S corp reasonable salary: setting reasonable compensation for S corp owners and officers

Most of what circulates about this subject is rule of thumb presented as law. The table below separates the two. Every row is either published IRS guidance or a holding from a decided case, and the source is named so you can check it.

The distinction sounds academic until you are the one being examined, and then it changes everything about how you prepare. If the rule were a floor, you would only need to clear it. Since the rule is characterization, what matters is whether the money you called a distribution looks like payment for work you did. That is why the strongest position is not the largest salary, it is the best-documented one, and why owners who took a defensible figure from market data usually fare better than owners who took a bigger number from nowhere.

Non-cash benefits sit inside the same analysis, and they follow a different rule for owners than for staff. A 2% shareholder is not treated as an employee for most fringe benefit purposes at all, under Rev. Rul. 91-26, and a 5% owner is a key employee at any salary, which strips the $50,000 group-term life exclusion the rest of the workforce keeps. Both outcomes show up as imputed income on the same W-2 the reasonable compensation figure lands on, so price the benefits before you decide the salary rather than after.

Sources: IRS, S corporation compensation and medical insurance issues; David E. Watson, P.C. v. United States, 668 F.3d 1008 (8th Cir. 2012); Joseph Radtke, S.C. v. United States, 712 F. Supp. 143 (E.D. Wis. 1989), aff'd 895 F.2d 1196 (7th Cir. 1990).

The IRS publishes the factors courts have considered in determining reasonable compensation. The left column is the IRS wording. The right column is what each factor turns into when somebody is looking at your file, because several of them are quietly asking about documents you may not have created yet.

Notice how much of this list is internal evidence rather than market research. Pay to non-shareholder employees, bonus timing and the existence of a compensation agreement are all things you control and can get right in an afternoon. The one factor that requires outside data is what comparable businesses pay for similar services, and that is the one the government leads with.

David E. Watson, P.C. v. United States is cited in nearly every article on this subject, almost always for the headline: a CPA paid himself $24,000 while taking around $200,000 in distributions, and lost. The useful part is never quoted. The government expert, an IRS general engineer named Igor Ostrovsky, showed his work, the district court adopted his figure of $91,044, and the Eighth Circuit affirmed. Here is the method, in the order he applied it.

Two details are worth dwelling on. First, the expert priced the non-owner job and then adjusted upward, rather than starting from an owner figure. The survey said an owner in a firm that size would receive about $176,000, but that number blended compensation with return on investment, so it was the wrong input. Reasonable compensation is the labor component only. Second, he revised his own opinion downward, from an initial $184,876 to the final $91,044, after learning more facts. The court treated the revision as a sign of a properly updated report rather than a weakness.

The transferable lesson is that a defensible number has a paper trail with a source, a date and an arithmetic step you can point at. That is the same standard a defensible salary band has to meet for any other role in the business.

When the court explained why the salary was understated, it set out its reasoning in six numbered findings. Read as a checklist in reverse, they describe exactly what a defensible file contains.

There is one more finding that never makes the summaries. The taxpayer testified that the firm partners had decided to distribute $2,000 a month to each partner regardless of the seasonality of the business. The court noted that there were no documents reflecting these salary discussions, and no other partner testified. The salary decision may well have been made exactly as described. Without a record of it, that made no difference.

These three are widely described as approaches the IRS recognizes. That is a little generous. The IRS publishes the nine-factor list, not a menu of methods. The methods are practitioner and valuation conventions for producing evidence about those factors, which is a useful thing to be clear about before you rely on one.

The independent investor point deserves emphasis because it is repeated so often without its context. In Exacto Spring a C corporation wanted to deduct a large salary and the IRS said the salary was too high; the Seventh Circuit asked whether an outside investor would still be happy with the return, found a return over 20 percent, and let the deduction stand. An S corporation owner arguing for a low salary is in the mirror position. A handsome investor return alongside a token salary is precisely the pattern that says the salary was too small.

If a percentage is useful at all, it is as a sanity check after the fact. Derive the figure from the market, then look at where it lands as a share of what the business produced. If the answer is wildly out of line with comparable firms, that is a prompt to re-examine the inputs, not a reason to override them.

The IRS does not publish a reasonable compensation dataset, and there is no official calculator. In Watson the source was an industry association survey specific to accounting practices, with regional adjustments. The equivalent source available to everyone is the federal Occupational Employment and Wage Statistics program, which reports mean and percentile wages for roughly 830 occupations across the nation, every state, and around 530 metropolitan and nonmetropolitan areas. The tax-exempt world runs the same exercise under a different statute, and its regulations are far more explicit about what the data has to look like: the examples in the nonprofit reasonable compensation rules reject a wide national range for a job title and accept the identical survey once it is segmented by organization size and geography, which is a useful standard to hold your own file to.

The full specification, including what the survey covers, what it excludes and the recent change that started publishing specific estimates for high percentiles that used to return only a footnote, is set out on the BLS salary data page. If you are weighing paid survey subscriptions instead, salary survey providers compares what the commercial vendors actually sell.

Redo this annually rather than once. The figure that was right when the business was two people stops being right when it is fifteen, and a salary that never moves while distributions climb recreates the exact pattern in the case law. The same discipline applies to everyone else on the payroll, which is what a salary structure is for.

The same band does double duty. It supports the owner compensation position, and it prices the next hire in that role from the same evidence base. If the role is one you will eventually fill with an employee, exempt vs non exempt classification decides whether that person is owed overtime, and job posting salary ranges covers what has to appear in the advertisement in a growing number of states.

Sources: IRS, S corporation compensation and medical insurance issues; IRS, S corporation employees, shareholders and corporate officers; IRS, Paying yourself; IRC 162(a)(1); David E. Watson, P.C. v. United States, 668 F.3d 1008 (8th Cir. 2012); Joseph Radtke, S.C. v. United States, 895 F.2d 1196 (7th Cir. 1990); Veterinary Surgical Consultants, P.C. v. Commissioner, 117 T.C. 141 (2001); Exacto Spring Corp. v. Commissioner, 196 F.3d 833 (7th Cir. 1999). This page is general information about published IRS guidance and decided cases, not tax or legal advice. Compensation decisions turn on your own facts, so confirm the position with your CPA or counsel before acting.