Make Prediction
Test Prediction
Iterate
These three key steps of the scientific process are often omitted from an R&D credit analysis.
SPRX R&D Credit Tip – use predictions effectively to enhance the accuracy and efficiency of your R&D credit process.
Consider the Source
Too many R&D credit CPAs believe that the only way to determine the percentage of time a person (“SME”) spends on qualified research activities is to sit down with the person (or their supervisor) and ask, “what percentage of time do you spend on qualified research”? This question has probably caused more debate between taxpayers and the IRS over the years than any other question.
The problem is that the CPA is asking a tax question to a non-tax professional.
Does the SME know the correct definition of “qualified research”? After all, it took the U.S. Treasury over 20 years to write the definition and taxpayers and the IRS have been disagreeing on the definition ever since. To ask a non-tax professional to define a term that tax professionals can’t define seems questionable. At best the SME will give a rough estimate.
Estimating Qualified Time
R&D credit CPAs have built their careers on the 1930 Second Circuit court case Cohan v. Commissioner. In this case the court allowed Mr. Cohan’s estimates of travel expenses when he could not produce the actual receipts. The court reasoned that Mr. Cohan obviously spent something on travel and so an estimate was allowed. Today the ability to estimate is known as the Cohan Rule. The IRS has explained that applying the Cohan Rule requires that the taxpayer present credible evidence that provides a rational basis for the estimates.
It is well established that taxpayers can use estimates when certain conditions are met. We’ll assume that that includes an SME estimating the percentage of time spent on qualified research activities when the SME has not recorded every hour worked during the year.
What happens when the SME estimates are inaccurate or too low?
Prediction Provides Guidance
Accurate predictions can help guide the SME in making estimates.
Once the SME has made an initial estimate it is difficult to alter the estimate. Sure, you can just change the number, but when the IRS asks the SME for clarification or justification of the estimate, there is a risk that the SME will say “I was told to say a different number”. The IRS has been known to bait SMEs during interviews specifically so that they will make such a statement.
Even the most cooperative SMEs are at a significant disadvantage in estimating qualified time because they don’t have a reference point. They don’t often have the benefit of knowing who others have defined qualified research or have estimated time. Predictions are useful guides.
A.I. Prediction Models
Artificial intelligence is a valuable predictive technology. Prediction models are trained on thousands of individual use cases. A human would have to spend a lifetime in an industry to gain similar experience. A.I. prediction models are used in many industries because of unprecedented experience and processing capacity. This certainly applies to predicting the percentage of qualified research time for a group of individuals.
When employee data is run through a prediction model and model estimates the qualified percentage of time based on the equivalent experience of several industry experts. The predicted value is just that, a prediction.
The SME can then be asked to validate the prediction. The SME is able to act with higher precision, time and confidence when they have the reference point of a prediction model.
A Better Experience
SMEs report higher satisfaction scores when they have predictions to aid their R&D credit estimates. They report spending less time on the task and feel greater confidence in their results. The annual task of estimating qualified research time dropped from several hours to less than an hour when they had the benefit of validating an estimate.
Some taxpayers are finding qualified costs they’ve missed in the past when they run all employee data through the prediction model. When less time is spent on the core departments, time can be allocated to validating other departments. The predictive model has helped areas that need additional investigation.
At SPRX.tax we build the tools that save you time…and money