The Internal Revenue Service issued final rules in January on how to define “research” that qualifies for a 20% federal tax credit.

The definition is important to power, mining and telecoms companies experimenting with new technologies. The more tax benefits a company can build into its project, the less the project will cost at the end of the day.

The tax credit will be difficult to claim for most projects.

In order to claim a credit, a company must show it spent money on experiments that have the aim of improving technology. The company will have to jump through four hoops to do this.

First, the aim must be to discover new information. The IRS gave the example of a manufacturing company that makes widgets, but wants to use a new material. The company lacks experience with the material, but how to use the material is within the common knowledge of other skilled professionals in the industry. This is not “research.” However, where a company wants to build a bridge that can carry a higher volume of traffic than other bridges without deterioration, its work on the technology to build the bridge does qualify. The IRS said it does not matter if someone else has already built such a bridge if the technology is a closely-guarded secret.

Second, research is a process of experimentation. The company should have more than one hypothesis for how to achieve a result and be uncertain which is better. It should run tests to determine which hypothesis is better.

Third, the activity must precede commercial operation. Activity after a project is in commercial operation is not research. The IRS said tooling up for production, trial production runs and trouble shooting are not research. Thus, a power company could not claim the cost of a turbine as research on grounds that the turbine was the first of its kind off the production line. However, the turbine manufacturer might claim that it was still engaged in research if it ran a test model before the model was in production.

Fourth, it is not research simply to adapt an existing product or process to a company’s needs. An example is customizing software so that a utility can use computers to dispatch electricity.

The cost of computer software developed for a company’s own internal use never qualifies for credits, except to the extent the software will be used in research. The IRS said it would not allow tax credits to be claimed for fixing year 2000 problems with computer software.

If a company qualifies for an R&D credit, then it can compute the credit in one of two ways. Under one approach, the credit is 20% of the amount by which the company increased its research spending above a base. For example, if research spending in 1999 is $6 million, but the company’s “base” spending on research was $4 million, then the credit is computed against the $2 million increase. The base is gross receipts for the year times the fraction of the company’s gross receipts that it spent on research during a five-year period from 1984 through 1988. Companies that had no research during this period are arbitrarily assigned a base of 3% of annual gross receipts. Research spending must exceed this amount before there is any credit.

Calculations are done by treating all business entities that are more than 50% owned as a single taxpayer.

The government will not let a company treat more than half its research spending in a year as an increase in its research spending. For example, if research spending mushroomed one year, the government would limit the credit for that year to 20% of half the research spending that year.

The other way to compute credits is under a sliding formula. A company would have to spend more than 1% of its gross receipts in a year to get a credit. The credit would be 1.65% of research spending above 1% of gross receipts, 2.2% of such spending above 1.5% of gross receipts, and 2.75% of research spending above 2% of gross receipts.

The R&D tax credit expires on June 30, 2004, but there is growing support in Congress to make it permanent.