Four reasonable compensation mistakes S-corp clients make

Every S-corp owner who works in the business has to settle on a reasonable compensation figure, and many of them aren’t sure how. The standard, from Reg. §1.162-7(b)(3), is the amount that would ordinarily be paid for like services by like enterprises under like circumstances. Answering that takes wage data and a way to apply it, and few owners have either on hand.

A salary set too low invites reclassification of distributions as wages, with back employment taxes, interest and a possible 20% accuracy-related penalty under §6662. A salary set too high means paying employment tax every year on income that could have been distributed. The preparer is exposed too, since an unreasonable position on a signed return can trigger a penalty under §6694.

Mistake 1: Using a fixed ratio

The one owners tend to have heard is 60/40, meaning 60% of profit as salary and 40% as distributions. It gets repeated with enough confidence that many owners assume it’s an IRS guideline.

However, no such ratio exists in the Code, the regulations, or the case law. IRS Fact Sheet 2008-25 lists nine factors drawn from court decisions, among them: the owner’s training and experience, their duties, the time they devote to the business, and what comparable businesses pay for similar services. A percentage of profit isn’t on the list.

Mistake 2: Setting salary by what the owner needs or what the business can afford

Some owners work backward from cash, so the salary is whatever covers their personal expenses, or whatever is left after the business pays its bills. When it comes up with a preparer, it’s often framed as a question about affordability.

The standard doesn’t take affordability into account. Reasonable compensation is measured by the value of the services performed and it generally doesn’t change with profit, distributions or the company’s cash position.

Mistake 3: Taking a chatbot’s answer at face value

Some owners now ask an AI tool for the number. Given a role, hours and a location, it returns a figure with wage data and citations attached.

In our testing, the citations often don’t check out. Pull up the BLS series the tool names and the figures frequently aren’t there, and some of the cited sources don’t exist.

Even when the citations check out, the method behind the number is harder to see. A justifiable figure depends on several choices, and a chatbot tends to skip them:

• Which year of wage data was used, and whether older data was adjusted for wage growth since then
• Which geography was used, since the same occupation pays very differently across metro areas
• Which percentile of the wage distribution was used, and why it fits the owner’s experience
• Which occupation was matched to each of the owner’s duties, and how the hours were weighted

A chatbot will often apply a national median from whatever year it has on hand to a single job title, and present that as if it were a careful analysis. Few owners know enough to check the sources or the method, so the number goes on the return as is.

Mistake 4: Picking a number with no method and no documentation

The simplest approach is to pick a number that feels right, or keep whatever was set when the election was made, and write nothing down.

This one is the hardest to defend. A 60/40 split is wrong, but the owner can explain how they got there. Reasonable compensation is evaluated after the fact, and the examiner wants to know what the owner did that year, how the salary was arrived at, and whether the decision was reasonable at the time. Without a record, all the owner can offer is what they remember after the notice arrives, and that looks like reasoning written after the fact.

What a number that holds up looks like

So what is required to produce a defensible number for reasonable compensation?

• A specific description of the owner’s duties. An owner-operator often sells, manages staff, keeps the books and does the technical work, and those jobs pay very differently. The analysis should price each of those jobs separately instead of pricing the title.
• Wage data from a recognized source. The Bureau of Labor Statistics Occupational Employment and Wage Statistics program is the most widely used public source, matched to Standard Occupational Classification codes and adjusted for the owner’s metro area. Every figure should trace to a source an examiner can look up.
• A transparent method. The cost approach, sometimes called the “many hats” method, prices each task at the market wage for that work, weights it by hours, and adds them up. The year, geography and percentile choices are stated, and the arithmetic can be checked line by line.
• A dated record. The analysis and a board resolution adopting the salary, written in the year the salary was set and revisited when the owner’s duties, the business, or the wage data change materially.

Why owners skip a proper study, and where the firm comes in

Part of the reason owners don’t end up with a defensible number is that a proper study has been expensive. A custom reasonable compensation study from a tax professional is usually a one-off engagement, and the fee can run into the thousands. For an owner paying themselves under $100,000, that eats up most of the payroll tax savings that made the election worthwhile, so the owner skips it and looks for a free answer.

How WageProof approaches this

We built WageProof to help tax preparers support their S-corp clients with this analysis. It produces a reasonable compensation report built the way this post describes: a task-by-task cost approach using BLS wage data for the client’s metro area, with the year, geography, and percentile choices stated and the methodology published in full. The report is dated and goes in the client’s file as the year’s documentation.

A single report is $199. Firm plans bring the cost down to as low as $25 per report and put the firm’s own branding on the report, so it can be offered to clients as a service of the firm. There’s a full sample report on the site.