University Patenting in Wales, Scotland and Northern Ireland: a Comparative Analysis

Introduction

The creation of knowledge-rich economies is a global phenomenon impelled, in part, by the continued liberalisation of world trade, and reinforced and rewarded by the strengthening of world law in intellectual property rights. In many countries, universities have been recognised as important repositories and generators of knowledge of commercial value and their research ‘missions’ have been extended beyond the creation of outputs as freely available public goods. The new technologies and new ideas that are transferred out of a university tend to have a disproportionate impact on the institution's economic hinterland (Audretsch and Lehmann, 2005). Thus both national and regional governments and agencies have assumed an interest in the stimulation and transfer of commercially valuable university research.

This paper reports unique data that provide insights into the comparative extent and effectiveness of university patenting in three UK regions/nations: Wales, Scotland and Northern Ireland. These regions are of interest in themselves because they are three distinct elements of the UK devolutionary settlement (Cooke and Clifton, 2005) following the establishment of the Scottish Parliament and Welsh Assembly in 1999 (devolution in Northern Ireland has been in flux over this period, for well known political reasons). Over the coming decades, observers will be able to assess the extent to which the differing degrees of political and economic autonomy and styles of government succeed in delivering improvements in the economic performance of these peripheral regions. Devolution is, however, imposed on different cultures and historical legacies, economic, political, legal and intellectual. So the gathering of baseline data is critical to the running of this natural experiment.

The core of the paper consists of increasingly refined patent productivity comparisons for individual institutions – refined in the sense that the measures more effectively capture innovations of commercial value. The measures are, respectively: patent filings, patent grants, patents having a commercial co-assignee, and patent citations. A brief attempt is made in the final section to place the main findings into a broader context.

From a regional perspective, the significance of university patent outputs is that they provide information on an important component of the region's innovation system. An innovation system is the mix of formal and informal institutions and policies geared to the creation and exploitation of new knowledge, a concept widely applied at the regional level (Cooke, Heidenreich and Braczyk, 2004). Devolved government has allowed a degree of discretion in the development of the innovation systems in the three regions. They, nevertheless, remain nested within the national (UK) system and it will be useful to begin with an account of recent relevant policy developments at that level.

University patenting: the policy framework

In the UK, rights to intellectual property arising from university research were formerly granted to a government corporation, the British Technology Group, but were ceded to the universities themselves in 1983. Although this gave institutions greater freedom to develop intellectual property portfolios, it was not until the mid-1990s that patenting activity ‘took off’ (as measured by patent applications; Beale, 2005). In this respect, the UK has lagged a considerable way behind the USA. There, universities have generated rapid growth in patents following the 1980 Bayh-Dole Act which allowed universities to patent findings that were generated from federally funded research (Jaffe, 2000).

Concerns about the potential conflicts between commercialisation and the traditional missions of universities (see, e.g., Owen-Smith and Powell, 2001; Poyago-Theotoky, Beath and Siegel, 2002; Stephan et al., 2007) have been common to both countries and it is more likely that the tardy UK response has been the result of weaker incentives for academic staff to become involved in commercial activity. That, at least, seems to be one of the basic premises of a series of recent policy reviews on university business links.

A report on the commercialisation of public-sector science (National Audit Office (NAO), 2002, pp. 22) noted that ‘insufficient “kudos” attaches to filing a patent’, and accordingly recommended the strengthening of incentives. Building on the Baker Report (HM Treasury, 1999), it advocated that research establishments (either the discipline-based Research Councils or those, primarily universities, in receipt of research council funding) should aim to develop, either individually or in collaboration, a critical mass of intellectual property so as ‘to take advantage of the incremental nature of most research’ (NAO, 2002, pp. 28). The Report's plea for stronger incentives was answered in the Science and Innovation Framework, 2004–2014 (HM Treasury, 2004), which led to the Higher Education Innovation Fund for England and Northern Ireland, accompanied by a similar boosting of third-mission funding in Scotland and Wales.

Of course, the creation and exploitation of intellectual property is not the only avenue for university business technology transfer. Student sponsorships, research contracts and consultancy, and business spin-outs are alternatives. Nevertheless, official reports put considerable emphasis on intellectual property. The Lambert Review of business-university collaboration (HM Treasury, 2003) questioned the quality, longevity and employment-generating potential of university spin-outs and recommended greater emphasis on licensing as a means of commercialisation.

This stress on intellectual property has not been uncontroversial. In their overview of European university patenting, Guena and Nesta (2006) single out the NAO (2002) report for its uncritical advocacy of the intellectual property route. They argue that patent success is highly skewed, thanks to a small number of very successful patents, and that most university technology transfer offices in the USA and Europe earn little or no income from the sale or licensing of intellectual property. The uneven commercial success of patents was, though, one of the reasons why the NAO report recommended the creation of institutional portfolios of sufficient mass. As for the overall returns to intellectual property, the Higher Education-Business and Community Interaction Survey, 2004–2006 (Higher Education Funding Councils, 2007) reports that between 2000–2001 and 2005–2006, aggregate annual revenues of UK universities varied between £40 and £58 million, compared to costs of between £13 and £17 million. A breakdown by standard statistical region for 2005–2006 shows revenues exceeding costs for Wales and Scotland and for all but one English region (the North East). For Northern Ireland, however, revenues were considerably less than costs.

The implication is that potential rewards are available from patent assets but thought needs to be given to their effective management; in particular, applications need to be pursued on a carefully selected basis. Thus, without denying the importance of other forms of technology transfer, a comparative analysis of university patent holdings is of interest, not only to individual universities as a means of benchmarking performance, but also to regional governments aiming to exploit university generated knowledge as part of their knowledge-economy and science policy strategies. These strategies, following UK policy imperatives all stress the importance of university technology commercialisation (see, e.g., Scottish Executive, 2001a, b; Welsh Assembly Government, 2004, 2006; Hewitt-Dundas, Roper and Love 2007).

The data

The Higher Education-Business and Community Interaction Surveys contain aggregated data on universities' self-reported intellectual property holdings from 2000–2001. The data presented in this paper come from a different source. They were commissioned from MicroPatent Professional Services by IP Wales, a Welsh Assembly Government business-support initiative (Beale, 2005). The data relate to patent filings and patent grants in the years 1983–2005 for all higher education institutions in Wales, Scotland, Northern Ireland and the Republic of Ireland and for all (16) English members of the self-selected ‘Russell Group’ of leading research universities. For the purposes of this paper, these English universities are generally excluded since they are not representative of overall university performance in England (though occasional benchmark comparisons are made to the 15 of these having science and technology faculties). Universities in the Republic of Ireland are also generally excluded from detailed consideration because other aspects of the comparison (notably staff counts) are not obtainable on a consistent basis.

Unlike the patent information published by the Higher Education Funding Councils (2007), the present data are reported for individual institutions, with separate counts for the five main national and international patent offices. Patents with commercial co-assignees are individually identified, and supplementary information is provided on patent classes and citation rates. The five offices are: Europe (EPO), Great Britain (UKPO), Japan (JPO), United States (USPTO) and the World Intellectual Property Organisation (WIPO) via the Patent Cooperation Treaty (PCT). Patent applications were not separately published by USPTO prior to 2000; thus for 1983–1999 only patent grants are counted. It is perfectly possible to apply for a patent in individual countries other than those stated but such applications are excluded from the data. However, patents granted by EPO, JPO, USPTO and via the PCT are generally regarded as high-value patents. Publication of documents typically lags behind filings by 18 months and, for this reason, the data, which were collected in 2005, are incomplete for the period 2003–2005.

It has been claimed (e.g., Saragossi and van Pottelsberghe de la Potterie, 2003) that university patent holdings underestimate the number of patented inventions generated by universities. The university portfolios in Beale (2005) were assembled by including documents on which the institutions were named as an assignee (owner) or co-assignee and documents that, regardless of the assignee name, were ‘family members’ of the university assigned documents. In principle, a family consists of a priority application (the first application to be filed) and subsequent applications in different patent offices on the same invention (citing the priority). However, as Dernis, Guellec and van Pottelsberghe (2001) explain, matters can be rather complicated in reality, with applications citing multiple priorities or different applications citing the same priority, both suggestive of a degree of ‘downstream’ development of the original invention. The definition used here is due to International Patent Documentation Center (a database now maintained by EPO), in which all patents with common priorities are regarded as belonging to a single family. When a document within a family was not assigned to a university, its probable institutional source was inferred by examining the rest of the family members. Family expansion increased institutional patent attributions by 23 per cent, going some way towards meeting the concerns of Saragossi and van Pottelsberghe de la Potterie (2003).

Patents are a measure of technology outputs. To compare productivity, an input measure is needed. This is taken to consist of counts of academic staff in schools/departments in the sciences and engineering in the research-oriented ‘pre-1992’ universities, that is, those institutions in possession of a university charter before the abolition of the ‘binary divide’ in 1992 which allowed polytechnics to claim university status. These counts are obtained from submissions to the 2001 Research Assessment Exercise (Higher Education Research Opportunities, 2002; the Research Assessment Exercise is undertaken at 5–7 year intervals as a means of informing the distribution of public research funds to universities). The pre-1992 universities all have an explicit research mission and submit returns in all, or nearly all, relevant Research Assessment Exercise ‘units of assessment’, or disciplines. This is not true of most of the post-1992 universities, or former polytechnics, whose comparatively weak research capabilities lead them to concentrate submissions in the few areas where they hope to make an impact. Thus, entire schools/departments may fail to leave any trace in the Research Assessment Exercise returns. Aggregate regional outputs, but not inputs, of post-1992 universities and other higher education institutions are reported below. Of the 68 units of assessment, 27 are considered here to qualify as science and engineering and, therefore, to be patent-relevant.

One of the problems of using the Research Assessment Exercise is that only the staff deemed to be ‘research active’ is submitted. For the purpose of evaluating university patent productivity it is preferable to include all staff (in fact, even the many Research Assessment Exercise rankings published in the UK education media are adjusted for the proportion of staff submitted). The returns do not give the exact proportion of research active to total staff but do give proportions within (six) bands. By using the midpoints of these bands, the reported numbers can be scaled up to yield an estimate of the total staff complement for each unit of assessment for each institution. These estimates are reported in Table 1 for Wales, Scotland and Northern Ireland on an absolute and per capita basis (using the 2001 Census of Population).

These estimates are of interest in their own right as indicators of the university based innovative potential in each region and, in that respect, Table 1 tells a striking story. The number of academic staff based in the science and engineering departments of pre-1992 universities relative to population is nearly twice as high in Scotland as it is in Wales and Northern Ireland. When adjusted for quality (by weighting staff according to the appropriate Research Assessment Exercise ranking) there is a marginal change – in Scotland's favour: research mass there is over twice as high as it is in Wales. Whatever the patent productivity measures show, the fact is that Welsh universities have been attempting to compete on the basis of half the resources of Scotland. This picture is unlikely to be distorted as a result of the exclusion of the post-1992 universities, as is implied by the fact that these institutions account for 8.5 per cent of Scottish patent filings compared to 4.5 per cent for Wales and zero for Northern Ireland.

Patent filings

For Scotland, there is considerable variation in filings productivity, with Strathclyde, at 1.1 filings per member, markedly ahead of the other institutions (Table 2). Unlike Aberdeen, Dundee, Edinburgh and Glasgow, Strathclyde does not have a medical school but it does have a pharmacy department (for comparison, filings per member of the English Russell Group universities is 0.59). There is a substantial difference in the performances of Scotland's two Russell Group members, Edinburgh and Glasgow. Edinburgh, which has Scotland's largest concentration of science and engineering academics, has a filings productivity barely above that of the two smallest groupings (among pre-1992 universities), St. Andrews and Stirling. Among the new institutions, Glasgow Caledonian and Napier College accounted for 65 of the 129 filings. Only 1 per cent of applications were filed in Japan and only 10 per cent at UKPO (presumably because UK protection can also be had via EPO and WIPO).

For Wales, there is a clear-cut difference between Cardiff University (including, for our purposes, the University of Wales College of Medicine with which it is now merged) and the other three pre-1992 universities. Cardiff accounted for 375 (74 per cent) of the 507 filings from Welsh institutions and has a productivity rate twice as high as its neighbours. In post-1992 institutions, 22 of the 27 filings by came from Glamorgan University. Thanks to Cardiff, Wales's overall old-university productivity exceeds that of Scotland: 0.52 per member compared to 0.44. The distribution of patent office applications is broadly similar to that of Scotland except for a greater concentration in UKPO, possibly reflecting less confidence in the global value of the innovations.

For Northern Ireland, there is little difference between Belfast and Ulster, except for scale. The overall rate of 0.27 filings per member is comparatively low. There is also a greater propensity to use multi-country routes (EPO and WIPO).

For two of the regions, Wales and Scotland, information is also available, at the regional level, on the types of invention that are being patented. As a route to commercialisation, patents are particularly favoured in the medical sciences and especially pharmaceuticals. Scotland is well endowed with medical schools and pharmacy departments whereas (at the time the data were collected) Wales had just one of each (both at Cardiff). Interestingly, this does not seem to be reflected by the distribution of patents across the International Patent Classification class levels. Of the 110 classes, eight account for over three quarters of the filings in the dataset (Table 3). While Wales trails Scotland and England in filings per researcher in class A61, which includes pharmaceuticals, it performs well in organic chemistry (C07) and biochemistry (C12, which includes genetic engineering). In relative terms, Wales's weakness is most pronounced in sections G and H, electronics and communications.

Patent grants

Patent filings may be a useful measure of the level of commercially oriented research activity, but their utility as a measure of the value of outcomes of that activity is compromised by the fact that application success rates may vary from institution to institution. The data provide some insights into these variations. For EPO, it is possible to compare filings over the period 1983–2005 with the number of grants. This is not the same as showing how many applications over that period went on to be granted since recent applications may be granted after 2005. Nevertheless, given the length of the period, the ‘issuance ratio’ (the ratio of patents granted to applications not granted) gives a good indicator of overall success (Table 4). The same can be done for US data but only for the years 2000–2005, the shorter period giving a less robust measure of success.

For some institutions, the small numbers involved suggest that the issuance rates should be treated with caution. Among the more active institutions, Strathclyde again stands out. Not only does it have the highest filings rate but it also (on the basis of EPO and USPTO data) has the highest issuance rates. The overall rates for Wales and Scotland are similar, though those for Northern Ireland are markedly lower. It is interesting that issuance rates are higher for US applications than those for EPO, suggesting either that US criteria are easier to satisfy, or that they are subject to shorter lags, or both.

The ranking of institutions based on the total number of patents at EPO and USPTO for 1983–2005 and the corresponding (granted) patent productivity rate is close to that based on filings productivity; suggesting that, as cross section data, filings may not, in fact, be a poor measure of outcomes (Table 5). This may not be true over time, as universities become more experienced and more selective about what inventions to patent. (For comparison, the productivity rate for the elite English universities is 0.14.)

Patent collaboration

Universities can develop patents on their own, in collaboration with industrial partners, or in collaboration with other universities and public research bodies. Evidence from Belgian universities (Sapsalis and van Pottelsberghe, 2007) suggests that having a commercial co-assignee has a weakly significant positive effect on patent value, while co-assignment with other universities and public research organisations has a strongly significant positive effect.

The present data show considerably more instances of commercial (compared to intra-university) collaboration and these are discussed first. Commercial cooperation in patent activity was low in the early years for most institutions (Table 6; and the data for 1983–1992 are compacted in the table) but has grown steadily on average since then. Table 6 attributes to each institution patents with commercial assignees and co-assignees that have been identified according to the patent-family definition, and illustrates how performances have changed over time. The average, however, conceals idiosyncratic patterns for a few institutions, casting fresh light on the outcomes related above. The last three columns of Table 6 show: the total count for the entire period; the count for the period end (2000–2005); and the ratio of end count to total count. The ratio gives a rough indication of whether or not institutional activity is accelerating or declining relative to other institutions.

Perhaps the most surprising results are those for Strathclyde. It was noted above that this university has high total filings and high EPO/USPTO patent grants; in both cases higher than Glasgow. Yet, as reported in Table 6, its portfolio of patents with commercial co-assignees, for the entire period, is somewhat smaller than that of Glasgow and Cardiff and only a little larger than that of Dundee. Yet more remarkable is that most of Strathclyde's documents refer to years prior to 2000; only 24 per cent are post-2000 (about half the Scottish average). Note also that the previous ranking of Edinburgh and Glasgow is reversed when it comes to the end count: total count ratio. Three quarters of Edinburgh's patents with commercial co-assignees are post-2000 compared to fewer than half for Glasgow. The late starters catching up is also evident from the results for the post-1992 Scottish institutions, for which 82 per cent of such patents relate to the last 5 years.

Cardiff, like Strathclyde, also seems to have stalled drastically. Over the entire period, Cardiff outperforms all other institutions (Table 6). Yet only seven of its 45 documents relate to the end period. It might be thought that the fall off in commercial collaboration at Cardiff, Strathclyde (and also Aberdeen) reflects a change in commercialisation strategy but, if so, it goes very much against the trend of the leading UK research universities. The end count: total count ratio for the English Russell group, as a whole is 0.63, while that for the three leading UK science and technology universities (Oxford, Cambridge and Imperial College) is 0.76.

The figures for Northern Ireland are dominated by Belfast which has much in common with Edinburgh and the post-1992 Scottish institutions and is pretty much a mirror image of Cardiff: 80 per cent of its patent attributions are from the end period. For the last 5 years the total count for the region exceeded that of Wales.

Insofar as any slowing down of commercial collaboration in patenting is a problem, it is clearly more of a concern for Wales. The fall in Cardiff's numbers has not been made good by the remaining universities. To be on a par with Scotland in per capita terms, Welsh outputs need to be at least half of those for Scotland. Yet for this kind of patent, the ratio stands at about one third for the entire period and around a quarter for the last 5 years.

When it comes to collaboration with other universities, a similar picture emerges. The patent filings documents reveal 10 instances of co-assignment between universities in Scotland and 34 between Scottish universities and English members of the Russell Group (including six with triple assignment). There were, by contrast, zero intra-Wales collaborations and only seven (all from Cardiff) with other Russell Group universities. For Northern Ireland, the corresponding figures are zero and two. There was no evidence of collaboration between universities in Northern Ireland and the Republic of Ireland.

Patent citation analysis

When a patent or application is published it may serve as a reference for inventions that are subsequently filed by the same inventor or others. If an inventor or an examiner makes reference to an earlier patent document, some technical relationship is implied, but the nature of that relationship may vary. The earlier work may be similar or it may be a different solution to the problem that is addressed by the new application. If a patent has had a significant impact on a technical field, the number of times it is cited by others will tend to be higher. Self-citation (by the inventor or assignee) implies continued research and development investment in the technical area. Citation by others implies some recognition of technical reliance on or advancement over the prior work.

Naturally, one would expect citation frequency to increase with the age of the patent since the opportunity for citation increases with time. It is also the case that citation practices vary across patent offices, with much higher frequencies being recorded on USPTO documents. Since the percentages of filings in the USPTO are very similar for the three nations/regions (Table 2), this effect can be ignored. Citations data have been compiled for EPO, USPTO and WIPO documents as part of an analysis of the full set of documents including those of English Russell Group and Republic of Ireland universities. The mean citation rate for the entire collection is 0.25 citations per patent. A simple summary comparator is provided by the proportion of documents having citations in excess of two standard deviations above the mean. Scottish documents at 3 per cent are above the two standard deviation threshold compared to 4 per cent for Wales and 1 per cent for (All-) Ireland.

If this seems good from a Welsh perspective, more troubling is that the citation frequency in that region has been consistently low from 2000 onwards. Although one expects lower average frequencies for newer documents, among the newer documents higher rates do predict innovations of emerging interest. What is apparent is that both Scottish and (All-) Irish universities are generating documents of emerging interest while Welsh universities are signally failing in that regard. Although mean citation rates for this restricted period are close to zero for all three regions, the two standard deviation markers are 3.9 for Scotland, 2.1 for Ireland and only 0.7 for Wales. This echoes the remarkable downturn, noted in Table 6, in patent families with commercial co-assignees at Cardiff compared with, say, Belfast and Edinburgh.

Implications and conclusions

This paper's focus on patents is not intended to deny the importance of other means of transferring ideas from universities or of obtaining revenue streams from business. Even so, the brief review of UK technology policy makes clear the importance central government places on commercialisation through the development of intellectual property rights, and this is echoed in the economic strategy and science policy documents of the devolved administrations.

The significant differences to emerge from the data, at least between Scotland and Wales, relate not to the main quantity measures relative to the number of researchers (filings and grants) but to quality measures (gross numbers of patent collaborations, the growth of commercial collaborations, and recent citations) and they echo findings on patenting from industry itself. Moore and Mainwaring (2006) show that, in the production sector, Welsh firms are at least as likely to engage in patenting as their Scottish counterparts but are less likely to hold patent clusters that typically arise from innovation routes with ongoing potential. For Northern Ireland, not only are the quantity measures poor but the very low rate of conversion of applications into grants is a matter of particular concern (see also, Hewitt-Dundas, Roper and Love, 2007). Recall that Northern Ireland is one of the few UK regions not to profit from the commercialisation of university intellectual property. It is very likely that these poor returns to commercialisation are closely connected to failure to reach the grant stage, and implies a rather indiscriminate approach to application selection. There is some suggestion of recent growth in outputs at Belfast but, overall, the picture is discouraging.

The quantity comparisons become less favourable to Wales and Northern Ireland when they are related to total population rather than researcher numbers and reflect a historical legacy of high investment in higher education in Scotland (relative to the whole of the UK). In this respect, Scotland has been favoured by the application, since 1979, of the Treasury's ‘Barnett formula’ for allocating government expenditure to Wales, Scotland and Northern Ireland. According to Mackay (2006), Scotland receives 7 per cent more per capita than Wales, even though its gross value added per capita is 17 per cent higher. Northern Ireland's even more favourable treatment is a result of its political ‘troubles’. The number of universities in Scotland also appears to help maintain stability in patenting performance, with emerging weaknesses in some institutions being offset by growing strengths in others. In Wales and Northern Ireland, the dominance of Cardiff and Belfast are potential sources of volatility.

In assessing these findings for the three regions, it is important to take account of the different sizes of the regional economies and the number of universities they are host to. In relation to the first, university technology spillovers tend to be localised but transfer can only be effected if the local economy is capable of absorbing the inventions. It has been suggested (Hewitt-Dundas, Roper and Love, 2007) that business expenditure on research and development is a good proxy for the capacity to absorb university technology outputs. In 2004, business spending on research and development in Scotland was £494 million, over twice that in Wales (£226 million) and over four times that in Northern Ireland (£116 million; Statswales, 2007). A small private research base is unlikely to sustain sufficient mass in each of the disciplines offered by (even a small number of) universities to benefit from the inventions of those disciplines.

On the other hand, where absorptive capacity is high, a larger university research base is beneficial in allowing the development of large, risk-spreading patent portfolios, as recommended by the NAO (2002). Indeed, that report suggests that institutions ‘may benefit from combining to create a significant body of intellectual property . . . in the same, or similar, market or geographical sectors’. Not only is this an option more readily available to universities in Scotland on account of the sheer size of its higher education sector, it is a route that they seem more eager to pursue, on the evidence of intra-university patent collaboration as well as other co-operative projects such as the web-portal ‘university-technology.com’ set up by all 13 universities to promote licensing opportunities.

Scotland does, of course, have mature national institutions like the Royal Society of Edinburgh, to promote the interests of higher education and help cement cooperation, but it might have been thought that something similar would have happened in Wales as a result of universities' membership of the federal University of Wales. However, even before the weakening of that body through the secession of Cardiff, it betrayed no willingness or ability to help formulate an all-Wales research strategy, let alone coordinate the development of multi-institution patent portfolios and promote collaborative commercialisation. Wales's new science policy (Welsh Assembly Government, 2006) stresses the need for improved commercialisation of university research but offers little strategic insight into how this might be done. Given the relative smallness of its higher education sector and three of its four pre-1992 universities, there is surely a case for adopting the NAO recommendation to create a combined body of intellectual property. For Northern Ireland, options have been more limited on account of its size and recent troubled history. It may now become possible for its universities to develop stronger collaborative links to their counterparts in the Republic and to take advantage of the recent rapid growth of the Irish economy.