Earnings Get A Tariff Boost

Q2 2026 earnings season is almost complete, and corporate America is putting up numbers not seen in nearly five years. With 88% of S&P 500 companies reporting, the earnings growth rate for the S&P 500 stands at an explosive 50.4% year-over-year — the second consecutive quarter of earnings growth above 20% (28.6% last quarter) and the strongest pace since Q3 2021. The tech companies, which report late in the season, did not disappoint: Information Technology earnings grew 70.4%, after 54.3% last quarter. Revenues are growing at 15.0%, the fastest since Q2 2022 and double-digit for a second straight quarter.

On the surface these results describe an unbelievably booming economy. The problem is that while the actual economy is booming, it is not running at the level that the earnings numbers point to. The Federal Reserve’s June projections have real GDP growing 2.2% this year and the unemployment rate ending the year at 4.3%. Add roughly 3.5% inflation to that 2.2% real growth, and nominal GDP is growing at about 6% per year. Yet S&P 500 revenues are growing at more than twice the pace of nominal GDP, and earnings are growing at more than three times the pace of revenues.

Part of the answer sits in plain sight, and it is worth setting aside first. Two outsized one-time valuation gains are inflating the aggregate this quarter. Alphabet posted a $98 billion net gain in other income, primarily unrealized markups of its equity stakes, reportedly led by SpaceX and Anthropic. Amazon booked a $53.4 billion gain, mostly from its investment in Anthropic. Neither is operating profit: for the most part nothing was sold — the holdings were revalued, and the paper gains flowed through reported earnings as accounting rules require. Exclude those two companies entirely — a crude cut, since it discards their genuine business profits along with the gains — and the index’s earnings growth falls from 50.4% to 32.0%; remove only the gains themselves, after tax, and it comes out near 29%.

But that subtraction deepens the mystery rather than resolving it: 32.0% earnings growth on 15.0% revenue growth is still an extraordinary result. Some of the revenue story is known — the AI investment boom continues to concentrate enormous spending in the tech sector, and that spending flows directly into the revenues of the semiconductor, cloud, and data center infrastructure companies. However, operating leverage on tech capital expenditures cannot explain a 17 percentage point gap between revenue growth and earnings growth for the index as a whole, even with the valuation gains removed. So we have a mystery: where is this earnings explosion coming from?

Exhibit 1

An earnings boom the economy cannot explain

Year-over-year growth, Q2 2026

All figures as stated in this article. Earnings and revenue growth with 88% of S&P 500 companies reported, per FactSet Earnings Insight, August 7, 2026; nominal GDP growth per Federal Reserve June 2026 projections plus prevailing inflation.

To solve it, let’s go back one year, to when we examined Q2 2025 earnings last August. In that article, we found tariffs hiding in the cross-section of results. Aggregate earnings grew a healthy 10.3% on 6.0% revenue growth. But firms earning the majority of their revenues domestically — the firms actually paying tariffs — grew earnings just 8.7%, while firms with majority-foreign revenues grew earnings 13.3% despite identical revenue growth. The conclusion was that companies were absorbing tariff expenses in their margins rather than passing them on to consumers, and that a weakening dollar (down 7.1% that quarter) was inflating the reported earnings of the foreign earners.

Then, on February 20 of this year, the Supreme Court changed everything. In a 6–3 decision authored by Chief Justice Roberts, the Court held that the International Emergency Economic Powers Act (IEEPA) does not authorize tariffs, striking down the 10% baseline tariffs, the country-by-country reciprocal tariffs, and the fentanyl- and immigration-related tariffs on China, Canada, and Mexico. The sector tariffs imposed under Section 232 (steel, aluminum, autos) and Section 301 (China) survived. But the bulk of the tariff edifice — the part that raised customs duties collections from $79 billion in calendar 2024 to $264 billion in calendar 2025 — was gone.

The ruling applied retroactively, too: approximately 330,000 importers had paid an estimated $166 billion in IEEPA duties, and the government now has to give it back — with statutory interest, at 6% for corporations, compounded daily from the date of deposit. Customs and Border Protection began issuing refunds in April. As of July 31, roughly $100 billion had been repaid (duties plus interest), refund claims totaling about $129 billion had been accepted for processing, and total refunds are projected to reach roughly $130 billion before interest — with $49 billion paid out in June alone.

Exhibit 2

Giving it back

IEEPA tariff duties and refunds, cumulative, as of July 31, 2026

As stated in this article. U.S. Customs and Border Protection declaration filed with the Court of International Trade, data as of July 31, 2026; projected total per Cato Institute. The refunded figure includes statutory interest.

Now the earnings mystery resolves itself, because the Court’s ruling hits Q2 2026 earnings from two directions simultaneously. First, the base effect: the year-ago quarter we are comparing against is precisely the quarter in which tariff expenses were compressing the margins of domestic-revenue firms. The heaviest of those expenses — the steep country-specific IEEPA rates — have vanished, so the margin recovery alone manufactures several points of year-over-year earnings growth without anything getting better in the underlying economy. Second, the windfall: refunds of duties expensed in prior periods flow back through current income statements, along with the interest.

Some rough arithmetic shows the magnitude. The S&P 500’s aggregate earnings ran at roughly $550 billion per quarter last year. About 2/3 of the refund dollars paid or authorized during Q2 accrued to S&P 500 members. If these are recognized in the quarter — about $57 billion before tax, call it $45 billion after — that would be about 8 percentage points of “earnings growth.” Set the Alphabet and Amazon valuation gains aside, and the index still grew 32.0% against last quarter’s 28.6%. Take the refund arithmetic out and this quarter’s growth would actually trail last quarter’s — and that is before even counting the first channel, the margin recovery built into the year-over-year comparison. So, most of what looks like an earnings boom, in other words, is a one-time gift from the Supreme Court.

The acid test will be the same cross-section that revealed the tariffs in the first place, now running in reverse. The domestic-revenue firms that suffered the tariff drag last year should be the fastest growers this year: they get the expense relief, they get the refunds, and they are insulated from the currency reversal. Because the dollar has flipped too — climbing from a January low near 96 on the DXY index to a late-June peak above 101, up on the year even after easing in recent weeks — currency translation has turned from tailwind to headwind for the foreign earners that led a year ago. If the full-quarter results show domestic-majority firms outgrowing foreign-majority firms by roughly the margin they trailed by last year, we have our culprit confirmed.

The test has a shelf life, though. Washington has been rebuilding the tariff wall under other authorities — a flat import surcharge through late July, then new Section 301 duties of 10 to 12.5% on 60 trading partners — so the expense relief will partly fade in the quarters ahead, and exactly how is not yet knowable: the review setting country-by-country rates for the sixteen economies under investigation is expected to conclude as soon as this month. The refund windfall, by contrast, keeps flowing no matter what gets rebuilt — roughly $30 billion of the projected total, plus interest, is still to be paid.

The ruling also resolved — prematurely — the inflation question. Last year we argued that firms would pass tariff costs on to consumers slowly, over many quarters, producing a drawn-out rise in inflation into the 3.5% to 4.0% range rather than the one-time price adjustment many were predicting. That is what happened: CPI inflation climbed steadily as firms worked tariffed inventory through their cost of goods even after February, and an energy shock amplified it — the headline rate reached 4.2% year-over-year in May, with energy prices up 23.5% on the year, accounting for over 60% of that month’s increase, while core inflation, which excludes food and energy, stood at just 2.9%.

The tariff half of that pressure is now ending mid-stream: as late as May, the New York Fed’s business surveys still found roughly 45% of tariff-paying firms with price increases in the pipeline, but the ruling had erased the bulk of the duties behind that pipeline — increases tied to the vanished duties will now never arrive, and some past increases are being unwound. June CPI fell 0.4% for the month (helped by a 5.7% drop in energy), bringing the annual rate down to 3.5%, with core inflation at just 2.6%. Prices are sticky downward, so the price level will not fully retrace. But the inflation impulse from the struck-down duties is gone — the replacement duties will start a pass-through cycle of their own, though from flatter rates — and the Fed’s own projections still show PCE inflation falling from 3.6% this year to 2.3% in 2027.

The fiscal picture is the one place where the news is unambiguously bad. Tariff revenue was supposed to help close the budget deficit, and briefly it did. Now the machine runs the other way: the rebuilt tariffs collect at nearly the old pace, while the Treasury simultaneously pays out $130 billion-plus of refunds along with 6% compounded interest. Relative to the revenue path projected a year ago, the refunds and their interest swing fiscal 2026 and 2027 by that full amount even with collections restored. This may be a large part of why the 10-year Treasury yield sits stubbornly at about 4.65% while inflation falls — the bond market is charging a growing term premium for a deficit whose newest revenue source has come back with a $130 billion refund bill attached. It will be interesting to see how the bond markets react as the remaining country-by-country Section 301 tariffs land.

For investors, the implication is about what happens when the gift stops giving. Wall Street analysts have extrapolated the earnings boom: consensus calls for 27.4% earnings growth in Q3 and 30.0% for calendar 2026, and the market is priced accordingly — the trailing P/E stands at 28.2, the highest since the dot-com era, with the forward P/E at 20.0 against a ten-year average of 19.0. But a meaningful slice of 2026 earnings is one-time refund income that will not happen again. In 2027 the comparisons flip: earnings growth will decelerate sharply, and the headlines may describe an earnings slowdown. Consensus is already there — the 2027 estimate is 13.6%, less than half the 2026 pace. That slowdown will be largely arithmetic, not economic — the mirror image of this year’s acceleration.

A year ago, the cost of tariffs was invisible in aggregate earnings and visible only in the cross-section. This year, the refund windfall is invisible in the aggregate multiple and visible only to those who look for it. As in law, so in earnings season: it pays to read the footnotes.