JIAHON Group · JIAHON Think Tank perspective · Road to NT$100 Million, The Remedy. Published by JIAHON Think Tank | Written by Marcus (Senior Vice President) | Published 2 October 2026 | jiahon.com.tw
Part One measured three roads on the same ruler; Part Two found the switches that can flip the sign of starting a company. Neither answered one question: when. The popular answer is heroes and courage. This article does not discuss stocks and does not talk about heroes; it answers with the National Development Council’s business cycles, a few international studies and the same ruler — windows are not rare, and in Taiwan one comes on average every four years; what is rare is the person who, when the window comes, has fuel in hand and a floor underfoot.
In Brief
Parts one and two measured three roads to NT$100 million and the switches that flip the sign of entrepreneurship. Neither answered the question of timing. This part does—without stocks and without heroes. Taiwan’s National Development Council has dated seven business-cycle troughs since 1998, roughly one every four years; windows are not rare. What is rare is reaching one with fuel—an industry ticket, written knowledge, deployable cash—and a floor that caps the downside. The evidence is lopsided: in late 2008, 86% of financially constrained U.S. firms said they bypassed attractive investments, against 44% of unconstrained firms, with similar patterns in Europe and Asia (Campello, Graham and Harvey, 2010); of 4,700 public companies tracked through three recessions, 17% did not survive and only 9% came out stronger (Gulati, Nohria and Wohlgezogen, 2010); graduates and startups that meet a downturn carry the scar for years (Kahn, 2010; Sedláček and Sterk, 2017). Run through the series’ own model, a window has a price: for an operator without a ticket, it must lift growth to about 19% a year for the first five years—more than twice the 8% baseline—to flip the sign. A window amplifies fuel; it does not replace it.
1. The question Parts One and Two did not answer
Part One put the lottery, a salary and a startup on the same ruler and concluded that, for the average entrant, none of them is positive. Part Two found three switches — the ticket, the margin ladder, the growth rate — plus a floor, and showed that the sign of starting a company can be written by the person doing it.
Neither answered one question: when.
There are plenty of popular answers, and most are the same story: someone went all in when nobody else dared, and it worked. The problem with stories like that was covered in both earlier parts: the planes that did not come back never appear in the story (Part One); the people left at the end of a coin-flipping contest first have to prove it was not luck (Part Two).
This article asks a different way. How often does a window come? When it comes, who gets it? How good does a window have to be to be worth waiting for?
First, the scope: this article does not discuss stocks. There is already plenty of research on timing the stock market, and its conclusion is mostly: don’t. This article is about windows for work and for companies: when to change seats, when to expand, when to enter.
2. Windows are not rare: in Taiwan, one every four years on average
The National Development Council has dated 7 business-cycle troughs since 1998: December 1998, September 2001, February 2005, February 2009, January 2012, February 2016 and April 2023. The intervals between them are 33, 41, 48, 35, 49 and 86 months — 48.7 months on average, about four years. The most recent fell from a peak in January 2022 and took 15 months of contraction to reach bottom.
Figure 1: Taiwan’s business-cycle troughs as dated by the National Development Council, 1998–2023 — 7 troughs, intervals of 33, 41, 48, 35, 49 and 86 months, 48.7 months (about 4 years) on average. Source: NDC reference dates and peak/trough determinations; horizontal axis to scale by year and month.
What this means: in the ten years you spend building a ticket in the industry, you will meet two or three troughs on average; with a long cycle, perhaps only one — between 2016 and 2023 the two troughs were 86 months apart. A window is not once-in-a-century luck; it comes, and it comes again.
What is rare is something else: the person who, when the trough arrives, has fuel in hand and a floor underfoot.
3. The same window, two outcomes
In the fourth quarter of 2008, Murillo Campello, John Graham and Campbell Harvey surveyed chief financial officers across 39 countries in the U.S., Europe and Asia, collecting 1,050 responses. Among U.S. companies, 86% of financially constrained firms said they had bypassed attractive investments because they could not raise external funds; among unconstrained firms, the figure was 44%; results in Europe and Asia were similar. In the same crisis, constrained firms missed good investments at almost twice the rate of unconstrained ones.
Figure 2: The same crisis, the share of firms that bypassed attractive investments (U.S. companies) — financially constrained 86%, unconstrained 44%. Source: Campello, Graham, Harvey (2010); Europe and Asia similar.
Ranjay Gulati and colleagues at Harvard Business School followed 4,700 public companies through three recessions, in 1980, 1990 and 2000: 17% did not make it — bankrupt, acquired or taken private; only 9% came out stronger than they went in, beating competitors in sales and profit growth by at least 10%. Companies that only cut costs had about a 21% chance of pulling ahead afterwards; those that only expanded, about 26%; at the top was a group within those that did both, about 37%: they saved costs by improving operational efficiency, relying less on layoffs, while investing more in marketing, R&D and new equipment.
Figure 3: 4,700 public companies after three recessions (1980, 1990, 2000) — did not survive 17%, survived but did not pull ahead 74%, came out stronger 9%. Source: Gulati, Nohria, Wohlgezogen (2010), Harvard Business Review.
For people without fuel, a window can even do harm. The economist Lisa Kahn followed U.S. college graduates who graduated in recession years: for each percentage point higher unemployment in the year of graduation, starting pay was 6–7% lower, and still 2.5% lower fifteen years later. Taiwan’s unemployment rate in August 2009 was 6.13% — the same month was a headwind for new graduates without experience and a hiring window for companies with cash in hand.
Startups are the same. Petr Sedláček and Vincent Sterk analysed cohort data on U.S. startups and found that companies born in a recession include a lower share with high-growth potential, and the gap follows that cohort for many years. A trough does not automatically hand good opportunities to newcomers.
These studies point to the same thing: a window does not divide opportunity evenly; it amplifies what each person already holds. For those with fuel and a floor, it amplifies into expansion against the trend; for those without, into harm. This is the same structure as the allocation problem in Section 9 of Part One — those with tickets don’t enter, those who enter have no ticket; the window only makes that gap larger.
4. Fuel: what to store in ordinary times
Fuel has to be stored in ordinary times; start storing once the window opens and it is too late.
| Fuel | Individual | Company |
|---|---|---|
| The ticket | Industry knowledge built in a seat that touches decisions (Part One, Section 9) | The trust that makes customers take your call and try your samples |
| Written knowledge | Your own notes, work and cases | A knowledge base: engineering answers from it, a second location opens by following it (Part Two, Section 4) |
| Cash | Money you can commit beyond living costs | Money you can commit beyond operating working capital |
None of the three fuels is given by the window. The ticket takes ten years; knowledge is written one article at a time; cash is kept from the good years.
5. The floor: decide first how much you can afford to lose
The floor does not decide how much you can earn; it decides whether you can act when the window comes. In the Campello survey, financially constrained U.S. firms burned through about a fifth of their liquid assets in 2008; 70% said they sold more assets than usual to raise funds (37% for unconstrained firms) — they could see the window; they just could not stand firm.
An individual’s floor has three parts: living costs that can last a stretch with no income, skills you can take with you, and never signing a personal guarantee that reaches through to you.
A company’s floor also has three parts: cash beyond operating working capital, fixed costs you can hold down, and no cross-guarantees.
The floor has to be built in the good years. Starting to save cash at the trough is like starting to fix the roof when it rains.
6. Three kinds of window
| Window | When it opens | Who it favours | Signal |
|---|---|---|---|
| Business-cycle trough | A contraction period dated by the National Development Council | Companies with a cash floor that can hire and add equipment against the trend | The business-cycle monitoring indicator turns blue; peers cut back |
| Technology change | A new generation of specification appears and old capabilities no longer apply | People carrying an adjacent ticket who learn the new capability first | Customers start asking about specifications you have never made |
| Supply chain break | The existing supply network cannot keep up | Suppliers who can deliver verifiable samples quickly | Large manufacturers search online for unfamiliar suppliers |
Within technology change, only the kind that makes old capabilities obsolete opens the window to newcomers. The organisational scholars Michael Tushman and Philip Anderson studied three U.S. industries — cement, domestic airlines and minicomputers — from each one’s beginning through 1980: for the two breakthroughs that destroyed existing competence, 7 of the 11 companies that introduced them first were new entrants; for the six breakthroughs that enhanced existing competence, 35 of the 37 first adopters were incumbents. In other words, when old capabilities become obsolete, people carrying an adjacent ticket who are willing to learn something new first have a better chance than incumbents holding on to the old capability; while the old capability still works, the window mostly goes to the incumbents. Liquid cooling is one such technology change right now (see “Liquid-Cooled Servers Are Coming: What Should Production-Line MRO Change?”): customers start asking about specifications you have never made.
For a supply chain break, the signal is at the customer. Large manufacturers rarely go looking for suppliers they do not know; when they start searching online for unfamiliar names, it usually means the existing supply network can no longer keep up. This is a window mid-stream companies often meet and often miss.
7. TSMC in 2009: a textbook window in the financial crisis
In January 2009, TSMC told its investor conference that first-quarter revenue would fall from NT$64.56 billion in the previous quarter to NT$32–35 billion, roughly half; market analysts estimated capacity utilisation below 40%. In May, chairman Morris Chang stepped in to bring back employees who had been laid off earlier; in June he took on the CEO role again, and after revenue had halved he increased investment: capital expenditure went from US$2.671 billion in 2009 to US$5.936 billion in 2010, about 2.2 times. Chang later said that what he had seen was the potential of 28 nanometres in smartphones.
Figure 4: TSMC capital expenditure, 2009 vs 2010 — US$2.671 billion to US$5.936 billion, about 2.2 times. Source: TSMC Form 20-F (fiscal 2009 and 2010).
This example is often told as a story of courage. Put back into this article’s framework, it is really three things holding at once:
Fuel: process technology and customer trust.
Floor: a financial position that could withstand revenue halving and still have room to invest more.
Window: the year global semiconductor revenue was forecast to fall by 30%.
Take away any one, and the ending would not have been this.
TSMC is a giant, but the logic scales down. At the April 2023 trough, among your peers, who was hiring, who was adding equipment, who was cutting back?
8. The window threshold: calculated on the same ruler
Is a window worth waiting for? Calculate it once with the same model as Parts One and Two.
The sub-model in Section 10 of Part One: the owner of a surviving company starts at NT$1 million of annual income; at 8% annual growth the twenty-year net difference is −NT$11.02 million; at 12% it is +NT$0.39 million, and the sign flips positive. Converted to the owner’s total income over twenty years, the condition for flipping is about NT$71 million.
What a window can do is make the first few years grow faster. How much faster? Starting from NT$1 million in year 1, the window makes the following years grow faster, after which growth returns to 8% a year:
| Years of window acceleration | Growth needed each of those years | Each year after |
|---|---|---|
| 3 years | about 26% | 8% |
| 5 years | about 19% | 8% |
| 7 years | about 16% | 8% |
| 10 years | about 14% | 8% |
Figure 5: The window threshold — years of acceleration and the annual growth needed (year-1 owner income NT$1 million; back to 8% a year after the acceleration period; reference lines at 12%, the flip threshold without acceleration, and 8%, normal growth). Source: derived from the Section 10 sub-model of Part One (see the full verification table).
In other words, for someone without a ticket to flip the sign on a window, the first five years need nearly 19% growth a year, more than twice the normal 8%. At 15% a year for the first five years, the net difference is still about −NT$4.25 million; at 20% a year, it becomes about +NT$1.70 million.
And reality may be harder. Sedláček and Sterk found that startups born in a recession have a lower share with high-growth potential and stay smaller on average. If a company started in a window gives its owner only NT$800,000 in starting annual income, the first five years need to grow at about 24% a year.
A window cannot replace a ticket. Without a ticket, a window cannot change the sign; with a ticket — Part One’s route B is already +NT$5.60 million — the window is acceleration.
Waiting also has a cost. Part One said the network that works at forty is less than half as useful at fifty-five, which means the ticket depreciates by 4.5% or more a year. Taiwan’s troughs come on average every four years, so the practical rule is: once you have the ticket, wait for the next trough, not the one after it.
9. What to do between two windows
For most of a life, the window is not open. What to do in that time was already covered in the two earlier parts; here it is just put in order:
Add fuel: stay in a seat that touches decisions, and write down what you know (Part One, Section 9; Part Two, Section 4).
Build the floor: keep cash in the good years; do not sign guarantees that reach through to you personally.
Roll dice you can afford to lose: small sample runs, new items, new customers — every roll must be one you can afford to lose (Part Two, Section 5).
What not to do: do not go all in, do not borrow to add leverage, and do not use fuel you do not have to bet on someone else’s story.
10. What this means for mid-stream supply chain companies
Build the floor in the good years — saving cash at the trough is too late.
In the 2008 survey, 86% of financially constrained U.S. companies missed attractive investments. Cash kept from the good years is fuel you can use directly at the trough.
The window’s signal is at the customer — when large manufacturers look for unfamiliar suppliers, the window is open.
There are three conditions for catching it: engineering can answer, samples can be delivered, lead times can be kept. All three also have to be ready before the window opens.
When the window comes, hire first — good people can be hired only at the trough.
In the period when unemployment rose to 6.13% in August 2009, companies with a floor could hire people they normally could not. Equipment can still be bought next year; the people of that year, once missed, are missed.
Conclusion
Part One said all three roads are negative for the average entrant. Part Two said you write the sign. This article fills in the final piece: when to write it.
Windows are not rare; in Taiwan one comes on average every four years. What is rare is having fuel in hand and a floor underfoot when the window comes. The popular word for this is courage, but from the numbers, courage is only the last step — the ten years of preparation before it decide whether that step can be taken.
A window amplifies fuel, not courage.
Frequently asked questions
How often does Taiwan’s economy reach a business-cycle trough?
The National Development Council has dated 7 troughs since 1998, about one every 4 years on average; the most recent was April 2023.
Is a recession a good time to start a company?
It depends on whether you have fuel. U.S. research found that startups born in a recession have a lower share with high-growth potential. Calculated with this series’ model, someone without industry experience needs about 19% growth a year for the first five years to flip the sign.
Should a company cut back or expand in a downturn?
Harvard Business School followed 4,700 public companies through three recessions: companies that only cut costs had about a 21% chance of pulling ahead later; those that only expanded, about 26%; at the top, about 37%, was the group that did both and saved costs through operational efficiency rather than layoffs. To go on the offensive, you first need cash: in the 2008 CFO survey, 86% of financially constrained U.S. companies bypassed good investments.
Do you lose out by graduating or changing jobs in a downturn?
U.S. research shows that for each percentage point higher unemployment in the year of graduation, starting pay is 6–7% lower, and still 2.5% lower fifteen years later. A window is a headwind for people without experience, so it is steadier to build skills you can take with you before making the change.
What are “fuel, floor and window”?
Fuel is the industry ticket, written knowledge and cash you can commit; the floor is the range you can afford to lose; the window is a business-cycle trough, a technology change or a supply chain break. Entering makes sense only when all three hold at once.
Appendix: the data grade of this article
| Parameter | Grade | Source |
|---|---|---|
| Trough dates and intervals (48.7 months on average) | Hard | NDC business-cycle reference dates and peak/trough determinations |
| 86% vs 44% (U.S. companies; Europe and Asia similar), 39 countries, 1,050 CFOs, about a fifth of liquid assets burned, 70% sold more assets (37% for unconstrained firms) | Hard (survey) | Campello, Graham, Harvey (2010) |
| 4,700 companies, 17%, 9%, 21%, 26%, 37% (the subgroup that did both and saved costs through operational efficiency) | Hard (research) | Gulati, Nohria, Wohlgezogen (2010) |
| Survived but did not pull ahead 74% | Derived | 100% − 17% − 9% |
| Starting pay 6–7% lower, 2.5% lower after fifteen years | Hard (research) | Kahn (2010) |
| Recession-born startups have a lower share with high-growth potential and are smaller on average | Hard (research) | Sedláček, Sterk (2017) |
| 7 of 11, 35 of 37 | Hard (research) | Tushman, Anderson (1986) |
| TSMC first-quarter guidance NT$32–35 billion (previous quarter NT$64.56 billion) | Hard (company announcement) | TSMC results of 22 January 2009 (Form 6-K) |
| Global semiconductor revenue forecast to fall 30% | Hard (company investor conference) | Commercial Times, 10 February 2009, reporting TSMC’s investor conference |
| Capacity utilisation below 40% | Analyst estimate | Commercial Times, 10 February 2009, quoting market analysts |
| Laid-off employees brought back in May; CEO role resumed in June | Hard (autobiography, company annual report) | Business Today excerpt of Morris Chang: Autobiography; TSMC 20-F (effective 12 June 2009) |
| Capital expenditure US$2.671 → 5.936 billion | Hard (company annual report) | TSMC Form 20-F, fiscal 2009 and 2010 |
| Saw 28 nm and smartphones | Account of the person involved | Commercial Times, 9 December 2024 |
| Unemployment rate 6.13% in August 2009 | Hard | DGBAS (reported by Commercial Times) |
| Window thresholds (about 26%, 19%, 16%, 14%) | Model derivation | Section 10 sub-model of Part One of this series |
| Scenario with NT$800,000 starting annual income | Assumption | For illustration |
| Ticket depreciates 4.5% or more a year | Inferred | Converted from Part One’s “less than half by fifty-five” |
| Signals of a supply chain break | Practical observation | JIAHON shop-floor experience |
Road to NT$100 Million series: Part One, “The Odds and Costs of Three Roads” | Part Two, “You Write the Sign” | The Remedy, “The Window Opens Only for Those with a Floor” | Part Four, “Margin Is the Price of Trust”
References
[1] JIAHON Think Tank (2026), “Choosing a Road to NT$100 Million — Lottery, Salary, Startup: The Odds and Costs of Three Roads” (Part One). https://jiahon.com.tw/three-roads-to-100-million-part-one/
[2] JIAHON Think Tank (2026), “You Write the Sign — The Same Numbers, Read Upward” (Part Two). https://jiahon.com.tw/you-write-the-sign-part-two/
[3] National Development Council, “Reference Dates of Taiwan’s Business Cycles,” “Determination of the 12th Business-Cycle Trough,” “Determination of the 14th Business-Cycle Trough,” “Determination of the 15th Business-Cycle Trough.” https://www.ndc.gov.tw/
[4] Murillo Campello, John R. Graham, Campbell R. Harvey (2010), “The real effects of financial constraints: Evidence from a financial crisis,” Journal of Financial Economics 97(3): 470–487. https://doi.org/10.1016/j.jfineco.2010.02.009
[5] Ranjay Gulati, Nitin Nohria, Franz Wohlgezogen (2010), “Roaring Out of Recession,” Harvard Business Review, March 2010 (R1003C). https://hbr.org/2010/03/roaring-out-of-recession
[6] Lisa B. Kahn (2010), “The long-term labor market consequences of graduating from college in a bad economy,” Labour Economics 17(2): 303–316. https://doi.org/10.1016/j.labeco.2009.09.002
[7] Petr Sedláček, Vincent Sterk (2017), “The Growth Potential of Startups over the Business Cycle,” American Economic Review 107(10): 3182–3210. https://doi.org/10.1257/aer.20141280
[8] Michael L. Tushman, Philip Anderson (1986), “Technological Discontinuities and Organizational Environments,” Administrative Science Quarterly 31(3): 439–465. https://doi.org/10.2307/2392832
[9] TSMC (22 January 2009), “TSMC Reports Fourth Quarter EPS of NT$0.48,” U.S. SEC Form 6-K. https://www.sec.gov/Archives/edgar/data/1046179/000114554909000084/h03042e6vk.htm
[10] TSMC, Form 20-F for fiscal 2009. https://www.sec.gov/Archives/edgar/data/1046179/000095012310034985/h04138e20vf.htm; Form 20-F for fiscal 2010. https://www.sec.gov/Archives/edgar/data/1046179/000095012311035858/h05002e20vf.htm
[11] Commercial Times (10 February 2009), “First-quarter revenue may halve.” https://www.ctee.com.tw/news/20090210700013-439901; Commercial Times (12 June 2009), “Morris Chang retakes command of TSMC”; Commercial Times (9 December 2024), “Shakespeare’s ‘tide’ moment: the key to TSMC’s expanded investment in 2009.” (Chinese-language articles)
[12] Business Today (21 November 2024), “Why would TSMC need layoffs? Morris Chang on returning as CEO 15 years ago” (excerpt from Morris Chang: Autobiography; Chinese-language). https://www.businesstoday.com.tw/article/category/183015/post/202411200041/
[13] Commercial Times (23 September 2009), “August unemployment 6.13%, last among the Four Little Dragons.” (Chinese-language)
[14] JIAHON Group (technical article), “Liquid-Cooled Servers Are Coming: What Should Production-Line MRO Change?” https://jiahon.com.tw/liquid-cooling-production-line-mro/
Note
This is an opinion article. All figures are drawn from the public statistics, research and company filings listed, with their data grade stated; the window threshold is derived from the model in Part One of this series and involves no internal financial data of any specific company. This article does not constitute investment, entrepreneurship, tax or legal advice.
License
This work is published by JIAHON Group. All rights reserved. Quotation and redistribution are permitted with proper attribution and a link to jiahon.com.tw. Commercial use—including but not limited to paid courses, reports, or consulting services incorporating this content or its frameworks (including “fuel, floor, window” and “the window threshold”)—requires prior written authorization from JIAHON. For commercial licensing enquiries, please contact us through the website.