A strong-dollar headline gets pasted onto every US multinational before the filing is open. Foreign sales convert into fewer dollars, so reported revenue and earnings can fall while the local business was fine. I believe that part. I've also hung the same headline on the wrong day. The basket was wrong too, and the 10-K shrugged.
The index I actually open is the Nominal Broad U.S. Dollar Index, FRED series DTWEXBGS, published by the Board of Governors from the H.10 foreign-exchange release. It's a trade-weighted average of the dollar against a broad set of trading partners, covering goods and services. FRED quotes it as an index with January 2006 equal to 100. Nobody seasonally adjusts it. It's nominal, so a move here is an exchange-rate move, with no adjustment for inflation differences between countries. A print of 120 means the basket is stronger than it was in January 2006. I used to turn that into a 20 percent move in the euro, or into a 20 percent move in your company's Mexican peso exposure. Neither of those.
Where the basket has been
The daily file starts on January 2, 2006. Highest close in it: January 13, 2025, at 130.0413. When I downloaded the series the latest close was October 30, 2026, at 121.6656, which is 8.3757 index points under that peak, or 6.4 percent. A year earlier, on October 30, 2025, the close was 121.2832. From that day to October 30, 2026 the index rose 0.3824 points. That's 0.3 percent, a flat tape for the last twelve months. The move people remember is the retreat from the January 2025 peak, and most of that retreat was already in the books by the end of 2025. December 31, 2025 closed at 119.7456.
The lowest close of 2026 through October 30 was January 29, 2026, at 117.4396. I keep it in the table because an autumn 'the dollar is ripping' headline has to argue with a year that already spent time lower. I have lost that argument to a headline before. The series low since 2006 is much older than this window, and it doesn't belong in a 2026 earnings note.
| Date | DTWEXBGS close |
|---|---|
| Jan 13, 2025 | 130.0413 |
| Jun 30, 2025 | 119.4088 |
| Oct 30, 2025 | 121.2832 |
| Dec 31, 2025 | 119.7456 |
| Jan 29, 2026 | 117.4396 |
| Jun 30, 2026 | 120.9248 |
| Oct 30, 2026 | 121.6656 |
Those closes are the daily nominal broad index from FRED series DTWEXBGS. January 13, 2025 is the highest close in the file, and October 30, 2026 is the last close in the file I used. Blank days show up in the download. October 12, 2026 is one, a holiday or a missing print, and I left them out.
The year-end spot is the wrong average
I make this mistake when I'm in a hurry. An income statement is translated at the rates that applied while the revenue was earned. For a practical reading that means the average over the fiscal year, plus whatever hedges the company already had on. The close on the last day of the year is a spot. It remeasures the balance sheet. As a summary of the year's sales, it's poor.
Microsoft and Procter & Gamble both end the fiscal year on June 30. The broad index closed at 119.4088 on June 30, 2025 and at 120.9248 on June 30, 2026, so the spot was higher at the end of the later year, by 1.52 points, about 1.3 percent. Stop there and you say the dollar strengthened into their fiscal 2026. I would have, and I would have braced for an unfavorable translation.
The year-long average goes the other way, which is annoying if the note is already written. Stay in the FRED file, leave the blanks out, and July 1, 2024 through June 30, 2025 is 250 prints whose average is 124.5847. That window is fiscal 2025. Move to July 1, 2025 through June 30, 2026 and you again have 250 closes, this time averaging 119.8431, which is 3.8 percent lower. The January 2025 spike is what separates them. It sits in fiscal 2025 and is missing from fiscal 2026, so a slightly stronger June 30 spot can live inside a year whose average dollar was softer. Both companies then reported favorable foreign exchange. I should have trusted the average. The anniversary spot is how I got the sign wrong.
Microsoft printed two points, then a stress case
Microsoft's Form 10-K for the year ended June 30, 2026 says revenue was $331,839 million, up from $281,724 million, which the company calls an 18 percent increase. Operating income was $155,237 million, up from $128,528 million, a 21 percent increase. In that same discussion, revenue and operating income both included a favorable foreign currency impact of 2 percent. Earlier, the filing says fluctuations in the dollar relative to certain foreign currencies increased reported revenue and did not have a material impact on reported expenses from international operations.
I read the 2 percent as two points sitting inside the growth rate, in an 18 point revenue increase and a 21 point operating-income increase. I won't relabel the leftover as a constant-currency growth rate. Microsoft did not print that leftover as its own number, and currency can mix with price and volume in ways a subtraction hides. Two points on an 18 point year is a real contribution. It's also small. Expecting the 6.4 percent retreat from the January 2025 peak to show up as a 6 percent revenue tailwind means you imported the index. Microsoft hedges, and its currency mix isn't that basket. The company measures a fiscal year, not a peak-to-trough trade.
The geographic note is the half I used to skip, and the warning is in the footnote. Revenue from the United States was $170,794 million, revenue from other countries $161,045 million. The United States line includes billings to OEMs and certain multinational organizations, because the company finds it impracticable to pin down the geographic source. 'Other countries' is the cleaner foreign number, and even that is just under half of $331,839 million. It isn't a map of where the software is used. See Microsoft.
The market-risk table is a different document from that 2 percent. It estimates the potential loss in future earnings or fair values, including associated derivatives, from hypothetical market moves. For foreign currency against revenue, a 10 percent decrease in foreign exchange rates is shown as a $13,653 million hit, in the earnings column, as of June 30, 2026. Investments, run through the same hypothetical, are a $5 million item. That gap is how you can tell the revenue exposure is the one they consider worth measuring. Nothing there says the shock equals a 10 percent move in DTWEXBGS. The paragraph I was reading also never defines the quote convention. I keep $13,653 million as Microsoft's own stress case, and I leave the broad index out of it.
P&G's sales growth was mostly the exchange rate
Procter & Gamble's Form 10-K for the year ended June 30, 2026 is blunter about the sales line, which I prefer. Net sales were $87,032 million, against $84,284 million the year before. The narrative says net sales increased 3 percent to $87.0 billion, driven by favorable foreign exchange of 2 percent and pricing of 1 percent. Unit volume and mix were unchanged. Organic sales, which the company defines as sales excluding acquisitions, divestitures and foreign exchange, increased 1 percent.
Two of the three points of reported sales growth were the exchange rate. Volume didn't grow. Tell yourself 'P&G grew 3 percent' and you've described the dollar translation plus a point of price, a split the filing already made. On earnings, foreign exchange impacts increased net earnings by approximately $224 million, which the company attributes to a strengthening of certain currencies against the dollar. Transactional charges are in that figure, and so is the translation of foreign subsidiaries. Most FX comments I see arrive as a growth-rate point. This one arrives as dollars. $224 million on net earnings of $16.1 billion, the rounded total the same section uses, isn't the whole story of the year. It's large enough that I write it next to the holding instead of waving at 'FX.'
The geographic split matches the exposure. In the note, United States net sales were $41.7 billion and international net sales were $45.3 billion. The MD&A says more than half of sales are generated outside the United States. Greater China, the United Kingdom, Canada, Japan and Germany together were approximately 21 percent of net sales. No other single country was more than 10 percent. What you're looking at is a pile of smaller currencies. The broad index is a better backdrop than a euro headline, and still a bad substitute for the 2 percent and the $224 million. See Procter & Gamble.
Coca-Cola, where the index barely moved and the peso did
Coca-Cola's year is a calendar year, so the June 30 lesson stays with the other two. The Form 10-K for the year ended December 31, 2025 reports net operating revenues of $47,941 million, up from $47,061 million, an increase of $880 million, or 2 percent. The bridge the company prints for that 2 percent is volume 1 percent, price and mix 4 percent, foreign currency exchange rate fluctuations negative 2 percent, and acquisitions and divestitures negative 1 percent. The currency line includes hedging. The unfavorable piece, the company says, was mainly a stronger dollar against the Mexican peso, the Argentine peso, the Ethiopian birr and the Turkish lira, partly offset by a weaker dollar against the euro, the pound, the South African rand and the yen.
I ran the same broad index over calendar 2024 and calendar 2025, and again I only kept days with a close. The 251 closes in 2024 average out to 122.8930. For 2025 I count 250 closes, and they average 122.7203, lower by 0.17 index points, about 0.14 percent. Flat basket. Coca-Cola still reported a 2 percent revenue headwind from currency, which is why the filing has to be opened. The Fed's trade weights are not Coca-Cola's revenue weights. A year can be quiet in DTWEXBGS and loud in the peso and the lira.
Earnings were the louder line. The same 10-K says currency fluctuations, again including hedges, unfavorably impacted consolidated operating income by 12 percent. Latin America's revenue bridge shows a 12 percent unfavorable currency column, and the segment discussion says operating income took an unfavorable currency impact of 21 percent. Latin America operating income was $3,742 million, against $3,792 million the year before, in a year when price and mix in that segment was an 11 percent help to revenue. Own the stock and the 21 percent is the figure I'd argue about. A revenue headline is going to quote the consolidated 2 percent.
The 2 percent needs a home, so here is the scale. The risk-factor section says the company derived $28.8 billion of net operating revenues in 2025 from operations outside the United States, out of the $47,941 million total. The outside piece is the majority of the company. The 10-K also said, based on spot rates and hedges then in place, that the company expected currency to be a favorable item for full-year 2026 revenue. That sentence is a February 2026 expectation in the annual report. It isn't a result, and I'm not going to update it from memory. What it does show is the sign flipping in the company's own telling, from one page to the next. Hedges and a moving spot will do that. See Coca-Cola.
| Filing | FX impact the company printed |
|---|---|
| Microsoft, year ended Jun 30, 2026 | Favorable 2% inside both revenue growth and operating-income growth |
| P&G, year ended Jun 30, 2026 | Favorable 2% of the net sales change; about $224 million more net earnings |
| Coca-Cola, year ended Dec 31, 2025 | Unfavorable 2% of the revenue change, including hedges; unfavorable 12% of operating income |
| Coca-Cola Latin America, 2025 | Unfavorable 12% of the segment revenue change; unfavorable 21% of segment operating income |
What I stopped doing with the headline
A dollar headline still reaches the portfolio through revenue earned in another currency. Listing country and the country on the sales footnote come apart often enough that I keep them in your US portfolio might not be a US portfolio. The habit of marking exporters down when the home currency rises is a sector habit, and it lives in which sectors react to which shocks. Neither of those hands me the number in this year's 10-K. The index is the backdrop. Mix, hedges, and the line the hit landed on are an MD&A job, and for Coca-Cola that hit was a different size in operating income than in revenue.
Valuation stays on its own page. A move in the dollar can change the discount rate people use, and that channel is about when cash arrives, which is why bond yields hit growth stocks. Translation of a sale that already happened is slower, and it's literal. I want that literal thing, in dollars or in growth-rate points, before a currency headline gets to reprice the whole position in my head.
What I write next to the holding
- Put the index close on the same line as the company's fiscal year. October 30, 2026 at 121.6656 does not describe a year that ended on June 30.
- Average the daily closes across the fiscal year before the anniversary spot gets a vote. For the June 30 companies, the June 30, 2026 spot was 1.52 points higher than June 30, 2025, and the fiscal-year average was 3.8 percent lower.
- Start from the FX sentence the company already printed. P&G's was about $224 million of net earnings. Microsoft only put 2 points inside an 18 percent revenue increase, and Coca-Cola split its figure into 2 percent of revenue and 12 percent of operating income, hedges included.
- If one region is doing the work, read that row twice. Coca-Cola's Latin America currency column was 12 percent of the revenue change and 21 percent of operating income, in a year when the broad index's calendar average barely moved.
- The geographic footnote comes before the word 'foreign.' Microsoft's United States revenue includes OEM and multinational billings the company says it cannot place. P&G's international net sales were $45.3 billion against $41.7 billion in the United States.
- A hedge moves the date, so the effect you 'should' have seen last quarter can be sitting in a forward book. Microsoft's $13,653 million figure is a 10 percent hypothetical on future earnings after derivatives, not last year's translation.
On a quiet day I decide which holdings actually earn outside the dollar, and the index stays in the macro note until the filing gives me a percent or a dollar figure. A 6 percent move in DTWEXBGS from the 2025 peak is a fact about a basket. It gets to be a fact about a stock after the company says so, and not before.
Educational reading of the dollar and reported results. Not investment advice. Index levels are daily closes of FRED series DTWEXBGS (Nominal Broad U.S. Dollar Index, Board of Governors of the Federal Reserve System, H.10), January 2006=100, through October 30, 2026: DTWEXBGS. Fiscal-year and calendar averages are the unweighted mean of those closes, blanks skipped. Company figures are from Microsoft's Form 10-K for the year ended June 30, 2026 (filing), Procter & Gamble's Form 10-K for the year ended June 30, 2026 (filing), and Coca-Cola's Form 10-K for the year ended December 31, 2025 (filing). Rates, hedges and mix change.