of an extra dollar's worth of R and D by 25 cents.

Because an extra dollar

of R and D this year increases the base amount during the next three years,

it reduces the credits available then. Assuming that the credit remains in effect for the following three years, the reduction (in cents) really equals only about 6 cents, if the interest rate is about 15 percent. And for firms that do not intend to raise this year's R and D expenditures above the base level or that do not have an income tax liability against which to apply the credit, the price reduction is far less than 6 cents. According to Eisner, Albert, and Sullivan (1983), "the effective rate of credit for firms with tax liabilities and increasing R and D is in the ,,10 order of only a few percent. Although relatively little is known about the price elasticity of demand for R and D, the few available estimates' suggest that it is rather low, perhaps about 0.3. If this is the case, and if the tax-credit-induced price reduction is 6 percent or less, the tax credit would be expected to increase R and D by 1.8 percent or less, which compares with the 1983 confidence interval from the above survey of 0.6-1.8 percent. Thus, based on these crude estimates, the survey results seem eminently reasonable.

To What Extent Have Activities Been Redefined as R and D?

11

Long before the tax credit was introduced in 1981, economists and others pointed out that any such credit would encounter difficulties with regard to the definition of "research and development." Obviously, the credit provides firms with an incentive to redefine activities as R and D. The Treasury Department has asked Congress to tighten the definition of R and D. In May 1983, Assistant Secretary of the Treasury John Chapoton testified that: "The vagueness of this definition allows the assertion

that the cost of developing virtually every product marketed in the economy should be treated under the .. credit as qualified research expense. It is difficult to specify what pre-production expenses are research and 1,12 experimental expenditures and what pre-production expenditures are not. To what extent has reported R and D expenditure increased merely because of the tax-credit-induced redefinition of activities as R and D? To obtain information on this score, we posed this question to a randomly selected subset of the firms in our sample. The results indicate that, taken as a whole, the reported R and D expenditures of these firms increased by about 4 percent per year during 1982-83 for this reason alone. Obviously, this estimate is very rough. Nonetheless, it is interesting to note that American experience in this regard, as measured by this estimate, is strikingly similar to the reported behavior of Canadian and Swedish firms described below. If our experience parallels theirs, this redefinition of activities as R and D will tend to peter out in the mid-1980s after reported R and D expenditures have risen in total by about 13 or 14 percent for this reason alone.

The

Clearly, this shift in definition makes it difficult to use the reported figures on R and D spending to estimate the effect of the credit. reported figures are based on a rubber yardstick; there is no simple way to sort out increases in R and D activity from the effects of more inclusive definitions of R and D. One important advantage of the survey technique described above was that the firms were asked to estimate the effect of the tax credit on their R and D expenditures, holding constant the definition of R and D. Thus, the resulting estimates are based on an unchanging yardstick.

The R and D Tax Allowance in Sweden

From 1973 to 1983, Sweden had an R and D tax allowance. What were its effects? In Sweden, current R and D expenditures are deductible in the year in which they are incurred, and capital assets used for R and D must be depreciated over their normal economic life. In addition, in the early 1980s, there was an R and D tax allowance equaling 5 percent of a firm's R and D expenditure plus 30 percent of the increase over the previous year. To estimate the effects of this allowance, we carried out a survey of Swedish firms. The universe of firms was divided into two strata, one stratum being composed of the top 14 R and D spenders, the other being composed of all other firms. A random sample was chosen from each stratum; the total sample turned out to include more than 80 percent of all company-financed R and D in Sweden.

According to the firms in this sample, company-financed R and D expenditures would have decreased by about 1 percent in 1981 and 1983 if the tax allowance had not been in existence. As in the United States, it appears that the tax-incentive-induced increase in company-financed R and D was less than the revenue loss to the government.

There are a number of reasons

why the effects were small. Many firms did not have any income tax liability against which to apply the allowance. Even if a firm had such liabilities, the reduction (in cents) of the after-tax price of an extra dollar's worth of R and D to a firm that increased its R and D expenditure (and that intended to continue to increase its R and D expenditure for each year in the foreseeable future) was only about 3 or 4 cents, if the interest rate was about 15 percent.

Because the tax allowance was in existence for about a decade, it was possible to perform somewhat more extensive statistical analyses than in

the United States, where there have been only a few years of experience with the tax credit. Based on our analysis of the official Swedish data, there was no evidence that the tax allowance resulted in a significant increase in the proportion of sales devoted to R and D. Thus, the statistical results were consistent with the survey findings. Although there are many dissimilarities between the Swedish and American economies, and any comparison between them must be viewed with caution, it is interesting that the estimated effects of the 13 R and D tax incentives in the two countries are very similar.

At the beginning of 1984, the Swedish government eliminated the R and D tax allowance. According to officials of the Swedish Royal Academy of Engineering Sciences, this was due in part to growing questions concerning the effectiveness of the allowance.

Direct R and D Tax Incentives in Canada

Canada was a pioneer in the adoption of direct R and D tax incentives. Both current and capital expenditures on R and D in Canada are fully deductible in the year incurred. In addition, during the early 1980s, there was both an R and D investment tax credit and a special research allowance. The investment tax credit (which was taxable) was 10-25 percent of current and qualified capital expenditures on R and D, the percentage varying with the size of the firm and the location of its R and D. The special research allovarice permitted corporations to deduct from their taxable income an amount equal to 50 percent of the increase in operating and capital expenditures for R and D.

To estimate the effects of the Canadian R and D investment tax credit and the special research allowance on company-financed R and D, a survey like that in the United States and Sweden was carried out. We selected a stratified random sample of 55 firms, which account for almost 30 percent of the company-financed R and D expenditures in Canada. Based on these

firms' estimates, it appears that in 1982 the investment tax credit increased company-financed R and D expenditures (when the definition of R and D is held constant) by about 2 percent, and that the special research allowance increased them by about 1 percent. Thus, the effect of these tax incentives seems to be about the same as the U.S. or Swedish tax incentives. As in the United States and Sweden, the increases in company-financed R and D expenditures seem to have been considerably less than the cost to the government in reduced tax revenue. Because of the availability of longer time series in Canada, it was possible to do more with econometric techniques there than in the United States or Sweden. While the econometric results are subject to a variety of limitations, they seem quite consistent with the survey results. 14

In 1984, Canada changed its system of direct R and D tax incentives, the stated purpose being "to improve their effectiveness, to make them of more immediate value to firms performing R and D and to facilitate financing ,,15 of R and D. The special research allowance was eliminated, and the investment tax credit was increased by 10 percentage points for all R and D performers. In addition, to increase the extent to which these credits are actually utilized, firms now can renounce their unused credits and allow outside investors to claim these incentives. While this new system has 16 attracted considerable public attention (and controversy), it is so new that very little information is available concerning its effects.

Conclusions

While there is agreement in many quarters that American public policy should promote a greater investment in civilian technology, the available evidence does not indicate that the R and D tax credit, in its present form,