Understanding Trends in Industrial Research Performance in Canada Ron Freedman, Partner, The Impact Group

Transcription

1 Understanding Trends in Industrial Research Performance in Canada Ron Freedman, Partner, The Impact Group July 2008 Introduction The five year period since the tech bust of 2001 saw an explosion in the number of Canadian companies performing R&D (research and development) (Figure 1). Statistics Canada reports that in Fiscal 1994 a total of 11,132 companies performed research. By Fiscal 2004 the number of firms that were conducting R&D rose dramatically - to 17,222 - an increase of over 54% during the period. What does this increase represent? Have corporate leaders discovered the importance of research and put their firms on a new path toward innovation and competitiveness, or is something else happening? Figure 1. Number of Industrial R&D Performers, ,450 4,959 3,695 3,663 13,363 12,087 11,132 10,771 9,804 9,648 9,783 9,967 10,849 17,222 15, Understanding the answer requires us to know something about how the official statistics on industrial R&D are collected. StatCan conducts an annual survey of the 1,000 largest known R&D performers, in order to collect in-depth information about their activities. But this survey accounts for only about 6% of the total number of companies that apparently conduct research. Data on the other 94% of firms come to StatCan through the Canada Revenue Agency (CRA) 1. 1 Formerly called Revenue Canada. 1

2 CRA provides StatCan with data on the companies that successfully apply for an SR&ED tax credit 2. Starting in 1994, StatCan began to use these data to supplement the survey that it administers, in order to develop a more complete picture of industrial R&D. Today, the total of the StatCan survey responses plus the CRA data yields the picture on industrial R&D activity across the country. Thus, the largest driver of the official statistics (accounting for 94% of the total 3 ) is the number of companies claiming and receiving approval for their SR&ED tax credit claims. The system is not theoretically perfect, because companies that perform research might not submit a tax credit claim and will thus be absent from the official records. However, these missing firms tend to be smaller companies spending small amounts on research, so their absence does not affect the overall spending results very much, although they undoubtedly influence the data on the number of spenders. In any event, the Canadian system of measuring corporate R&D is vastly superior to that of almost all other countries; other countries largely rely on surveys that fail to count many more firms than StatCan and CRA do together. A further benefit of the Canadian system is that CRA exercises a degree of due diligence over the submitted claims, so that ineligible R&D claims are presumably denied. StatCan counts only the successful CRA tax credit claimants. Thus, the number of firms performing R&D in Canada is the product of two factors: the number of firms applying for SR&ED tax credits multiplied by the approval rate of firms by CRA. If more or fewer companies apply for tax credits and the approval rate is stable, then the number of apparent R&D performers will rise or fall accordingly. In contrast, if the number of applicants is stable and the approval rate changes, then the apparent number of R&D performers also changes correspondingly. All things being equal, a higher approval rate increases the number of performers and a lower rate decreases the number. There is a high degree of churn in the population of R&D performers. A recent Impact Group study 4 revealed that even in the good years (pre tech-bust) the proportion of committed R&D firms - those that perform R&D on a consistent basis (in this case 6 or 7 years in a 7-year period) - totalled only 15.9% of all R&D performers (Figure 2, Table 1). Fully 56.9% of companies that conducted research did so for only 1 or 2 years out of a 7-year period, indicating that they were only occasional performers. 2 Scientific Research and Experimental Development. 3 In fact, the SR&ED component of the database is higher than 94% because not all of the 1,000 firms surveyed by StatCan return a completed survey. 4 The Demographics of Industrial Research in Canada, Toronto. January

3 % Performing Research Figure 2. Persistenceof R&D Performers # of Years Performing Research, Percent of Firms Cumulative Total Table 1. Number of Years Performing Research, (%) Years of R&D (Out of 7) Total Number of Firms 8,583 5,723 3,309 2,246 1,568 1,375 2,357 25,161 Percent of Firms Cumulative Total Cumulative Total Source: Statistics Canada. Special Analysis. Number of Companies = 25,161 Our earlier study concluded that a majority of firms that performed R&D - that is, successfully claimed R&D tax credits - were not committed R&D companies, but rather companies that were pursuing one- or two-time tax benefits. Using the same ratios to classify the 17,222 industrial R&D performers in 2004, would yield the following nominal breakdown of years of performance (Table 2): Table 2. Probable Distribution of 17,222 Fiscal 2004 R&D Performers Years of R&D (Out of 7) Total Percent of Firms Number of Firms 5,873 3,909 2,273 1,533 1, ,619 17,222 3

4 So, in this scenario, only a small number of the 17,222 R&D performers in 2004 (a total of 2,566) would be considered regular R&D performers, conducting research for 6 or 7 years in a 7-year period, (versus 9,782 companies that conducted research for 1 or 2 years only). Corporate R&D Performance Trends What has been happening with the number of companies applying for SR&ED tax credits and the number being approved? Data indicate three distinct patterns of application in the 15-year period from (Figure 3). Between 1990 and 1994 there was a strong rise in the number of SR&ED applicants - from 8,187 to 11,864 - an increase of 45%. From 1994 to 1999 there was a small drop in applicants - from 11,864 to 10,999l - a decline of 7%. From 1999 to 2004 the number of applicants exploded - rising from 10,991 to 20,132 - a jump of 83%. The number of successful applicants similarly jumped from 9,967 to 17,222. What is driving the huge increase in SR&ED applicants and claimants? 25,000 20,000 Figure 3. SR&ED Applications & Approvals, ,132 17,739 15,000 10,000 5, ,628 13,725 17,222 11,442 11,459 12,340 15,729 11,864 10,728 10,999 9,330 10,518 11,052 10,882 13,363 12,087 8,187 11,132 10,771 9,804 9,783 9,967 10,849 9,648 4,450 4,959 3,695 3, Applicants Approved First, we will address the question of successful applicants. Is it possible that a significant change in SR&ED approval rates caused the number of R&D companies to increase? Newly-released CRA data allow us to calculate the approval rate for SR&ED claimants (Figure 3). 4

5 100% 90% 80% Figure 4. Derived Approval Rate, % 88.7% 89.9% 87.9% 94.0% 89.9% 90.6% 85.5% 85.5% 88.1% 88.7% 70% 60% 50% 40% 45.1% 42.3% 39.3% 43.3% 30% 20% As indicated in Figure 4, the approval rate for CRA claims was 93.8% 5 in ; in other words 93.8% of all SR&ED applications were approved in that year. There was a slow but steady drop in the approval rate afterwards, such that by 2004 approvals were running at 85.5% of applications. So, while the number of companies applying for tax credits grew rapidly during the period (from 11,864 to 20,132), the percentage of applications being approved dropped somewhat. This is consistent with economic theory, which would postulate that as the number of applications expanded, their marginal quality declined. Apparently, the small decline in the approval rate was overwhelmed by the large increase in the number of claimants, and the the number of successful claimants duly rose. The decline in the percentage of applications being approved by CRA between 1994 and from 93.8% to 85.5% - makes some intuitive sense. Prior to this, a relatively stable number of companies was applying for tax credits (although the number slowly declined from 1994 to 1999). This implies a steady-state system with a reliable group of R&D performers. On balance, 5 The derived approval rate is the number of firms whose SR&ED applications were approved, as a percentage of the firms applying to the program. 6 Because StatCan only began to use CRA data in 1994, it is not possible to calculate the approval rate of SR&ED claims prior to that. 5

6 at the beginning of the period the quality of the claims being submitted to CRA was consistently high, with 93.8% of claims being approved. However, from the perspective of declining marginal returns, the quality of the next tranche of applications - the incremental applications - was not likely to be as high as the initial group, suggesting that the approval rate should decline, which it did. While the number of applicants expanded by 54% from 2000 to 2004, the approval rate declined only by 8.9% (From 93.8% to 85.5%) 7. Thus, it is not unreasonable to conclude that the growth in the number of companies applying for tax credits drove the increase in the number of R&D performers (i.e. successful SR&ED applicants), rather than an increase in the approval rate. How, then, can the increased rate of applications be explained? There are several possibilities. SR&ED Applications on the Rise There are several possible explanations for the growth in SR&ED applications that followed the tech bust (of 2001). One is that more CEOs 8 made a concerted effort to improve their companies productivity and competitiveness by launching R&D programs, investing more in research and subsequently claiming SR&ED tax credits. If that were the case, it would signal a sea change in corporate commitment to R&D, and it would have important policy implications 9. However, StatCan data on corporate R&D spending (Figure 5) do not support that hypothesis; spending actually declined after the 2001 peak, which implies that there was no substantial expansion of corporate R&D performance, by one measure at least It seems unlikely that post-2001 there was a large expansion in the number of new technologyoriented companies - companies that would be more likely to apply for SR&ED tax credits. It is true that the tech bust resulted in the layoff of many scientists and engineers at larger technology-oriented firms. Often, these individuals formed new technology-based SMEs. Often, the new firms went on to develop new products, conduct new research and apply for SR&ED tax credits. 7 Private discussions with individual CRA claims officers suggests that under pressure from industry associations, politicians and individual companies, that CRA may have let its standards for claims approval slip somewhat during the period, which would account for the (still) comparatively high approval rate of the incremental, post-2000 claims. 8 This implies over 9,000 CEOs, corresponding to the number of new SR&ED applicants after For example, it would imply that the SR&ED program has spurred incremental R&D performance. 6

7 14,000 12,000 Figure 5. Current Intramural Expenditures ( ) 10,668 11,370 11,772 10,000 8,000 6,000 4,000 2, ,108 5, ,702 6,253 7,296 7,496 7, , e 8,762 9,241 12,015 11,425 11, e 2000r 2001r 2002r 2003r 2004p 2005p Current Intramural Expend. (Constant 1997 $million) It is probable that this phenomenon had some marginal effect on SR&ED applications and approvals, but likely not in sufficient numbers - thousands of new applicants - to explain the dramatic increased witnessed. (After all, if the number of R&D performers did not grow during the tech boom in the years leading up to 2000, when investment was flowing into technology and spending rose dramatically, we do not consider it likely that the inverse was true after 2000, when funds were short and companies were closing or downsizing.) There are a number of other possible explanations for the apparent growth in performers that occurred after the tech bust: Changes in the regulations essentially making new classes of companies eligible for the program. Changes in the administration of the SR&ED program leading to more applicants and approvals 10. Other external factors influencing the number of applicants. 10 We have already discounted the possibility that changes in interpretation produced a higher approval rate - in fact, the opposite appears to be true. 7

8 Growth in R&D Performers - Other Factors at Work If the evidence does not support the conclusion that a fundamental change in corporate culture is at the root of the growing number of R&D performers - that is, more CEOs becoming committed to R&D - and there is no clear evidence that rising SR&ED approval rates are responsible, what can the explanation for the growing number of R&D performers possibly be? Our investigations lead us to examine a number of additional factors at work. One factor is that new software guidelines were introduced in the late 1990s and then revised shortly afterwards. According to Denys Cooper 11, these changes were in response to a previously high rejection rate for software filings. These likely produced a jump in the number of eligible software claims. Secondly, CRA expanded its role in Until that time CRA had acted largely as the auditor of company requests for tax credits or refunds. It adjudicated corporate tax filings to determine whether companies had met the eligibility requirements. In 1998 CRA adopted a second role, in parallel with its auditing role; it became an active promoter of the SR&ED program to industry 12. CRA began to aggressively promote industry awareness of the program; for example, it launched local workshops designed to make companies more aware of the program. It also began to advertise the program more widely, and to work with industry associations to inform their members about the program. CRA reported that 13 : Efforts... began in 1998 to restore confidence in the program and respond to industry concerns about consistency and predictability in program delivery... New policy guidelines... set out the CCRA s procedure for addressing claimant concerns and unresolved issues resulting from a review and for conducting a technical review of a claim respectively... Joint government-industry workshops help claimants apply for the credit and comply with its requirements. Sector-specific committees... developed issue papers to improve program consistency and fairness across economic sectors, and National Technology Sector Specialists respond(ed)to the needs of major Canadian industry sectors. The Account Executive service provid(ed) a central point of contact and guidance on the program. In summary, through wide-ranging partnerships forged with industry, we have created a rich source of information and guidance about the SR&ED program. We... streamlin(ed) paperwork and simplifi(ed) the claim process. It is easier for companies to benefit from the program, and feedback shows that industry is more confident about it. (Emphasis added.) 11 Denys G. T. Cooper is currently a Guest Worker - Technology and International at NRC- IRAP. We are grateful to him for the following points. 12 For example, CCRA s 2000 Performance Report states: We are delivering an appropriate mix of programs to protect the tax base, while still encouraging taxpayers to participate in key tax incentive programs such as the Scientific Research and Experimental Development Program. 13 Annex: Supplementary Performance. Information by Business Line Annual Report to Parliament. 8

9 In parallel with CRA s new promotion orientation, some provincial government agencies - notably ADRIQ in Quebec - began to promote awareness of SR&ED in their provinces. Together, these measures undoubtedly boosted the application rate. An additional factor, we believe, in the growing number of program applicants, is a byproduct of the emergence and growth in the SR&ED tax credit consulting industry. A Google search of the term SR&ED Consultants brings up a list of 3,250 hits; mostly specialist firms offering SR&ED tax credit application services to other companies. As explained by one firm - and typical of firms in the industry - it offer(s) a flexible fee structure based on a contingency option, a per diem option, or a combination of both. These fee choices ensure that a client pays for only those services that they require for completing a successful SR&ED ITC claim. In other words, firms in the SR&ED consulting industry approach companies that typically have not applied for SR&ED tax credits, offering their services on a contingency basis. The consulting firms offer to complete the SR&ED paperwork in return for one third of the returns, if successful. There is no risk to the R&D performing company. And on the earlier evidence, more than 85% of claims will be approved by CRA, meaning that the consultants risk of failure is also quite low. A change in SR&ED program guidelines in the early 2000s required companies to submit their claims within 18 months of completion, which probably helped the SR&ED consultants to sell their services to firms that might previously postpone their filings. In effect, we hypothesize that CEOs were not necessarily buying into the improved benefits of R&D and changing corporate R&D performance accordingly, but rather, were being encouraged to pursue tax credits, often for routine business expenditures that in past they did not claim. They were prompted to do this because they had nothing to lose by applying for a tax credit through a third party, and a high assurance of success. However, nothing in their core business had changed, only the fact they began to apply for tax credits. Even if these factors were responsible for the rise in the apparent number of R&D companies, does this in fact matter? After all, if the work for which the companies were successfully applying for a tax credit legitimately qualified as R&D according to the CRA s interpretation then we could as easily conclude that a vast amount of R&D activity was being consistently underreported in the pre-2004 period, and that the surge in new applications and approvals was simply setting the record straight. The fact that the CRA s approval rate fell implies that the agency was even more stringent in adjudicating claims post-2000 than in the previous period. But what if the SR&ED approval rate data are misleading? What if they mask an underlying trend to in fact loosen the approval guidelines? Does this contradict the evidence? We saw earlier (Figure 3) that the number of firms applying for SR&ED tax credits exploded by over 88% from 1997 to 2004 (10,728 applicants to 20,132 applicants) and that the number of firms being approved rose by 78% (from 9,648 to 17,222). This implies that over 80% of the new applicants were being granted tax credits. Is it realistic that 80% of firms that previously did 9

10 not apply for tax credits for research were in fact conducting research all along but either did not know that, or did not know they could apply for tax credits? We find this conclusion to be somewhat suspect. For one thing, the CRA utilizes a restrictive OECD definition for R&D that sets a high standard for the definition of eligible research: To qualify for the SR&ED program, work must advance the understanding of scientific relations or technologies, address scientific or technological uncertainty, and incorporate a systematic investigation by qualified personnel. Work that qualifies for SR&ED tax credits includes: experimental development to achieve technological advancement to create new materials, devices, products, or processes, or improve existing ones; applied research to advance scientific knowledge with a specific practical application in view; basic research to advance scientific knowledge without a specific practical application in view; and support work in engineering, design, operations research, mathematical analysis, computer programming, data collection, testing, or psychological research, but only if the work is commensurate with, and directly supports, the eligible experimental development, or applied or basic research. (Source: CRA. Emphasis added.) How likely is it that new SR&ED applicants were in fact performing experimental development, applied research or basic research all along? Not very, we would argue. We find it hard to fathom that in a space of eight years ( ) that 7,574 firms - 80% more than before - launched (or discovered they already had) in-house programs of scientific research and experimental development that would meet the CRA s high standard. Did a change to the CRA eligibility guidelines to allow for more software projects to qualify for tax credits influence the number of applicants and approvals? Undoubtedly, such changes (e.g. in the software eligibility guidelines) did increase the number of applicants and the number of successful claimants. Is it possible that during the period of interest, that CRA was in fact lowering the bar for approvals - allowing a higher proportion of claims than it did previously? Such a claim runs counter to expectations - and our data - because all things being equal, it is in the CRAs interest to limit the government s tax claim exposure and not raise it 14. Even after the influx of new claimants, the overall SR&ED approval rate claimants was still at a relatively high level % in 2004 versus 93.8% in Put differently, an 80% increase in the number of claimants yielded only an 8.8% drop in the approval rate. This strikes us as being an unexpectedly small decline in approval levels, given the magnitude of the increase in applicants. limit. 14 The SR&ED program is essentially un-capped. The government is obligated to fund all eligible claims, without 10

11 Conclusion The sharp rise in the apparent number of industrial R&D performers from 2000 onward is probably a consequence of multiple factors: Changes in the SR&ED program guidelines to allow a broader range of eligible claims (e.g. software). Reduced oversight and less stringent standards of claims assessment on the part of CRA. Increased marketing of the program by CRA and others. The growth of the SR&ED consulting industry. In our opinion it is unlikely that a vast number of companies substantially altered their basic approach to investing in or performing research, between 1997 and Rather, in increasing numbers, many companies began to apply for R&D tax credits for activities - often routine activities - they did not previously apply for. The growing number of consultants who will prepare a firm s tax credit claim on a contingency basis means that putative R&D performers have no risk or cost for submitting a tax credit and are thus applying in greater number. If this analysis is correct, it leads to the conclusion that the SR&ED program itself is tending towards becoming a general business subsidy program, rather than a focussed R&D support program. We readily accept that a number of the assertions in this paper need to be tested more thoroughly, and encourage other researchers and analysts to contribute to the discussion. *** 11

12 Notes 1. SR&ED applicant data were provided by CRA. 2. SR&ED approval data were provided by Statistics Canada. 3. Other data come from The Demographics of Industrial Research and Development in Canada, The Impact Group. Toronto. January We thank Dr. Denys Cooper for contributing to this analysis. 12