Where the Money Goes: Corporate Tax Breaks, Buybacks and the Pay Gap

In Part 1 we said SNAP and Medicaid aren't socialism, and that the fair worries about them are about cost and results. This part turns the same questions toward the money that goes to companies. The usual case is simple: cut business taxes and companies will invest, hire, and raise pay. That's a real theory. So what happened when we tried it?

The Joint Committee on Taxation estimated three large corporate tax breaks for fiscal 2024: the lower rate on foreign subsidiary income ($57.0 billion), the research credit ($20.4 billion), and the export-income deduction known as FDII ($13.8 billion). That's about $91 billion by our sum, close to the $95.8 billion in SNAP benefits paid in fiscal 2025. (JCT cautions that its estimates aren't meant to be added, and these predate the 2025 tax law.)

The two sides also overlap. In a July 2026 report, GAO estimated that 13.8 million adults on Medicaid and 10.6 million adults in SNAP households worked at some point in 2024, and about two-thirds of them worked full time. In the 11 states that shared employer data, 46 companies were among the top 25 employers of Medicaid or SNAP recipients in at least two states, and 17 of those were among the 50 largest Fortune 500 companies by number of employees. The employer data cover only those 11 states, so they are not a national ranking.

Who actually gains when corporate taxes fall? Congress's own scorekeepers have an answer. Over the long run, JCT assigns 25% of the corporate tax to workers and 75% to owners of capital, and in the short run it assigns 100% to capital. CBO uses the same 25/75 split, and Treasury puts labor's share at 18%. (Some economists argue labor bears more, and CRS notes that debate.) So even before looking at any data, the government's own models say most of a corporate tax cut goes to shareholders and owners. A corporate tax cut doesn't split evenly between employer and employee.

The 2017 Tax Cuts and Jobs Act cut the corporate rate from 35% to 21% (CRS R45736). The Congressional Research Service looked at the first year.

• Wages. Real wages grew 2.0% in 2018, slower than the economy's 2.9%. For production and nonsupervisory workers it was 1.2%. CRS found "no indication of a surge in wages in 2018."
• Bonuses. Announced bonuses totaled about $4.4 billion, or $28 per worker, equal to 2% to 3% of the corporate tax cut.
• Buybacks. CRS reported that much of the money went to a record $1 trillion in announced stock buybacks by the end of 2018.

S&P Dow Jones Indices counted what S&P 500 companies actually spent: a record $806.4 billion on buybacks in 2018, up 55.3%, plus $456.3 billion in dividends, up from $419.8 billion. Together, a record $1.263 trillion went to shareholders. S&P's own analyst said corporations "used their savings from lower tax rates to push dividends and buybacks to record levels" (S&P Dow Jones Indices). Next to that, $4.4 billion in bonuses is about one third of one percent, by our math.

The best study so far uses actual tax records. Four economists, publishing in the September 2026 American Economic Review (Kennedy, Dobridge, Landefeld and Mortenson), compared similar firms that did and didn't get the full rate cut. The firms that got it grew their capital stock 4.6% and employment 1.3% more, a real gain. But below the 90th percentile of each firm's pay scale, the earnings effect was statistically indistinguishable from zero. The authors estimate 60% of the gains went to firm owners, 8% to executives, 32% to other top-10% earners, and 0% to the bottom 90% of workers. Counting workers' own stock ownership, "in the short run, 87 percent of private income gains flow to the top 10 percent" (AER).

Walmart is a fair example because its own numbers are public. On January 11, 2018, it announced an $11 starting wage costing about $300 million a year and one-time bonuses of up to $1,000, costing about $400 million. The bonus depended on length of service, and only workers with at least 20 years qualified for the full $1,000. The same day, Sam's Club announced it was closing 63 clubs, converting up to 12 into e-commerce centers. Months earlier, on October 9, 2017, before the tax bill passed, Walmart's board had approved a $20 billion buyback program. The bonus was about 2% of the buyback, by our math.

To be fair to Walmart: the buyback was planned before the law, raises came in a tight labor market, and it had raised pay before, announcing in 2015 a $9 entry wage and a $10 floor for current workers by February 2016.

Supporters say judging a tax cut by its first year misses the point. The goal was more investment, which raises productivity and wages over time. CRS itself found investment "grew significantly" in 2018, though it said the pattern did not clearly match the tax changes (CRS R45736), and the AER study found real gains in investment and employment. Buybacks aren't money burned, either: shareholders can reinvest it, and many are retirement savers. Those are real points. The question is whether "eventually" ever reaches the typical worker, and the evidence so far says not much.

This isn't only about 2017. For decades after World War II, typical worker pay grew in line with productivity. The Economic Policy Institute finds that from 1979 to mid-2026, net productivity grew 93.7% while compensation (wages plus benefits) for typical private-sector workers grew 32.7%. At the top, CEO pay at the 350 largest firms rose 1,316% from 1978 to 2025, compared with 28% for typical workers. CEOs made 20.6 times what a typical worker made in 1965 and 30.7 times in 1978. In 2025 it was 324.7 times (EPI's 2025 figure is projected from partial-year data).

The counterpoint. Economists across the spectrum have challenged EPI's figures. Martin Feldstein argued that once you count total compensation and use the same price index for both lines, pay tracked productivity closely (NBER, 2008). Heritage's James Sherk found productivity rose 100% from 1973 to 2012 and average compensation 77%, using one inflation measure for both (Heritage, 2013). They have a point: choices of measure matter. A Bureau of Labor Statistics study of 183 industries found the gap shrinks in 87% of industries when you use a different price index.

Why the gap still holds. That same BLS study found gaps in 83% of industries even with the same price index on both lines, and said those gaps "came from a declining labor share of income" (BLS). The critics' best case uses average compensation, which is pulled up by top earners; Sherk himself grants that median pay has not kept up with average productivity. The typical worker is the one the debate is about. EPI already includes benefits in its line, and says counting benefits "does not materially close this gap at all" (EPI). And no choice of price index turns 31 to 1 into 325 to 1.

None of this means companies are villains or that tax policy should never help business. It means the claim "a corporate tax break is a raise for workers" doesn't hold up well. Congress's own models assume most of the money goes to owners. The 2018 data showed record payouts to shareholders and small, one-time bonuses. The best tax-record study found the gains went mostly to the top 10%. If we expect SNAP to prove it helps the people it's meant for, we can ask the same of a tax break.

Previous, Part 1 of 3: Is it socialism? What the word means, what Republicans built, and how to talk about it.

Part 3 of 3: Same rules for every dollar. SNAP gets audited to the decimal point. The Pentagon has never passed an audit. We look at contracts, broadband money sent to satellite companies, and who has to prove results.