twelve and one-half percent credit at the margin, as can be seen in Table 2. Thus only 63.2 percent of total qualified R and D expenditures of those calculating a tentative credit in the R and D tax file were incurred by firms in a position to claim the full credit (defined as 99 percent or more of a full 25 percent credit). And this does not take into account the some 6 percent of expenditures, according to our Compustat data, undertaken by firms who would not have filed for the tentative credit because their 1981 expenditures were below their 1980 base expenditures. Comparison of the OTA data for 1981 with data from the Compustat tapes, from McGraw-Hill and from the National Science Foundation tabulations reveals some striking and informative differences. Total company funds for R and D grew by 16.2 percent from 1980 to 1981 according to NSF and 14.1 percent in the Compustat data. Restricting the Compustat firms to those with positive R and D growth, to make them comparable to the OTA sample, we still get a growth over base of only 21 percent, as against the 40.3 percent growth in the OTA sample. There is clearly a strong implication that many business taxpayers were indulging in creative accounting, classifying as qualified research and experimentation expenditures in 1981 activities which were not included in calculating the 1980 base. Tax credits were claimed on alleged increases in R and D which were some double other presumably reliable measures of the increases of R and D which actually occurred. The Compustat data for 1981 indicate a much larger proportion of R and D undertaken by firms with R and D greater than base for which there was some federal tax liability against which the credit could be claimed. This came to 80.7 percent of total expenditures in 1981. Interestingly that same proportion, as indicated in Table 3 of our National Tax Journal article was down to 52.7 percent in 1982, as a consequence of both declining growth in R and D and lower tax liabilities. This last points up the pro-cyclical nature of the R and D tax credit. In recessions, R and D expenditures, like all other expenditures, tend to slacken. Since the credit is tied to the rate of growth of R and D expenditures, it is particularly sensitive to such a slackening. Thus a decline merely in the rate of growth over base from, say, 12 percent to 6 percent would cut the credit in half. Further, with more firms suffering losses in a recession, tax liabilities against which the credit may be claimed are less numerous. On both counts, therefore, the R and D tax credit tends to be lower in recession when tax reductions would appear particularly desirable to stimulate the economy, and higher in booms when a tighter tax policy might appear useful to prevent inflationary excesses. Indeed, since the credit relates to increases in nominal R and D expenditures, inflation serves to increase the credit and reduce taxes, again the opposite of what would be indicated by appropriate counter-cyclical policy. A year ago, working with data from Compustat, McGraw-Hill and NSF, we presented estimates of the tentative credit for 1981 ranging from $442 million to $684 million, all very considerably higher than earlier Treasury and Joint Congressional Committee on Taxation estimates. But even our figures have proved substantially low in light of the very great increase in the "eligibility ratio" or ratio of qualified R and D expenses to total R and D, which apparently took place in 1981. The 1981 credits were of course low because they applied only to R and D expenditures after June 30. Credits for later years could easily be projected at considerably higher levels, generally one billion dollars and more for each of the years from 1982 through 1985. We are now able to present a set of updated estimates of Treasury revenue losses for 1981, 1982 and 1983, projected losses for 1984 and 1985 and, if the credit is extended in its current form, projected losses for 1986 and 1987. The estimates begin with analysis of a now presumably reasonably complete OTA file of returns of form 6765, Credit for Increasing Research Activities, for 1981. Comparison of "qualified research expenses" reported to the Treasury and reported R and D in the Compustat data permit us to calculate "eligibility ratios" of R and D used in calculating bases and totals for tax credit purposes with R and D reported on financial statements. We can then tie the Compustat data to McGraw-Hill's survey results of R and D in 1982 and 1983 and anticipated changes in R and D from 1983 to 1984 and from 1984 to 1987, to project tax credits in future years. As shown in Table 1, the OTA file now indicates a tentative credit of $872 million for 1981, of which 72 percent or $630 million could be claimed against current tax liabilities. The great bulk of the credit, as seen in Table 3, goes to very large firms, over 75 percent or $660 million of the tentative credit to companies with more than one billion dollars in assets. The credit claims to the Treasury were apparently held down significantly in 1981 by the amount of R and D, over nine percent, reported by firms calculating increases of more than 100 percent over their 1980 base period. The actual R and D reported in financial statements and listed in the Compustat tape does not show such a concentration of extremely high growth. Thus, when we apply common eligibility ratios for all firms, calculated so that estimated R and D for the second half of 1981 as well as base figures for 1980, for firms with R and D over base, are equal to the OTA totals, the credits calculated from the Compustat data are somewhat higher. Accepting this minor discrepancy, we may note in Table 4 that the Compustat-calculated tentative credit of $935 million in 1981 would rise to $1,474 million in 1982 and $1,775 million in 1983. Lacks of sufficient tax liabilities, against which the credit may be offset, reduces the credit currently available to $784 million in 1981, $934 million in 1982 and $1,416 million in 1983. Allowance for the potential threeyear carryback, however, brings the credit available back close to the tentative credit, or to $888 billion for 1981, $1,395 million for 1982 and $1,688 million for 1983. Aa may be noted in Tables 5 and 6, the credits again are highly concentrated among large firms. In Tables 7 and 8 we may see the distribution of both qualified R and D and the credit by the growth categories distinguished in the legislation. The Compustat data indicate only small proportions of R and D undertaken by firms reporting increases of more than 100 percent over base. We may infer that the large proportion in this category in the 1981 OTA file is a one-time phenonomen of firms suddenly discovering vast amounts of qualified R and D expenses in 1981 with little or nothing so listed for the base period of 1980. The 2.6 percent of 1981 qualified R and D undertaken by firms reporting research expenses less than those in the base period, and hence not eligible for any credit, grew to 7.4 percent in 1982 and 8.1 percent in 1983 according to Compustat data with eligibility ratios calculated to conform to tax returns. Table 9 indicates, for those firms for which current and back tax data were available, the portions of R and D undertaken by "takeup class," that is, class determined by the extent to which firms had current tax liabilities to which research and experimentation credits could be applied. In 1982, as may be seen in Table 10, some 41 percent of R and D was in the zero takeup class, that is, was undertaken by firms who could claim no credit against current taxes because they were not paying current taxes. That figure was sharply lower but still a substantial 22.9 percent in 1983. However, the carryback provision had a major effect, as may be seen in Table 9, in that while no credit could be claimed against current taxes in this zero takeup category, 85 percent of the tentative credit coud be claimed on the basis of carryback in 1982, and 76 percent in 1983. For all categories, the percentage of tentative credit which could be claimed rose from 63.4 percent to 94.7 percent on the basis of carryback for 1982, and from 79.8 percent to 95.1 percent for 1983. Tables 11, 12, and 13 offer breakdowns of some of this information by industry, for the years 1981, 1982, 1983, respectively. Tables 14 and 15 offer projections, from the McGraw-Hill surveys, of qualified R and D and the tentative credit for the years 1984 through 1987. These, of course, can be no better than the information from McGraw-Hill respondents on which they are based. They suggest though, as might be expected, that the incremental tax credit for research and experimentation will continue to be costly to the Treasury in 1984 and 1985, and on into the future if it is extended or made permanent. We project, as may be seen in Table 14, tentative credits of $1,702,000 in 1984, $1,544,000 in 1985, $1,539,000 in 1986 and $1,618,000 in 1987. While substantial portions of these tentative credits will not be available for each year's current tax return because of lack of current tax liabilities, the carryback provision, as we have noted, is likely to keep the full cost to the Treasury in each year not too far below the tentative credits we are projecting. We are continuing our efforts to ascertain what effect the incremental credit for research and experimentation has in stimulating R and D spending. As noted above, the nature of the credit is such as to inspire doubt as to its effectiveness. Our examination of the data thus far has done nothing to dispel that doubt, although it is fair to say that definitive analysis is difficult, in part because the credit is calculated on the basis of growth in R and D. Hence R and D surely grows more rapidly when and where the credit is greater but that gives us no clue as to whether the credit is stimulating the R and D expenditures on which it is based. Our analysis is also slowed by lack of current tax data. Up to this time we have only been able to analyze the Treasury tax data bearing on the credit in 1981 when, of course, it was in effect for just half a year. It might be argued that an impact would be found in later years when the credit was available for all 12 months of expenditures and when firms had a chance to alter their plans to fit the new legislation. Our main approach in analysis thus far has been to relate R and D spending to whether or not the credit was actually available. By this line of reasoning, for exam ple, we should expect firms with R and D well below base to have no incentive to increase R and D because they would not be likely to increase it enough to enjoy a tax credit anyway. In earlier analysis, reported last year, we could find no evidence of changes in behavior of firms with R and D below base and firms with R and D above base as between years before the credit was in effect and the years from 1981 on. Similarly, working with the current 1981 OTA file, we have endeavored to note the effects of takeup class and growth class on R and D. We might expect that, if the credit were effective, R and D would be greater, other things given, where firms had tax liabilities to which the credit might be applied. As may be seen on Table 1, though, and as we have found in more sophisticated analysis which we plan to discuss in subsequent report, we can find no evidence that the availability of the credit has increased R and D spending or that its lack of availability has discouraged it. It must be confessed, however, that this analysis of the OTA data may be defective because the OTA file has no information on carryback. Hence our measure of credit availability may be seriously flawed. We have endeavored to correct for this in analyzing Compustat data by year to see whether the ratio to tentative credit of credit claimed including carryback has some positive impact on R and D spending or, more subtly, the difference in effect of that variable calculated for years when the credit was not in effect, and hence could have no significance, and calculated for the years 1981 through 1983. Again, our statistical results offered no suggestion that the credit was having any positive effect. One thing that is clear is that the credit for research and experimentation is costing the Treasury substantial amounts of revenue, well over a billion dollars per year. The annual loss has been growing rapidly and will continue to grow, if more slowly, if the tax credit is extended. Yet analytical considerations, now supported by some data, raise serious questions as to whether the credit is doing much, if anything, to realize its ostensible goal of increasing investment in research and development. As we indicated at the outset, the presumption that more R and D spending may be desirable could not extend automatically to the conclusion that any particular tool is effective or desirable. One may even question whether support for research and experimentation should depend upon subsidies to business rather than on direct government action or support of nonprofit universities and research institutes. Much current research, in agriculture, in defense and in basic sciences is, after all, not done by private business. Perhaps support for non-business R and D would be more fruitful, with business left to concentrate on R and D spending which seems profitable without special tax incentives. If we do decide to go further on the route of encouraging business R and D spending, certain revisions of the current credit might increase its effectiveness. First, we might do well to eliminate the 100 percent limitation. While it does not apparently relate to a large proportion of R and D, its negative incentive effects can be considerable where it does come into play. Second, the credit might best be made refundable or converted into a direct subsidy. Aside from being aboveboard and allowing the Congress and the public to see clearly what the government encouragement of R and D is costing, a direct subsidy would free government support from the sometimes capricous effects of a tax system which is already so saddled with "incentives" which may less charitably be dubbed loopholes. Clearly, the current provision discriminates against firms without tax liabilities, either because, perhaps, they are new and still growing rapidly or because of low profits or substantial indulgence in other tax-reducing activities and accounting. Third, while retaining the incremental nature of the credit-which may in principle allow it to have a greater "bang for the buck"-we should eliminate the company-specific movements of the base. It is this feature which results in firms generally losing dollar for dollar in future credits the amounts that they gain in current credits from increasing their R and D expenditures, and actually encourages some firms to reduce their R and D. This provision could be changed by having a company-specific base but leaving it fixed at, say, the average of 1982, 1983 and 1984 qualified R and D expenditures (with possibly some exceptions or adjustments for new firms). This would imply rapidly growing amounts of R and D credits over the years but would mean that the effective rate of credit would be equal to the statutory rate, and would eliminate any incentive from the credit to reduce R and D. Somewhat more complicated to administer, but less costly in terms of tax revenues, would be a provision to start with a company-specific base but to adjust that base in accordance with national or industry movements in R and D. Thus, if a firm were in an industry where R and D in 1983 grew by 10 percent, its base in calculating its tax credit for 1984 would be raised by 10 percent. A firm's increased expenditures would contribute to raising the base and reducing future credits for all firms in the industry, but would have a trivial effect in raising its own base and reducing its own future credits. Having the base depend on industry behavior rather than company-specific behavior would achieve maximum incentive impact with minimal Treasury tax loss. Improvements along these lines would seem imperative if an R and D credit of any form is to be extended. Whether it should be extended in any form is another matter. Given the concern about mammoth structural deficits in the years ahead, this Committee and the Congress might well wish to avoid action that would add perhaps another one and one-half billion dollars annually to that future deficit, with little or no clear payoff. TABLE 1.-QUALIFIED R&D (QRD), BASE, TENTATIVE CREDIT AND CREDIT CLAIMED, BY TAKEUP CLASS, OTA FILE, 1981 TABLE 2.-QUALIFIED R&D CREDIT CLAIMED, BY TAKEUP CLASS AND GROWTH CLASS, OTA FILE, |