This month something happened that I could not explain with anything I had written on this site before.
Nothing I own fell. The American companies inside my index funds kept doing what they do. No crash, no panic, no headline. And my account balance dropped anyway — by more than any single bad market day has ever cost me.
The cause was the exchange rate. The won strengthened against the dollar through July, and because my retirement accounts are Korean accounts holding American assets, the same dollars converted into fewer won. My wealth in dollars was untouched. My wealth in the currency I actually live in shrank.
I started this blog four weeks ago. In that short time I have written about what the S&P 500 is, what an ETF is, why I buy on a schedule, and how the Fed sets rates. Each of those posts treated its subject as a thing standing on its own.
They are not standing on their own. They are three dials on the same machine, and the dials are connected to each other by linkages I could not see.
I have spent my working life around machinery. My job is to understand systems where you turn one thing and three other things move. So this post is my attempt to draw the linkage diagram — honestly, including the parts where the diagram does not work.
Fair warning: this is the longest thing I have written here, and the conclusion is not a tip.
The rule everyone knows, and why it keeps failing
Ask anyone what interest rates do to stocks and you will hear the same sentence: rates up, stocks down.
I believed it. It has a satisfying mechanical feel — raise the cost of money, and the price of everything bought with money falls.
The problem is that it is wrong often enough to be useless.
Start with something that happened this month. The Fed held its policy rate at 3.5%–3.75% and did not move it once. Meanwhile the ten-year Treasury yield went from 4.48% on July 1 to 4.67% on July 29 — it rose about a fifth of a percentage point while the central bank sat still.
If rates were one lever, that could not happen. Rates did not do one thing in July. Different rates did different things, and the one the Fed controls was the one that did nothing.
Zoom out further and the rule gets worse. There have been long stretches when rates and stocks rose together for years, and stretches when they fell together. If the relationship were a simple lever, that could not happen either.
Here is what the research actually says, and it is the first genuinely deep thing I learned writing this.
The relationship between stocks and bonds is not fixed. It changes sign, and what flips it is the kind of shock hitting the economy.
Equities and bonds respond to growth news with opposite signs, and to inflation news with the same sign. When growth is the dominant worry, bad news hurts stocks and helps bonds — rates fall as stocks fall, and the two move opposite. When inflation is the dominant worry, bad news hurts both at once — rates rise as stocks fall, and the two move together.
A widely cited model shows that the relative volatility of growth and inflation, plus the correlation between them, explains roughly 70% of the long-run variation in the US stock–bond correlation. That is not a footnote. That is most of the phenomenon.
So the honest version of the rule is:
Rising rates hurt stocks when the rates are rising because of inflation. When rates rise because growth is strong, stocks often rise with them.
The same number moving in the same direction means two opposite things depending on why it moved. Anyone who tells you rates up equals stocks down is describing one regime and calling it a law.
What a rate actually does to a share price
Underneath the regimes, there is a mechanism, and it is worth getting exactly right because everything else hangs off it.
A share is a claim on money the company will produce in the future. To value it today, you take those future amounts and shrink them, because money arriving in ten years is worth less than money arriving now. The amount you shrink by is the discount rate, and the discount rate is built on top of the interest rate on government bonds.
Raise the rate, and every future amount shrinks harder. The price falls. That part really is mechanical.
But here is the piece that explains why some stocks get hurt far more than others.
How much a stock is damaged by a rate rise depends on how far in the future its money arrives.
A utility company earning steady cash today has most of its value in the near term. A company whose profits are expected mostly a decade from now has almost all of its value in the far term — and the far term is exactly where discounting bites hardest. Compounding works in both directions. A higher rate applied over ten years does far more damage than the same rate applied over two.
This is why rate scares tend to hit growth and technology names hardest while boring dividend payers barely move. It is not sentiment. It is arithmetic about when the money shows up.
I hold both kinds through index funds, which means I own the arithmetic in both directions and do not have to pick.
The Fed only holds one end of the rope
Now a distinction I had completely wrong until I wrote about the FOMC.
The Fed sets a target range for overnight lending between banks. Right now that is 3.5%–3.75%. But the rate that discounts a company’s profits ten years out, and the rate that sets a thirty-year mortgage, is the ten-year Treasury yield — and on July 29 that was 4.67%.
Almost a full percentage point above the top of the Fed’s range. The Fed did not set that number. The market did.
And remember what that number did during July. The Fed held its rate flat all month. The ten-year rose from 4.48% to 4.67% anyway. Same month, same country, same economy — one rate frozen by decision, the other drifting upward on its own.
The Fed’s grip is tightest at the shortest maturity and loosens as you go further out. Long yields reflect what buyers and sellers collectively believe about inflation and growth over a decade, plus the compensation they demand for tying money up that long. The Fed influences those beliefs. It does not dictate them.
This has a consequence people rarely say out loud: the central bank can hold rates steady and your mortgage can still get more expensive. They are different rates, set by different mechanisms. Watching only the Fed announcement is watching one end of a rope and assuming you know what the other end is doing.
The currency leg, where the textbook simply fails
Now the third dial, and the part I find genuinely strange.
There is a standard theory for how interest rates should move exchange rates. It is called uncovered interest parity, and the logic is clean: if Korean deposits pay more than American ones, money floods toward the won, the won rises immediately — and then must be expected to fall afterwards, because otherwise you could earn the higher interest and the currency gain at the same time, forever, risk-free. Markets should not leave that on the table.
Clean logic. It does not hold.
This failure is documented well enough to have its own name — the forward premium puzzle — and it is one of the most durable anomalies in international finance. Studies stretching across decades find that high-interest-rate currencies have not depreciated as theory requires. They have frequently appreciated instead. NBER and BIS working papers have been circling the problem for years without a settled answer.
The proposed explanations — risk premiums, rare disasters that skew expectations, limits on how much capital can chase the trade — are all partial. Nobody has closed it.
I want to be careful about what I am claiming. I am not saying economists are foolish. I am saying that the cleanest available theory of how rates move currencies is known by the people who built it to be contradicted by the data. That is the honest state of the field.
And it matters for me directly. If someone tells you the won will weaken because Korean rates are lower than American ones, they are quoting a theory that has failed empirically for forty years.
And it is not a small effect. According to Federal Reserve H.10 data, the dollar bought 1,538.05 won on July 2 and 1,460.76 won on July 24 — the won gained about 5% against the dollar in under four weeks. For a Korean holding American assets, that is roughly a 5% haircut on everything, delivered by a variable that has nothing to do with any company they own.
What actually moved the won this July was not a rate story at all. Reporting points to foreign money flowing into Korean semiconductor shares, and to policy encouraging Korean companies to bring overseas earnings home and convert them. Real flows of real money — bigger than the rate channel, and largely unpredictable.
Why I get hit twice
Now put the three dials together, from where I am sitting.
An American investor holding an American index fund has one transmission path. Rates move, the discount rate moves, the value of their holdings moves. One input, one output.
I have two.
Path one: US rates move → the discount rate on US companies moves → my index funds move. Fast, and it happens the moment the market reprices.
Path two: US rates move → the interest rate gap between the US and Korea moves → capital flows shift → the won/dollar rate moves → the won value of everything I own moves. Slower, noisier, and — because of the puzzle above — not reliably in the direction theory predicts.
Two paths, one input, different lags, and no guarantee they point the same way.
Sometimes they offset. US rates rise, my funds fall, but the dollar strengthens against the won and the conversion cushions the blow. It looks like nothing happened. It was two large forces cancelling.
Sometimes they compound, and that is the version that hurts — the month where the assets fall and the currency moves against you, and you take both.
And sometimes, like this July, path two moves on its own for reasons that have nothing to do with rates at all. Foreign investors buy Korean chip stocks, the won strengthens, and a man’s retirement balance falls in Korea while nothing whatsoever happens in New York.
That is the diagram. One input, two paths, a third input that arrives from outside the diagram entirely, and a linkage in the middle that the textbook admits it cannot describe.
What engineers do when they cannot predict the load
Here is where my working life actually helps.
When you design something that has to hold weight — a bridge, a crane, a pressure vessel — you never design it for the load you expect. You design it for the load you expect, multiplied. Engineers call that margin a safety factor, and it exists for one simple reason: you cannot know every load the thing will ever see.
You know the trucks that cross the bridge today. You do not know that in fifteen years the trucks will be heavier, or that a storm will push sideways in a way nobody modelled, or that salt will quietly eat a joint you cannot see. So you build in margin. The bridge ends up heavier and more expensive than it strictly needs to be, and that surplus is the entire point.
A bridge designed to exactly match today’s traffic is not efficient. It is fragile. It performs beautifully right up until conditions move, and then it does not perform at all.
That is the shape of the problem I have been describing for the last two thousand words. I cannot know which regime I am in — and I will only recognise it afterwards. I cannot predict the currency, because the best theory available has been failing for forty years. I cannot even assume the relationship between rates and stocks will keep the same sign from one decade to the next.
It is not that this machine is complicated. It is that the machine gets rebuilt while you are still measuring it — partly because people are measuring it and acting on what they find. Any model I finish will be a model of a machine that no longer exists.
So I stopped trying to design for the load I expect, and started designing for loads I cannot predict.
Buying the same amount on the same day of every month is a safety factor. It is not optimised for anything. It performs adequately whether rates rise or fall, whether the won strengthens or weakens, whether the regime turns out to be about growth or about inflation. It gives up the best case in exchange for surviving the worst one.
I used to think of my monthly purchase as the boring option — the thing you do when you are not smart enough to do something better. Writing this changed that. It is not the absence of a strategy. It is what you build when you are honest about what you cannot know. I am not buying on a schedule because I failed to understand the machine. I am buying on a schedule because I understand it well enough to know it cannot be predicted.
That is a much more comfortable thing to believe, which is exactly why I should hold it loosely.
What I might have wrong
A safety factor is not free, and I am underplaying the cost. The extra steel in a bridge is steel you paid for and may never need. Someone who correctly read one regime and leaned into it will beat me over that stretch, and the margin I am carrying is the price of not knowing. I am choosing to be reliably mediocre over being occasionally excellent and occasionally wiped out. That is a preference, not a proof, and I should call it what it is.
The engineering metaphor may be flattering me. Markets are not machines. Machines do not have opinions about you, and they do not change behaviour because you looked at them. The comparison is useful for structure and misleading if pushed too far. I have probably pushed it slightly too far somewhere in this post.
I cannot actually do anything about the currency. I have described a two-path problem and offered a solution to one path. Hedging the other has real costs and its own risks, and I have not done the work to have a view yet. Naming a problem is not solving it, and I do not want the tidy ending here to disguise that.
One month is not evidence. July’s currency move felt enormous because it was recent and it was mine. Over the horizon that actually matters to me — twenty years — currency effects have historically been far less important than whether I kept buying. I am aware that I have just written three thousand words partly because a number annoyed me.
Frequently asked questions
Do stock prices always fall when interest rates rise?
No. The relationship changes sign depending on whether the dominant shock is about growth or about inflation. When rates rise because inflation is rising, stocks and bonds tend to fall together. When rates rise because growth is strong, stocks often rise alongside them.
Why does the Fed’s rate differ from mortgage and bond rates?
The Fed targets an overnight rate between banks. Longer rates, such as the ten-year Treasury yield, are set by the market’s collective view of inflation and growth over that horizon. On July 29, 2026 the policy range was 3.5%–3.75% while the ten-year yield was 4.67% — and that ten-year yield had risen from 4.48% at the start of the month while the Fed held its own rate completely still.
Why do higher interest rates not reliably strengthen or weaken a currency?
Standard theory (uncovered interest parity) predicts that higher-rate currencies should subsequently depreciate. Decades of evidence show they often appreciate instead. The anomaly is called the forward premium puzzle and it remains unresolved.
Why did my foreign investments lose value when the market did not fall?
If you hold foreign assets in an account denominated in your home currency, your returns combine the asset’s performance with the exchange rate. If your home currency strengthens, the same foreign assets convert into less money even when their price is unchanged.
Should I hedge currency risk on foreign investments?
That depends on your horizon, your costs, and what the money is for — it is not a question with one answer. Hedging removes currency movement in both directions and is not free. I have not made a decision on this for my own accounts.
Does any of this change how I should invest?
For me it did the opposite of what I expected — it made the plainest possible approach look more defensible, not less. But that is a conclusion about my situation and my temperament, not advice for yours.
I am not a financial adviser, and nothing here is investment advice. I am an engineer writing down what I am learning as I go. Please make your own decisions, and speak to a qualified professional about your own situation.