that pricing rules, which will act to limit how much value can be shifted to is deemed a ‘low tax’ location (as judged against parent ﬁrm country a low tax country. In addition, many European countries operate exit speciﬁc thresholds), then we set the tax variable, τ , equal to the parent fjt taxes that attempt to levy tax on the net present value of the expected ﬁrm country's statutory rate. A description of the country pairs for revenue stream on an intangible asset when it is moved out of the coun- which this is the case is given in Table 3.2. There is variation in whether try. Such tax provisions reduce (if not remove) any tax advantages to re- aparent ﬁrm country operates a CFC regime (some regimes are locating to a lower tax jurisdiction. If ﬁrms do intend, with some introduced during the period for which we have data) and in the positive probability, to re-locate the ownership of a patent in the future, applicant countries that are deemed low tax (which differ over time and if transfer pricing rules and exit taxes do not act to perfectly off-set when statutory rates change). This deﬁnition of whether CFC rules any tax advantages of doing so, this would reduce the importance of bind effectively assumes that the income received from a patent is corporate tax in the initial location decision. This is an additional reason deemed to be passive income, and that the share of passive income is that it is important that we allow heterogeneity in the importance of sufﬁcient to trigger the CFC rules. This is clearly an approximation. corporate tax across intellectual property. However, if we look across all location options that ﬁrms in our data The place where we need to make more restrictive assumptions face and that are deemed low tax by CFC regimes, then it is rarely the about the relationship between legal ownership and income from case that the parent ﬁrm has both inventors and holds legal ownership intellectual property is when we carry out the ex ante analysis of the of a patent application in the same location. The results we present Patent Box tax reforms and calculate the revenue implications of these below are robust to the alternative assumption that patent applications reforms. In order to do this we need to assume that the relationship with ownership located in countries where associated real innovative between legal ownership and income is not changed by the policy activity is also located would be treated as active income, so that CFC reform. rules do not bind. 3.2. Taxes 3.3. Descriptive statistics We measure the impact of tax on payoffs using the statutory tax The variables included in the model are deﬁned and summarised in rate. We assume that returns from intellectual property are expected to be sufﬁciently high that deductions such as capital allowances are Table 3.3. The top panel contains the observed location attributes we include. These comprise the tax rate that the parent ﬁrm would face if relatively unimportant, so that the effective tax rate faced by the ﬁrm is approximately the statutory tax rate (see Devereux and Grifﬁth it earned income from the application of intellectual property in the (2003), where Fig. 1 shows that the marginal effective tax rate location, a measure of the presence of real innovative activity in a asymptotes to the statutory tax rate as proﬁtability increases). location deﬁned as an indicator of whether at least one of the inventors Our identiﬁcation strategy relies on variation over time and across associated with the patent applications that form the idea are located in countries in the tax rate. Table 3.1 (columns (8)–(11)) summarise the that country and country-time varying observable characteristics. The variation in corporate tax rates. In general, main statutory tax rates fell latter includes a measure of intellectual property rights protection. in the two decades up to 2005, but with the timings of changes differing This is based on a measure developed by Ginarte and Park (1997) and across countries. The Scandinavian countries – Denmark, Finland, Park (2008). The countries we consider all have advanced systems of property rights and therefore rank relatively highly on the protection Norway and Sweden – reduced tax rates signiﬁcantly around 1990. Italy enacted a reduction of over 10 percentage points in 1998, as did of intellectual property. We deﬁne a country as having a strong intellec- tual property regime if it scores above the median of countries in our Germany in 2001. France and the UK have enacted a series of gradual reductions. sample. Other country-time varying variables include market size, as measured by Gross Domestic Product (GDP) and the technological There can be additional tax levied in the parent ﬁrm's home country as a result of Controlled Foreign Company (CFC) rules, which aim to innovativeness of a country, proxied by business R&D investment in prevent ﬁrms locating income in lower tax countries in order to avoid the country as a share of GDP. Table 3.2 Number of ﬁrms and CFC regimes by parent ﬁrm country. Patent ﬁrm Number of CFC regime Applicant countries for which CFC ever binds country parent ﬁrms introduced (no. of years) Belgium 28 –– Denmark 27 1995 FI (1 year), IE (all years) Finland 25 1995 IE (11 years) France 108 1980 CH (6 years), IE (all years) Germany 446 1972 CH (all years), FI (8 years), GB (1 year), IE (all years), NO (9), SE (10) Ireland 4 –– Italy 75 2002 – Luxembourg 8 –– Netherlands 58 –– Norway 7 1992 IE (14 years) Spain 8 1996 CH (7 years), FI (2 years), IE (all years) Sweden 48 1990 CH (1 year). IE (11 years) Switzerland 106 –– United Kingdom 135 1984 CH (2 years), IE (all years) Notes: Country codes: Belgium (BE); Switzerland (CH); Denmark (DK); Finland (FI); France (FR); Germany (DE); Ireland (IE); Italy (IT); Luxembourg (LU); Netherlands (NL); Norway (NO); Spain (ES); Sweden (SE); United Kingdom (GB). R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 19 Table 3.3 Variable deﬁnitions and summary statistics. Variable Deﬁnition Min Max Mean Standard deviation Location characteristics Tax rate (τ ) Statutory tax rate in applicant country; or statutory tax rate 10.00 61.70 41.79 11.34 fjt in parent ﬁrm county when binding CFC regime Real activity Dummy equal to one when any of the inventors associated 0.00 1.00 0.86 0.35 with the patent applications that form an idea are located in that country Strong intellectual property protection Measure of applicant countries' relative degree of intellectual 0.00 1.00 0.76 0.43 property rights protection Market size GDP measured in millions of constant PPP US dollars 0.01 12.56 1.90 1.97 Technological innovativeness Business investment in R&D as a share of GDP 0.29 3.20 1.53 0.40 Patent characteristics Large ﬁrm Large parent ﬁrms are those for which the total number of patent 0.00 1.00 0.79 0.41 applications is above the 80th percentile High quality Ideas are classiﬁed as high quality if over half of the associated patent 0.00 1.00 0.36 0.48 applications were ﬁled at each of the EPO, USPTO and JPO Electrical Instrumentation, computer, electronics, communications, electrical 0.00 1.00 0.37 0.42 Chemical Chemicals, pharmaceuticals, printing, petroleum 0.00 1.00 0.32 0.42 Engineering General and mechanical engineering 0.00 1.00 0.31 0.38 Notes: Statistics are based on all patent applications in our data. GDP is measured in constant PPP US dollars (expenditure measure) using a 2005 base year. Business Investment in R&D as a share of GDP (BERD) is from OECD Main Science and Technology Indicators. GDP and BERD are available at http://stats.oecd.org. We allow the valuations ﬁrms place on location characteristics to Together the large and statistically signiﬁcant standard deviations vary across patent applications. A summary of observable patent on the random coefﬁcients on tax and real innovative activity (in all (or idea) characteristics is given in the bottom panel of Table 3.3.In m size groups) indicates the presence of important correla- industry-ﬁr estimation we allow all coefﬁcients to vary with the industry the patent tions in payoffs, both across locations for a given patent, and across application belongs to and the size of the associated parent ﬁrm. This patents in a given idea. These correlations will generate patterns of sub- allows the model to capture, for example, that large ﬁrms are more stitution that will depart from the more restrictive patterns implied by a likely to have organisational structures that assist the location of standard multinomial logit model. intellectual property for tax purposes. The tax rate is interacted with a The remaining three rows of Table 4.1 describe the impact of measure of the idea quality, reﬂecting the possibility that ﬁrms' location having strong intellectual property protection, and the marginal choices may be more responsive to tax when they expect intellectual impacts of market size and technological innovativeness, on the property to earn higher returns. payoff function. For ﬁve of the six industry-ﬁrm size groups, a loca- tion having strong intellectual property protection is, all else equal, associated with ﬁrms obtaining higher payoffs from locating legal 4. Results ownership of their patents there (the exception is medium electric ﬁrms, for which the strong intellectual property rights dummy is Table 4.1 shows the estimated coefﬁcients of the choice model negative). Larger market size is associated with statistically signiﬁ- outlined in Section 2. The model is estimated using simulated maximum cantly larger payoffs for ﬁve of the size industry-ﬁrm size groups, likelihood (see Train (2003)). We allow all coefﬁcients to vary across and a higher degree of technological innovativeness is associated industry and ﬁrms size, indicated by the different columns. We include with statistically signiﬁcantly larger payoffs for four of the size a full set of location–industry–ﬁrm size ﬁxed effects (not reported in industry-ﬁrm size groups. Table 4.1, but available upon request). Table 4.2 shows the matrix of own and cross tax elasticities The top row of Table 4.1 shows that the mean marginal impact of implied by the choice model. It contains the elasticities of the tax on the payoff from placing legal ownership of a patent in a loca- share of patents located in each of 14 European countries with tion is negative and statistically signiﬁcant across all industries and respect to the rate of corporate tax set in each of these countries parent ﬁrm size groups. The second row shows that in both the elec- and in the US. These are calculated as described in Section 2.We trical and engineering industries the payoff for high quality patents report the matrix of elasticities using tax rates and the distribution is more sensitive to taxes. This is true both for large and medium of patent applications for the most recent year in our data, 2005. ﬁrms. In the chemical industry the payoff for a high quality patent Each cell shows the elasticity of the share of patents located in the is estimated to be marginally less responsive to tax than for lower country indicated in column 1 with respect to the tax rate set by quality patents for large ﬁrms, with there being no statistically the country in row 1. The emboldened diagonal shows the own tax signiﬁcant difference between the high and low quality patents for elasticities. For all locations, except Luxembourg, the own tax medium ﬁrms. Row three shows that there is a substantial degree elasticities are less than one in magnitude. of unobserved heterogeneity in the importance of tax on location There is a limited literature on the elasticity of the location of corpo- choice across ideas, the standard deviations of the random coefﬁ- rate income with respect to tax. De Mooij and Ederveen (2008) report cients on tax are both large and statistically signiﬁcant across all that empirical studies considering the effect of differences in statutory industries and size categories. tax rates on various measures of proﬁtability (with a view to indirectly The fourth row shows that, ceteris paribus, having real innova- capturing the effects on proﬁt shifting) tend to ﬁnd a semi-elasticity of tive activity in a location is associated with a higher payoff from around −1.2. As in this paper, Karkinsky and Riedel (2012) consider placing legal ownership of a patent in that location across all indus- the link between corporate tax rates and patent applications. They tries and size categories; the ﬁfth row shows that there is a signiﬁ- estimate a semi-elasticity that, depending on the functional form of cant amount of variation in the importance of this characteristic their model, implies that a 1 percentage point increase in the rate of across ideas. corporate tax translates into a 3.5%–3.8% fall in patent applications 20 R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 Table 4.1 Estimated parameters. Industry Electrical Engineering Chemical Size Large Medium Large Medium Large Medium Tax rate (mean) −3.48 −4.93 −4.91 −4.88 −6.54 −4.03 (0.13) (0.32) (0.14) (0.24) (0.13) (0.28) Tax rate x quality −0.67 −1.98 −0.69 −0.66 0.34 0.08 (0.14) (0.37) (0.15) (0.28) (0.14) (0.31) Tax rate (std. dev.) 3.59 3.67 2.35 3.17 3.36 4.24 (0.16) (0.40) (0.23) (0.27) (0.15) (0.27) Real activity (mean) 5.60 7.27 6.20 7.03 6.11 6.55 (0.03) (0.12) (0.05) (0.09) (0.04) (0.09) Real activity (stand. dev) 2.50 2.76 2.80 2.96 2.95 2.59 (0.03) (0.10) (0.04) (0.08) (0.04) (0.08) High IP property protection 0.63 −0.22 0.28 0.19 0.68 0.60 (0.04) (0.11) (0.06) (0.10) (0.06) (0.10) GDP −0.20 0.66 0.53 0.43 0.39 0.20 (0.02) (0.07) (0.03) (0.05) (0.02) (0.05) R&D expenditure/GDP 0.26 0.03 0.23 0.09 0.18 0.00 (0.02) (0.05) (0.03) (0.04) (0.02) (0.05) Notes: The dependent variable is location choice in one of the locations shown in Table 3.1. Estimation is based on 379,849 patent applications. Industry–location–ﬁrm size ﬁxed effects included. Large ﬁrms are those associated with a total number of patent applications above the 80th percentile in each industry. from that location. Direct comparison with our results is made difﬁcult allows us to simulate counterfactual policy situations. We illustrate by the fact that our model allows tax effects to vary across all locations. this by considering a recent set of policy reforms. We ﬁnd that the share of patents held in Luxembourg is most sensitive A number of European countries have introduced polices that to tax (the Luxembourg semi-elasticity is 3.9%) and least sensitive for offer substantially reduced rates of corporation tax on the income Germany (the German semi-elasticity is 0.5%). derived from patents, and in some cases other forms of intellectual Theoretically, we might expect smaller countries to have rela- property (these are often called Patent Boxes). Firms are able to tively high own tax elasticities, as a change in their tax rate will declare that some portion of their proﬁts are derived from either not affect the market rate of return, making the cost of capital the use or licence of patents, and these proﬁts are taxed at a lower more responsive to tax changes (see Wilson (1999)). This may be rate. Patent Box rules differ across countries, for example, in terms one of the reasons that such countries are more likely to compete of how eligible income is measured, how the rules that apply for corporate income using low rates; a change in the rate leads to when calculating how much income can be allocated to patents, a larger change in the relatively small tax base (see, for example, None of the countries and how the related expenses are treated. Bucovetsky and Hauﬂer (2007)). The own tax elasticities in Table 4.2 require that the R&D underlying the intellectual property took show some evidence of this; they are higher for the Benelux countries place in that country, as this is not permissible under European law. than for France and Germany. We use the most recent year of our data (2005) to simulate the The importance of allowing for observed heterogeneity and impact of the two sets of policies. First we consider the introduction of correlation in locations' payoffs can be seen by looking at the cross tax Patent Boxes in the Benelux countries, and second the later introduction elasticities. In a multinomial logit with no observed or unobserved het- in the UK. We simulate the impact of these policies on the share of new erogeneity all cross tax elasticities in a column would be the same — a patents for which legal ownership is placed in each of these countries reduction in the tax rate in location A would lead to patent applications using the choice model presented above. For illustrative purposes, we switching from other locations in proportion to their original shares. assume that the total level of patenting activity by European ﬁrms is This implausibly restrictive pattern of substitution is not imposed in not affected by the policy reforms. We also consider the impact of our more ﬂexible model, meaning that elasticities vary substantially these policy reforms on tax revenue; this requires the further assump- within a column. In particular, our model allows the data to capture tion that the relationship between where tax is levied and the location the fact that ﬁrms are more likely to choose to switch between locations of legal ownership is not altered by the policy reform. with similar characteristics (whether that be because the ﬁrm has in- The policies are summarised in Table 5.1. In 2007 Belgium intro- ventors located in several locations, or because locations have similar duced a Patent Box that reduced the tax rate on income derived from tax rates). patents from 34% to 6.8%, and the Netherlands introduced a Patent Box that reduced the rate from 31.5% to 10%. In 2008 Luxembourg reduced the rate from 30.4% to 5.9%. The UK government introduced a 5. Policy simulations Patent Box at the rate of 10% in 2013; the main rate of corporate tax in the UK was 30% in 2005, but had fallen to 24% by 2013. We simulate One of the advantages of estimating the model outlined above is that the impact of the reduction from 30% to the Patent Box rate. it captures patterns of substitution across locations, and it therefore Table 5.2 sets out the results of these simulations for the four locations that introduced Patent Boxes. A note of caution in interpreting these results is that the lowest tax rate we observe in the Note that in Table 3.2 we adopt of the convention of reporting tax elasticities. Roughly speaking these can be interpreted as telling us the % change in the share of patents in lo- cation A associated with a 1% change in the rate of tax in location B. These can readily be Evers et al. (2013) provide further details on the policies and incorporate the rules into converted to semi-elasticities (which, roughly speaking, tells us the % change in the share effective tax rates. of patents in location A associated with a 1 percentage point change in the rate of tax in For further discussion of the UK Patent Box see Grifﬁth and Miller (2010, 2011).A location B) by dividing by the appropriate tax rate. So for instance the 2005 rate of corpo- number of other European countries (Cyprus, Liechtenstein, Malta, Spain, and the Swiss rate tax in Germany was 38.3% and the German own tax elasticity is 0.201. Hence the Ger- canton of Nidwalden) have since introduced similar policies. man own tax semi-elasticity is given by 100 × (0.201 / 38.3) = 0.52%. The full set of results, including the impact on other countries, is available on request. R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 21 Table 4.2 Own and cross price elasticities of locations with respect to statutory tax rates. Belgium Denmark Finland France Germany Ireland Italy Luxembourg Netherlands Norway Spain Sweden Switzerland UK US Belgium −0.569 0.010 0.018 0.085 0.174 0.001 0.016 0.006 0.079 0.003 0.003 0.030 0.036 0.025 0.038 Denmark 0.016 −0.410 0.015 0.044 0.145 0.000 0.013 0.007 0.060 0.003 0.004 0.033 0.047 0.025 0.037 Finland 0.015 0.008 −0.465 0.096 0.190 0.000 0.015 0.003 0.103 0.002 0.002 0.046 0.031 0.021 0.043 France 0.013 0.004 0.017 −0.311 0.118 0.000 0.009 0.002 0.052 0.001 0.001 0.017 0.017 0.011 0.021 Germany 0.007 0.004 0.009 0.032 −0.201 0.000 0.006 0.002 0.035 0.001 0.001 0.012 0.016 0.008 0.012 Ireland 0.048 0.031 0.033 0.103 0.279 −0.404 0.063 0.017 0.162 0.007 0.009 0.048 0.176 0.054 0.352 Italy 0.011 0.006 0.012 0.041 0.095 0.001 −0.373 0.004 0.043 0.002 0.002 0.017 0.035 0.019 0.017 Luxembourg 0.062 0.040 0.029 0.126 0.450 0.002 0.055 −1.197 0.125 0.012 0.019 0.058 0.093 0.073 0.079 Netherlands 0.022 0.010 0.034 0.094 0.234 0.001 0.017 0.004 −0.569 0.003 0.003 0.037 0.039 0.027 0.048 Norway 0.036 0.022 0.030 0.095 0.279 0.001 0.030 0.013 0.099 −0.783 0.009 0.082 0.070 0.046 0.050 Spain 0.012 0.009 0.008 0.032 0.097 0.000 0.012 0.008 0.047 0.003 −0.336 0.017 0.007 0.016 0.010 Sweden 0.017 0.012 0.032 0.066 0.166 0.000 0.015 0.003 0.078 0.004 0.002 −0.432 0.033 0.026 0.033 Switzerland 0.023 0.017 0.024 0.086 0.282 0.001 0.028 0.006 0.093 0.004 0.003 0.036 −0.278 0.029 0.049 UK 0.014 0.008 0.013 0.038 0.106 0.000 0.015 0.004 0.053 0.002 0.002 0.024 0.018 −0.321 0.030 Notes: Each cell contains the elasticity of the share of patent applications in the country indicated in column 1 when the tax rate changes in the country in row 1. Numbers are calculated using tax rates and patent applications in 2005. data is 10% in Ireland, whereas two of the Patent Box rates are below this by a statistically and economically signiﬁcant amount. The results with level, and so are outside the observed range of taxes in our data. high quality patents are similar. We carry out the simulation on the full set of patent applications, In columns (6)–(8) of Table 5.2 we consider the impact on tax shown in the top panel. It may be the case that many patents do not revenue from income derived from patents. These combine two effects. earn much income, and so in the bottom panel we carry out the simula- The reduction in the statutory tax rate will reduce revenue, but the tion using only the high quality patents, under the assumption that increase in the share of income from patents will increase it. We these are the patents that are expected to earn the highest income. demonstrate the impact on tax revenue by computing the product of The estimates suggest that the location of these patents were on average the statutory tax rate in each country and the share of patent applica- more sensitive to tax. tions. We index this to 100 before the introduction of any Patent The ﬁrst column shows the actual share of patent applications in Boxes. In the upper panel of the table we assume that all patents are each location in 2005 (prior to the introduction of Patent Boxes). The equally valuable, and that the relationship between legal ownership second column shows the predicted share of patent applications in and taxable income is not affected by the reform. All countries experi- each location after the introduction of Benelux Patent Boxes. The ence a decline in revenue. Although the countries that introduce Patent standard error of these predicted shares are shown in parenthesis. The Boxes attract more new patents, the increased share is not sufﬁcient to third column expresses the % change from column 1 to column 2. The outweigh the effect of the lower tax rate. With all four Patent Box introduction of Patent Boxes in the Benelux countries leads to a large policies in place, revenues are less than half of their previous levels in and statistically signiﬁcant increase in the share of new patents whose these countries. Ernst et al. (2013) provide evidence that lower rates legal ownership is located in Belgium and Netherlands. The increase in of tax on patent income attract particularly innovative projects with Luxembourg is proportionally large, but is not statistically signiﬁcant. high earning potential. In the lower panel we consider the effect on There is no change in the share in the UK. There is a decline in the revenue when we consider only high quality patents. The picture is share of patent applications located in other non-Patent Box locations here is similar; the introduction of Patent Boxes results in a substantial (not shown). reduction in revenues. The fourth column shows the predicted shares after the introduction of the UK Patent Box (in addition to the Benelux Patent Boxes). The ﬁfth 6. Summary and conclusion column shows the % changes from column 1 to column 4. The UK Patent Box leads to a reduction in the share of new patent applications made by The literature has emphasised the downward pressure on corporate subsidiaries located in the Benelux countries, but for Belgium and the income tax rates that arises from factor mobility. There is also a large Netherlands they still have a statistically signiﬁcantly higher share literature that discusses the strategies ﬁrms use to shift income for tax than prior to the introduction of any Patent Boxes. The share of new purposes and to circumvent anti-avoidance rules, and that highlights patent applications made by subsidiaries located in the UK increases an important role for intangible assets. However, we know relatively Table 5.1 Patent Box regimes. Applicant country Year introduced Patent Box Effective rate (1) (2) (3) Belgium 2007 Applies to gross income from patents and supplementary certiﬁcates 6.8% Luxembourg 2008 Applies to net income from patents and some other forms of intellectual property 5.9% Netherlands 2007 Applies to net income from patents and some other forms of intellectual property. Policy substantially broadened in 2010. 10% United Kingdom 2013 Applies to net income derived from patents and similar types of intellectual property. 10% Notes: Effective Patent Box rates are those that were in place when the policy was ﬁrst introduced. Each policy is associated with criteria that deﬁne which income is eligible. All policies include licence and embedded income. Policies differ in the conditions under which acquired intellectual property is eligible. Net and Gross refers to development costs. A number of other European countries now also operate policies akin to a Patent Box, and a similar policy has been proposed in the US. 22 R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 Table 5.2 Impact of Patent Boxes on location of and tax revenue raised from new patents. Share of new patent applications Tax revenue Prereform Benelux Patent Boxes % change UK Patent Box % change Prereform Benelux Patent Boxes UK Patent Box All patents Belgium 2.39 3.53 47.6% 3.42 −3.1% 100 30 29 (0.34) (0.34) Luxembourg 0.33 0.60 83.9% 0.56 −7.0% 100 36 33 (0.34) (0.34) Netherlands 7.92 12.51 58.0% 12.16 −2.8% 100 50 49 (0.38) (0.38) UK 4.15 4.15 −0.1% 5.25 26.6% 100 100 42 (0.31) (0.32) High quality patents Belgium 1.90 3.29 73.2% 3.17 −3.9% 100 35 33 (0.35) (0.35) Luxembourg 0.42 0.76 82.8% 0.71 −7.3% 100 35 33 (0.35) (0.35) Netherlands 7.00 12.48 78.3% 12.09 −3.1% 100 57 55 (0.39) (0.38) UK 4.89 4.39 −10.2% 5.64 28.5% 100 90 38 (0.31) (0.33) Notes: The top panel provides numbers based on all patent applications; the bottom panel provides numbers on high quality patents only. Column 1 shows the actual share of patent applications in each location; columns 2 and 4 give the predicted share of patent applications in each location following the introduction of the Benelux Patent Boxes and following the additional introduction of the UK Patent Box. Standard errors are in parenthesis. Column 3 and 5 show the corresponding percent changes in shares relative to column 1. The ﬁnal three columns show revenue raised from new patents, assuming that all patents have equal expected values. Revenues are indexed to 100 in the pre Patent Box period. Numbers are based on simulations using data for 2005. little about the extent to which the location of intangible assets would allow us to directly estimate this margin, but this would be responds to tax. The evidence there is on the impact of tax on the an interesting avenue for future research. location of capital more generally has tended to suffer from the imposi- The introduction of Patent Boxes by several European countries in a tion of restrictive a priori assumptions placed on the underlying model relatively short space of time has given rise to concerns that countries of ﬁrm behaviour. From a policy perspective it is clearly important to are engaging in tax competition for patent income. In future work we understand how responsive ﬁrms are to corporate income taxes when intend to build on the framework developed here to consider whether they make location decisions. governments are engaged in a strategic game to attract income from In this paper, we estimate a model of ﬁrms' decisions over where to intellectual policy that ultimately will continue to exert downward locate the legal ownership of their patents. We ﬁnd that corporate tax pressure on corporate taxes. rates are an important determinant of location choice. We extend the current literature on the determinants of ﬁrm location choice by estimating a ﬂexible choice model, which accounts for both observed Acknowledgments and unobserved heterogeneity in behaviour. We are able to generate own and cross tax elasticities across locations that capture complex The authors would like to thank Alan Auerbach, Richard Blundell, patterns of substitution in the data. The model can be used to conduct Michael Devereux, Roger Gordon, Jim Hines, Tim Schmidt-Eisenlohr, ex ante analysis of policy changes. We ﬁnd that this heterogeneity is Joel Slemrod and seminar participants at IFS, University of Michigan, important for explaining location choices. Centre for Business Taxation at the University of Oxford and Our model also shows that other factors inﬂuence where ﬁrms Stockholm University for useful comments. We thank International choose to hold legal ownership of patents. For instance, ﬁrms are Tax Policy Forum (ITPF) and the Economic and Social Research Council more likely to locate patent ownership in countries where they have (ESRC) for ﬁnancial support under the Centre for the Microeconomic associated real innovative activity. This may reﬂect co-location external- Analysis of Public Policy (CPP) at the Institute for Fiscal Studies and ities, or the inﬂuence of tax rules which seek to limit the extent to which under grant RES-000-22-4268. income and real innovative activity can be geographically separated. Firms also value other non-tax location characteristics. Such factors, along with tax rules like the operation of CFC regimes that limit the References tax advantages of locating patent ownership in low tax jurisdictions, Abramovsky, L., Grifﬁth, R., Macartney, G., Miller, H., 2008. Globalisation and innovation. help explain why we do not see ﬁrms choosing to hold all legal IFS Working Paper, W08/10. ownership of patents in the lowest tax locations. Altshuler, R., Grubert, H., 2006. Corporate Taxes in the world economy: reforming the taxation of cross-border income. Departmental Working Papers 200626. Rutgers We use the model to consider the impact of the recent introduction University, Department of Economics. of preferential tax regimes for income from patents. These Patent Boxes Becker, J., Fuest, C., 2007. Quality versus quantity — the composition effect of corporate are likely to attract patent income, but our estimates suggest that taxation on foreign direct investment. CESifo Working Paper Series 2126. Bernard, A., Redding, S., Schott, P., 2007a. Comparative advantage and heterogeneous they will also lead to substantial falls in tax revenues. Of course ﬁrms. Rev. Econ. Stud. 74 (1), 31–66. some of this revenue loss might be offset by gains from attracting Bernard, A., Redding, S., Schott, P., 2007b. Firms in international trade. J. Econ. Perspect. 21 activities that yielded positive externalities; these would need to (3), 105–130. Berry, S., Haile, P., 2010. Nonparametric identiﬁcation of multinomial choice de- be takenintoaccountina calculationofthe welfareimpactof mand models with heterogeneous consumers. Cowles Foundation Discussion the policy. It is also possible that the tax reforms will affect ﬁrms' Paper 1718. decisions over whether to apply for a patent on a new technology Berry, S., Levinsohn, J., Pakes, A., 1995. Automobile prices in market equilibrium. or whether to rely on secrecy. We do not have information that Econometrica 63 (4), 841–890. R. Grifﬁth et al. / Journal of Public Economics 112 (2014) 12–23 23 Berry, S., Levinsohn, J., Pakes, A., 2004. Differentiated products demand systems from Grubert, H., Mutti, J., 2009. The effect of taxes on royalties and the migration of combination of micro and macro data: the new car market. J. Polit. Econ. 112 (1), intangible assets abroad. In: Reinsdorf, M.,Slaughter,M.(Eds.),International 68–105. Trade in Services and Intangibles in the Era of Globalization. University of Blundell, R., Grifﬁth, R., Van Reenen, J., 1999. Market share, market value and innovation Chicago Press. in a panel of British manufacturing ﬁrms. Rev. Econ. Stud. 66, 529–554. Hall, R., Jorgenson, D., 1967. Tax policy and investment behavior. Am. Econ. Rev. 57 (3), Bohm, T, Karkinsky, T and Riedel, N (2012) “The impact of corporate taxes on R&D and 391–414. patent holdings” mimeo. Hall, B.H., Thoma, G., Torrisim, S., 2007. The market value of patents and R&D: evidence Bucovetsky, S., Hauﬂer, A., 2007. Preferential tax regimes with asymmetric countries. Natl. from European ﬁrms. NBER Working Paper No. 13426. Tax J. 52, 269–304. Hines, J., 1996. Tax policy and the activities of multinational corporations. NBER Working Cohen, N., 2012. Corporate governance law: ﬁrm heterogeneity and the market for Paper Series, Working Paper No. 5589. corporate domicile. SSRN Working Paper. Hines, J., 1999. Lessons from behavioural responses to taxation. Natl. Tax J. 52 (2), de Mooij, R., Ederveen, S., 2008. Corporate tax elasticities: a reader's guide to empirical 305–322. ﬁndings. Oxf. Rev. Econ. Policy 24 (4), 680–697. Hines, J., Jaffe, A., 2001. “International taxation and the location of inventive activity”.In: Desai, M., Foley, C.F., Hines, J., 2006. The demand for tax haven operations. J. Public Econ. Hines, J. (Ed.), “International Taxation and Multinational Activity”.ISBN: 90, 513–531. 9780226341736. Devereux, M., 2006. The impact of taxation on the location of capital, ﬁrms and proﬁt: a Hines, J., Rice, E., 1994. Fiscal paradise: foreign tax havens and American business. Q. survey of empirical evidence. Oxford University Centre for Business Taxation Work- J. Econ. 109, 149–182. ing Paper Series, WP 07/02. Karkinsky, T., Riedel, N., 2012. Corporate taxation and the choice of patent location within Devereux, M., Grifﬁth, R., 2003. Evaluating tax policy for location decisions. Int. Tax Public multinational ﬁrms. J. Int. Econ. 88, 176–185. Financ. 10, 107–126. Krautheim, S., Schmidt-Eisenlohr, T., 2011. Heterogeneous ﬁrms, ‘proﬁt shifting’ FDI and Dischinger, M., Riedel, N., 2011. Corporate taxes and the location of intangible assets international tax competition. J. Public Econ. 95 (1–2), 122–133. within multinational ﬁrms. J. Public Econ. 95, 691–707. Lanjouw, J., Schankerman, M., 2004. Patent quality and research productivity: measuring Ernst, C., Spengel, C., 2011. Taxation, R&D tax incentives and patent application in Europe. innovation with multiple indicators. Econ. J. 114, 441–465. ZEW Discussion Paper No. 11–024. Lipsey, R., 2010. Measuring the location of production in a world of intangible productive Ernst, C., Richter, K., Riedel, N., 2013. Corporate taxation and the quality of research and assets, FDI, and intraﬁrm trade. Rev. Income Wealth 56, 99–110. development. ZEW Discussion Paper No. 13–010. Melitz, M., 2003. The impact of trade on intra-industry reallocations and aggregate indus- Evers, L., Spengel, C., Miller, H., 2013. Intellectual Property Box regimes: effective tax rates try productivity. Econometrica 71 (6), 1695–1725. and tax policy considerations. ZEW Discussion Paper, No. 13–07. NESTA, 2009. Innovation Index 2009; innovation, knowledge spending and productivity Finnerty, C., Merks, P., Petriccione, M., Russo, R., 2007. Fundamentals of international tax growth in the UK. planning. In: Russo, R. (Ed.), ISBN: 978-90-8722-016-7. Nevo, A., 2001. Measuring market power in the ready-to-eat cereal industry. Ginarte, J., Park, W., 1997. Determinants of patent rights: a cross-national study. Econometrica 69 (2), 307–342. Res. Policy 26, p283–p301. OECD, 2006. Report on the attribution of proﬁts to permanent establishments, Parts Graham, J., Tucker, A., 2006. Tax shelters and corporate debt policy. J. Financ. Econ. 81, I (general considerations), II (banks) and III (global trading). Cent. Tax Policy 563–594. Adm. Grifﬁth, R., Miller, H., 2010. Support for research and innovation. In: Chote, R., Emmerson, OECD, 2012. OECD Factbook 2011–2012: economic, environmental and social statistics. C., Shaw, J. (Eds.), The IFS Green Budget: February 2010, IFS Commentary 112. Pakes, A., 1985. On patents, R&D and the stock market rate of return. J. Polit. Econ. 93, Grifﬁth, R., Miller, H., 2011. Corporate Taxes and intellectual property. In: Brewer, M., 390–409. Emmerson, C., Miller, H. (Eds.), The IFS Green Budget February 2011, IFS Commentary Pakes, A., 1986. Patents options: some estimates of the value of holding European patents 117. stocks. Econometrica 54, 755–784. Grubert, H., 2003. Intangible income, intercompany transactions, income shifting, and the Park, W., 2008. International patent protection: 1960–2005. Res. Policy 37, choice of location. Natl. Tax J. 56 (1), 221–242 (Part 2). p761–p766. Grubert, H., Mutti, J., 1991. Taxes, tariffs and transfer pricing in multinational corporate Train, K., 2003. Discrete Choice Methods with Simulation. Cambridge University Press. decision making. Rev. Econ. Stat. 73, 285–293. Wilson, J.D., 1999. Theories of tax competition. Natl. Tax J. 52, 269–304.

http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

Journal of Public Economics Unpaywall

http://www.deepdyve.com/lp/unpaywall/ownership-of-intellectual-property-and-corporate-taxation-twzv5OKdDj

Ownership of intellectual property and corporate taxation

Loading next page...

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher: Unpaywall
ISSN: 0047-2727
DOI: 10.1016/j.jpubeco.2014.01.009
Publisher site: See Article on Publisher Site

Abstract

Journal

Journal of Public Economics – Unpaywall

Published: Apr 1, 2014

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Ownership of intellectual property and corporate taxation

Ownership of intellectual property and corporate taxation

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Ownership of intellectual property and corporate taxation

Ownership of intellectual property and corporate taxation

References

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies