This Earnings Season Is Crushing Expectations. But the Underlying Growth Is Weaker
Corporate America is having one of its best earnings seasons in years, but there’s more going on beneath the numbers. Tariff refunds are padding profits just as analyst valuations are diverging sharply, complicating the picture for investors trying to gauge what companies are actually worth.
The refund windall comes with a catch
Over 40 S&P 500 companies have reported roughly $9.6B in tariff refunds in the past quarter. At least $2.1B of that has already landed as cash.
The refunds flow from a Supreme Court decision that invalidated a key part of President Trump's tariff policy, ordering the government to return duties collected under the International Emergency Economic Powers Act.
Apple leads the pack with nearly $2.2B reported. Nike follows at $986M. FedEx sits at roughly $800M. Amazon received $640M, and General Motors took in $500M.
Some companies get to keep the extra cash. Others are simply passing it along, making earnings look stronger without changing the underlying business.
The refund payday depends on the business
The Nike versus FedEx comparison shows why the same refund produces different outcomes for different companies.
Nike paid import duties on merchandise it owned. Those costs were absorbed into retail pricing. Because individual tariff payments can't be traced to specific shoppers after the fact, Nike has no practical obligation to refund customers. The full $986M flows to earnings.
Nike’s results look much less impressive without the tariff boost. Its $0.72 in earnings per share included a $0.52 benefit from the recovery, leaving just $0.20 in underlying EPS as revenue slipped 1%.
FedEx operates differently. It collected duties and taxes on behalf of shipping customers, and each payment remains linked to a specific shipment.
FedEx is distributing its full $800M refund back to shippers and consumers in phases beginning in August. Its Q4 strength came from operations, not refunds: revenue rose 13%.
Amazon says it will proactively refund customers in cases where it can trace specific import charges passed on to them. For everything else, it plans to use refunds to lower prices broadly.
Caterpillar shows the limits of the boost. It recorded $392M in expected tariff recoveries last quarter but still expects roughly $2.2B in tariff costs this year, excluding recoveries. The refund helps, but the cost pressure remains.
Record beats, but the bar keeps rising
The tariff refunds are landing inside an already-elevated earnings season. 85.2% of companies exceeded Wall Street's EPS expectations through mid-August, the highest share since 2021. S&P 500 profit growth is running above 30% for the quarter.
Bank of America strategists expect growth to stay above 20% through the rest of 2026 before moderating into the mid-teens for 2027.
When earnings growth is above trend but decelerating, the S&P 500's median 12-month return has historically been 6.7%, compared with 14% when growth is still accelerating.
Stocks that beat estimates are already seeing muted market reactions. Bloomberg Intelligence data shows companies that beat on revenue, earnings, or both have on average seen flat one-day excess returns. Misses have triggered steeper selloffs.
Analyst disagreement adds another layer of fog
Layered on top of inflated earnings is a separate problem: analysts can't agree on what these stocks are worth. The S&P 500 is seeing its highest price target dispersion in a decade.
The average gap between the highest and lowest analyst targets now stands at 62.5% of the consensus, well above the 43% average since 2016.
Nvidia targets range from $180 to $743 against a recent price around $219. That spread isn't unique to Nvidia. On an equal-weighted basis, the typical S&P 500 stock had a dispersion of 44% in July.
A Yale and Indiana University study found that stocks with highly dispersed targets historically deliver weaker returns, even when the average target looks attractive.
High dispersion can also signal stale estimates: when bad news hits, some analysts update immediately while others lag, keeping the consensus artificially elevated.
Retail investors who use the consensus target as a buy signal may be reading a number that doesn't reflect current views.
For 2026, the combination of tariff-boosted earnings and wide analyst disagreement means the headline numbers require an extra layer of scrutiny before drawing conclusions about a company's actual trajectory.
