Have you ever watched a market rhyme with an old crisis and felt that familiar knot in your stomach? I had that feeling this week. Rising US Treasury yields, a sloppy Japanese yen, and a crowd that cannot stop talking about technology. Those three notes were playing in the mid-1990s too. Then the music broke. Currencies collapsed, banks failed, and whole economies in Asia were forced into recession. The question now is not whether the melody sounds familiar. It is whether the instruments are the same.
What Looks Familiar And What Does Not
A senior bank economist recently laid out a comparison that is hard to ignore. The setup before the Asian financial crisis of 1997 included climbing American bond yields, a yen that had already lost a huge amount of ground, and a market story built on new technology. Swap the internet for artificial intelligence and you get a chart that looks uncomfortably close to today. I do not think that means we are walking into a carbon copy of 1997. I do think it means investors should stop treating Asia as a simple growth story and start treating it as a demand story.
That shift matters. In the 1990s the region imported capital. Savings were not enough to fund investment. When dollar funding got expensive and the yen unnerved people, money left and balance sheets cracked. Today most of Asia exports capital. Higher US funding costs sting in a different way. They can cool the American appetite for chips, servers, and the electronics stack that has been carrying growth in South Korea, Japan, Taiwan, and Singapore. Demand vulnerability is the phrase that stuck with me. It is less dramatic than a currency crash. It can still flatten a cycle.
The Yield Shock That Set The Stage Then And Now
Go back to the mid-1990s for a second. The US 10-year yield climbed from about 5 percent in late 1993 to around 8 percent by late 1994. Even in the spring of 1997 it was still near 7 percent, roughly two percentage points above where it had been four years earlier. That was not a footnote. It repriced every dollar loan in the world.
Fast forward. The same benchmark sat near 0.5 percent in August 2020. Early this week it was hovering close to 4.79 percent. Yes, that climb took years. The uncomfortable part is the speed of the latest leg. From about 3.9 percent in February, the yield jumped roughly 80 basis points in a matter of months. I have found that markets forgive a slow grind. They punish a sudden step when balance sheets are already tight.
Washington also said it would lean harder into buybacks focused on the 10-year to 30-year part of the curve, at least doubling the maximum size of those operations. That is not a rescue. It is a plumbing move. Still, it tells you officials are watching the long end because the long end is doing real work in the economy. Higher long rates raise the cost of factories, data centers, and the kind of patient capital that funds an AI buildout. If that buildout slows, Asia feels it in export orders before New York feels it in headlines.
Rising dollar funding costs and a wobbly yen were key catalysts for stress when Asia still needed foreign money to pay its bills.
That line is the heart of the 1990s story. It is not the heart of the 2026 story. The region is not begging for hot money the way it once did. Households and firms save more. Current accounts in several economies run surplus. Sovereigns hold large buffers. So the same yield spike does not automatically trigger a funding run. It can still trigger a growth scare. Those two outcomes look similar on a price chart for a week. They are not the same for a year.
The Yen As A Global Mood Ring
In April 1995 the yen traded near 80 per dollar, a cycle low that felt extreme at the time. By April 1997 it had moved toward 130. That is a depreciation of about 55 percent. Investors who had borrowed yen to buy higher yielding assets elsewhere started to sweat. When that trade unwinds, it does not unwind politely.
Look at the recent path. From a low near 103 in January 2021 the yen slid toward 163 in July. That is a drop of about 57 percent, almost a mirror of the mid-1990s swing. Joint action from Washington and Tokyo then pulled the currency back toward 160. Markets are still guessing whether another intervention is coming. I will not pretend I know the next print. I do know that a currency this weak becomes a global funding story even when Japan itself is not in crisis.
Why? Because the yen is still a funding currency in the minds of many desks. A sudden bounce can squeeze carry trades. A sudden slide can make risk managers cut exposure to anything that looks like Asia. Neither move needs a local banking collapse to matter. The yen is a mood ring. When it shakes, people check every other position they own.
Perhaps the most interesting aspect is how little this has to do with tourism or import prices in Tokyo, at least for the rest of the region. The question for Seoul, Taipei, and Singapore is whether a messy yen episode tightens global dollar liquidity just as US rates are already high. That combination can delay capex. Delayed capex is delayed chip demand. Delayed chip demand is a softer print for Asian factories.
Tech Optimism Then, AI Optimism Now
Before 1997, markets were infatuated with the internet. New networks, new software, new hardware. The story was real. The valuations and the funding assumptions were not always real. Today the crowd has a new object of affection. Artificial intelligence needs power, chips, memory, networking gear, and a long list of components that Asia is unusually good at making.
That is the bull case in one sentence. It is also the vulnerability. If US funding costs stay high, some of those data-center plans get stretched. If equity markets wobble, some of those plans get cut. Asia does not need a local credit crunch for growth to fade. It only needs American buyers to pause.
In my experience, the most dangerous phase of a technology boom is not the first wave of spending. It is the moment when everyone assumes the first wave is a permanent floor. Electronics exports have been propping up growth in several export platforms. That is a gift. Gifts can be withdrawn. I would rather watch order books than listen to keynote speeches.
From Financial Vulnerability To Demand Vulnerability
This is the part that should change how you think about the region. In the 1990s Asia was on the wrong side of the capital account. Money came in to fund investment that local savings could not cover. When the money left, banks and corporates discovered they had borrowed short in dollars and invested long in local assets. That mismatch is what turned a yield shock into a crisis.
Today the region is largely on the other side. Many economies export capital. They invest abroad. They do not need a constant inflow of foreign deposits to keep the lights on. Higher US yields still matter, but they matter as a tax on global demand rather than as a fuse on local funding. A weak yen still matters, but it matters as a risk-off signal rather than as proof that Asian pegs are about to snap.
Instead of a financial vulnerability as in the 1990s, Asia now faces a demand vulnerability.
I like that framing because it is honest. It does not sell a fairy tale about invulnerability. It also does not sell panic. If US bond yields and funding costs weigh on the AI hardware boom, or if the yen unsettles global funding markets, demand for the region’s goods can buckle. Growth then fizzles. That is a slow bruise, not a broken bone. Slow bruises still change earnings, politics, and elections.
Why The Old Crisis Playbook Fits Poorly
People love a clean analogy. Clean analogies are often lazy. The 1997 episode had a few ingredients that are simply not sitting on the table in the same way.
- Fixed or tightly managed exchange rates that invited one-way bets
- Heavy short-term dollar borrowing by banks and firms
- Thin foreign-exchange reserves in several economies
- Current-account deficits that required constant external financing
- Weaker supervision and less transparent corporate balance sheets
Most of those items have been rewritten. Floating rates or more flexible regimes take pressure off reserve books. Reserve stocks themselves are larger. Banks fund more in local currency. Corporates still borrow in dollars, sure, but the share and the tenor look healthier in many places. Regulators remember 1997. Memory is not a guarantee. It is a speed bump.
So if you are waiting for a replay of currency collapse, capital flight, and cascading bank failures across the region, you may be waiting for a movie that already left the theater. The risk that replaced it is quieter. It lives in export volumes, factory utilization, and the hiring plans of chipmakers. Quiet risks are easy to dismiss until a quarter of missed shipments shows up in the data.
The AI Hardware Chain That Now Carries Growth
Walk through the map. South Korea sells memory and displays. Taiwan sits at the center of advanced logic. Japan supplies equipment, materials, and specialty parts. Singapore is a hub for trade, finance, and high-value manufacturing. When American firms race to build training clusters, those four economies feel the surge first.
That concentration is a strength in a boom. It is a bottleneck in a pause. I keep coming back to a simple test. If the cost of capital in the United States stays elevated, which projects still clear the hurdle rate? The giant hyperscale plans probably do. The second-tier buildouts, the speculative campuses, the “we might need this in 2028” warehouses of gear? Those are the first to slip. Asia does not get paid on press releases. It gets paid on shipments.
There is another layer. AI hardware is not a single product. It is a stack. A delay in one layer can stall another. Power constraints in the United States already complicate timelines. Add expensive money and you get a queue that stretches. Stretch the queue and you stretch Asia’s export cycle.
A Side-By-Side Look At The Two Eras
| Theme | Mid-1990s Setup | Current Setup |
| US 10-year yields | Rose toward 7 to 8 percent | Up from 0.5 percent to near 4.8 percent, with a sharp 2026 jump |
| Japanese yen | Weakened about 55 percent into 1997 | Weakened about 57 percent from 2021 highs before intervention |
| Market narrative | Internet optimism | AI hardware boom |
| Asia’s capital position | Importer of capital | Exporter of capital in most large economies |
| Main risk channel | Funding and currency stress | Export demand and capex delay |
| Policy buffers | Thinner reserves, rigid pegs | Larger reserves, more flexible rates |
Tables flatten nuance, I know. Still, this one helps. The left column is the rhyme. The right column is the reason you should not trade 2026 as if it were 1997 with better phones.
How Higher Yields Travel Into Asian Factories
Think of a US chief financial officer staring at a 10-year rate that has jumped 80 basis points since winter. The board still loves the AI story. The board also loves a clean interest-coverage ratio. So the company keeps the flagship cluster and delays the second site. That delay is invisible in Silicon Valley for a quarter. It is visible in a Taiwanese foundry booking meeting the same week.
Multiply that decision across a dozen buyers and you get a softer export pulse. Multiply it across a year and you get a growth scare in economies that had been leaning on electronics to offset weak domestic demand. This is not mysterious. It is arithmetic. High rates tax long-duration assets. Data centers are long-duration assets. Chips are the steel of that asset class.
Could buybacks on the long end of the Treasury market ease some of that tax? A little. Doubling the cap on operations is not the same as a structural decline in yields. Do not confuse a plumber with an architect. The plumber keeps the pipes from bursting. The architect decides how tall the building can be.
Intervention Talk And Why It Still Matters
Markets are weighing the chance of another official push to support the yen. Intervention is a headline machine. It is also a temporary tool. If rate differentials stay wide, the yen can drift again after the cameras leave. That drift is what funds managers care about, because drift invites leverage, and leverage invites crowded exits.
I have watched enough of these episodes to know the pattern. First comes the verbal warning. Then comes a visible print in the market. Then comes a week of relief. Then someone asks whether the relief changed the fundamental gap between US and Japanese policy rates. If the answer is no, the old trade creeps back. The creep is the risk, not the press conference.
For the rest of Asia the issue is correlation. A yen shock that forces global desks to cut risk can hit regional equities even when local current accounts look fine. That is annoying. It is also tradeable if you know you are dealing with a correlation spike rather than a solvency event.
What “Better Insulated” Actually Means
You will hear that Asia is better insulated. That sentence is true and incomplete. Insulated from what? From a classic sudden stop, yes, in most of the large economies. From a slump in US tech capex, no. Insulation is not a force field. It is a thicker coat. You still feel the wind.
Household savings, surplus current accounts, and deeper local bond markets are the coat. They mean governments can fund themselves without sprinting to offshore creditors. They mean banks can roll local liabilities. They do not mean a Korean exporter is indifferent to a cancelled server order. Do not mix those two ideas. Mixing them is how people get blindsided by “unexpected” growth downgrades.
- Check the capital account first. If a country exports capital, a yield spike is less likely to trigger a funding crisis.
- Then check the export mix. If electronics tied to AI are doing the heavy lifting, a US pause becomes the main threat.
- Then check policy space. Flexible rates and large reserves reduce the chance that a growth scare becomes a currency scare.
That three-step filter is how I keep the 1997 ghosts in the right box. Useful history. Wrong default template.
Where The Pain Would Show Up First
If the demand channel is the live wire, the first sparks are not bank runs. They are inventory adjustments. Chipmakers trim utilization. Component suppliers push out delivery dates. Shipping lanes that had been tight ease a little. Equity analysts cut export forecasts and pretend they saw it coming.
Labor markets in export hubs would be next. Overtime fades. Contract workers roll off. That is politically sensitive in places that sold the AI wave as a national project. Currency markets might still stay orderly. Orderly currencies and messy factories can coexist. That pairing would confuse anyone still fighting the last war.
Financial stress is not impossible. It is just not the base case in the same way it was three decades ago. The names that would feel a demand slump first are the ones whose revenues track global tech capex with little domestic ballast. Diversified service economies would lag the hit. That lag can create a false sense of safety in regional indexes that mix both groups.
A Personal Read On The Crowd
I will say this plainly. The market is better at spotting rhymes than endings. Everyone can see that yields, the yen, and tech fever showed up before. Fewer people want to sit with the boring conclusion that the transmission belt has changed. Boring conclusions do not go viral. They do keep you from selling the wrong thing at the wrong time.
If I had to pick a mistake investors will make, it is treating every wobble in Asian assets as the start of a 1997-style contagion. The second mistake is the opposite: treating surplus savings as a shield against any slowdown at all. Reality sits in the middle, which is where reality usually sits. Annoying, I know.
Another habit I keep seeing is the habit of talking about “Asia” as if it were one balance sheet. It is not. Commodity exporters, tourism recoveries, and electronics platforms do not share the same pulse. An AI pause hits the electronics platforms hardest. A yen shock hits anything priced off global risk appetite. Those are overlapping circles, not the same circle.
Policy Choices That Could Soften The Blow
Domestic demand can fill some of the gap if exports fade. That sentence is easy to write and hard to execute. Aging populations, high household debt in some markets, and cautious consumers limit the size of the offset. Still, targeted support for small firms, public investment that does not crowd out private capex, and a refusal to defend unrealistic currency levels can keep a demand shock from becoming a financial shock.
On the US side, the path of yields remains the swing factor. A stable long end that stops climbing is already a gift to planners of multiyear hardware projects. A further grind higher is a tax. Intervention in the yen can buy time. Time is useful only if someone uses it to reduce leverage in crowded carry trades. Time without de-leveraging is just a pause between two similar headlines.
How To Watch The Next Six Months Without Getting Hypnotized
Forget the anniversary essays for a minute. Watch a short list of live indicators. Export orders for electronics. New orders versus inventories at chip-related manufacturers. The 10-year Treasury yield and the shape of the long end. The yen’s realized volatility, not just its level. Announced delays in US data-center construction. None of those require a theory of history. They tell you whether the demand channel is opening or closing.
A simple watchlist: Yields: is the long end still lurching higher? Yen: is volatility rising with the level? AI capex: are projects slipping or only speeches changing? Asia: are electronics shipments cooling before the rest of trade?
If three of those four flash amber at once, growth forecasts for the export platforms should come down. If only the historical rhyme is flashing, you can keep your seat. History is a teacher. It is a terrible autopilot.
The Difference Between A Scare And A Crisis
A crisis in 1997 meant broken pegs, frozen interbank markets, and emergency programs. A scare in 2026 would mean missed export targets, softer industrial production, and a debate about how much of the AI boom was pulled forward. Both can hurt portfolios. Only one rewrites the political map of a region for a decade.
I would rather prepare for the scare and stay open to the crisis than the other way around. Preparing for the scare means stress-testing earnings that assume endless server demand. Staying open to the crisis means watching funding markets and the yen for signs that a correlation spike is turning into a liquidity event. Two dashboards. One brain. That is the job.
The region is not begging for hot money the way it once did. It is begging, quietly, for American buyers to keep spending on machines.
That is the sentence I would tape to a monitor. It keeps the comparison honest. It also keeps the optimism in check. Asia built real strength after 1997. Strength does not cancel exposure to the customer who pays for the boom.
What Investors Keep Getting Wrong About Contagion
Contagion in the old sense was a banking and currency chain reaction. One peg falls, another looks expensive, foreign banks pull lines, local banks hoard cash, trade finance seizes. That chain is harder to ignite now. What still travels fast is risk appetite. Equity correlations spike. Credit spreads in dollars gap. Portfolio flows reverse for a month. Then, if the local funding base is sound, the dust settles and the real-economy story takes over again.
People mix those two contagions because they feel the same on day one. They do not feel the same on day ninety. Day ninety is when you find out whether factories or balance sheets were the weak point. I would bet on factories in this cycle. I could be wrong. If I am wrong, the warning signs will show up in dollar funding markets and in the yen, not in a single disappointing export print.
A Note On Savings, Power, And The Next Bottleneck
Asia’s savings surplus is a shock absorber. It is also a reminder that the region is financing other people’s spending, including America’s technology build. That is a position of strength until the spender flinches. Then the surplus looks like unused capacity.
Power is the other bottleneck hiding in plain sight. You can print chips faster than you can permit substations. If US power constraints slow the installation of new clusters, Asia’s shipping schedules slip even if money is available. Yields and watts. Not a pairing you saw in 1997. Very much a pairing you should watch now.
Putting The Pieces Together Without The Drama
So where does that leave a reader who just wanted a straight answer? The similarities are real. Yields up. Yen weak. Technology in fashion. The differences are larger. Asia funds itself better. The live risk is a stall in US demand for the hardware that has been carrying regional growth. That stall would hurt. It would not automatically recreate the wreckage of the late 1990s.
If you manage money, separate the history lesson from the portfolio. Use 1997 to remember how fast dollar liquidity can matter. Use 2026 data to decide whether the customer is still buying. If the customer is still buying, the rhyme is a curiosity. If the customer blinks, the rhyme becomes a warning label on export-heavy earnings.
I keep circling back to that first knot in the stomach. It is useful. It means you are paying attention. It is not a trading signal by itself. The signal is in the transmission. Funding then. Demand now. Get that right and the rest of the conversation gets quieter, which is usually when the better decisions get made.
A Closing Checklist For Anyone Still Comparing The Two Cycles
Ask whether the economy in front of you imports or exports capital. Ask whether its growth is riding a single foreign capex wave. Ask whether its currency regime can flex without a political crisis. Ask whether banks fund in the same currency as their assets. If you answer those four questions honestly, you will stop treating every Asian dip as the sequel nobody asked for.
And if the yen lurches again next week? Check the carry, check the vol, and check the order books. Then decide. That sequence is less exciting than a crisis anniversary essay. It is also how you stay solvent while other people argue about rhymes.
