11 min read4 viewsinvesting

War on the Rocks: The GDP Line Went Up, and About Ten Billion of It Came From a Change in Counting

An underground city-hall car park converted into an emergency operations centre at dawn, a long table in the foreground holding a stack of evacuation maps and a radio set not yet plugged in, half a cup of cold coffee weighing down the paper, rows of empty folding chairs receding into the dim colonnade behind

Notes on the War on the Rocks episode of 16 November 2026, where Ryan Evans talks with Amanda Hsiao of RAND, Michael Hunzeker of George Mason University and Peter Mattis of the Jamestown Foundation about Taiwan's 28 November local elections and its defence preparations: defence spending went from roughly 2.4% to roughly 3.3% of GDP in a single year, about $10 billion of that from folding in military pensions and the Coast Guard; roughly $29 billion of approved US arms has yet to be delivered, and an $11 billion package is frozen as a bargaining chip. The guests' shared judgement is that the percentage measures nothing about effectiveness — what counts is plans tested under stress. Every number is as stated at the time of recording. Educational listening notes only — no investment advice, no price targets, and the judgements here may be wrong.

  • defence budget
  • podcast-notes
  • Taiwan Strait
  • societal resilience
  • inputs vs outcomes
Contents
  1. For years I watched only that line as a share of GDP
  2. About ten billion dollars came out of the arithmetic
  3. What I watch now, and where I still can’t see
  4. Further reading
  5. The one thing worth taking with you

An underground city-hall car park converted into an emergency operations centre at dawn, a long table in the foreground holding a stack of evacuation maps and a radio set not yet plugged in, half a cup of cold coffee weighing down the paper, rows of empty folding chairs receding into the dim colonnade behind

War is nothing but the continuation of political intercourse, with the admixture of other means.

—— Clausewitz, On War, Book VIII (1832; trans. J. J. Graham, 1873)

In the War on the Rocks episode of 16 November 2026, Ryan Evans talks with Amanda Hsiao of RAND, Michael Hunzeker of George Mason University and Peter Mattis of the Jamestown Foundation about Taiwan’s 28 November local elections and its defence preparations. The hardest set of numbers is this: Taiwan’s defence spending went from roughly 2.4% to roughly 3.3% of GDP in a single year, and about $10 billion of that came from a change in what gets counted — military pensions, the Coast Guard, items of that sort. Meanwhile roughly $29 billion of arms the US has already approved for Taiwan has yet to be delivered, and an $11 billion package was frozen as a bargaining chip after Trump’s meeting with Xi. All three guests judge that these percentages and dollar figures measure nothing about fighting power — what counts is what has been exercised under stress — and every number here is as stated at the time of recording.

For years I watched only that line as a share of GDP

Before this episode, my read on Taiwan’s preparedness rested on two things: what share of GDP went to defence, and what the US had agreed to sell. A rising share was good news; a longer arms list was one more notch of safety. I never examined the assumption underneath — that money spent becomes capability, and that spending it is administrative work, budgeted and executed on schedule.

My second old belief was shallower. What Taiwan has been building these past few years goes by “whole-of-society resilience”, which in plain terms means civilians, local government and civic groups all able to hold up and help themselves through a disaster or a war. I took that for the floor of Taiwan’s political consensus: people disagree about talking to Beijing or about unification, but surely “what if it actually happens” puts everyone on the same side. I treated the whole thing as a progress bar — moving means advancing.

About ten billion dollars came out of the arithmetic

Start with the passage that made me sit up. Between 2025 and 2026 Taiwan changed what its defence budget counts, folding in military pensions, the Coast Guard and similar lines. The reasoning holds: Washington demands higher Taiwanese defence spending while measuring itself and its NATO partners with a broader yardstick, so Taipei asking for the same yardstick is fair. The result was a jump from about 2.4% to about 3.3% in one year, roughly $10 billion of it from the new arithmetic. One guest added a line I wrote down: retirement and pension benefits do not make a military more effective.

Two bars compared side by side, the right one taller, with its upper segment boxed in a dashed outline and labeled "about $10B from a changed counting scope," showing that part of the rise comes from redefining what counts rather than from new spending.

From there the whole episode picks at the same fault: we fixate on inputs and use them as a substitute for outcomes. Deterrence is settled by outcomes, and outcomes are what Beijing assesses too.

The inputs really are piling up. Every ministry has been told to rewrite emergency plans originally written for natural disasters into versions aimed at a Chinese invasion or high-end coercion. Every agency is inventorying how much of each critical good it holds and how much it needs. Taichung has exercised what to do after Chinese missile strikes on its LNG facilities. Hospitals are being hardened and equipment moved into basements. The parts of the supply chain that ran through China are being identified, and Taiwan has started producing its own blood supply. What one guest said she cannot assess is what all of it adds up to — plans on paper and numbers in tables do not tell you whether Taiwan is ready. Her fix: test those plans repeatedly under increasingly realistic and complex conditions.

Then there is the political layer, which I used to treat as noise. The episode is blunt about it: when the president’s party does not control the legislature, getting a policy outcome takes politics, and the process gets ugly. One guest noted that critical defence programmes — backup satellite communications among them — have been deliberately targeted for zeroing out. Worse is the thinness of expertise: the research resources and staff available to members of the US congressional armed services committees far exceed anything Taiwan’s legislature can offer, Taiwan has a handful of real experts, and one of them does not want another term. The guests described having to explain in person that Taiwan had bought the missile systems, but without sensing it could not target them — and that this is the level at which the people voting on the budget have to be walked through it.

On the left, five stacked blocks labeled budget, plans and supplies; on the right, a single empty dashed box labeled "results drilled under stress," with the arrow between them marked as having no numbers.

The part on resilience becoming a party badge is the part I liked least. Grassroots work is lively — the episode mentions Forward Alliance, the Kuma Academy founded by Puma Shen, and groups that never make the news in America — and the top has moved too, with the Lai administration’s whole-of-society resilience committee prying open the stovepipes between ministries and the private sector so agencies that never spoke to each other now talk. But to show up at one of those trainings, a guest said, is nearly to declare which end of the political spectrum you stand on: overwhelming support among DPP-leaning voters, and among KMT-leaning voters some concern that these activities antagonise Beijing. The host’s reply: that kind of division over defending your own country is not a good sign. Which makes Puma Shen’s campaign interesting: the 9-in-1 fills more than ten thousand local seats, and in Taipei he did not run on resilience at all, going on blue-leaning media to talk about tree canopies, transport and childcare, to the surprise of both parties. Yet protection of critical infrastructure, power plant certification and whether coal stays in reserve land on mayors and county chiefs.

There is a disconnect I had not considered. The MND has shown no institutional interest in playing with the grassroots organisations. Effective deterrence has to begin and end with the military, one guest argued, and yet what is being discussed is municipalities deciding where air defence assets go — the sentence alone shows the degree of disconnect. At the other end, new resilience structures are being layered on top of existing civil defence laws and institutions, and how the old is to integrate with the new nobody can say.

A spectrum running from the Green camp on one end to the Blue camp on the other, with the marker for "attends resilience training" crowded against the left end where dots are dense, while the right end is sparse.

The last third turns the same yardstick on China, which I thought was the most honest stretch of the hour. Last month China ran a naval exercise in the waters between Taiwan and Japan and between Taiwan and the Philippines, some three dozen vessels including an aircraft carrier. The PLA clearly wants to signal capability, a guest said; but if we scrutinise Taiwan’s military for how much is substance and how much is show, the question has to be flipped. His team looked at the big named exercises through commercial open-source imagery and saw plenty of surface combatants moving and perhaps some replenishment, yet in the ports themselves no attempt to exercise a large logistics build-up — ships leaving quickly and coming back quickly. Whether those operations can be sustained and replenished in a degraded environment, he said, the jury is out. Another guest added the balancing point: nearly all the PLA shortcomings we talk about are ones the PLA itself identified, and it is exercising and testing against them, so what we are looking at is a learning organisation working through the faults it found in itself.

Finally, the eight people. There were reports of infiltrators from the mainland landing in Taiwan and being arrested, and the guests’ assessment is: we don’t know. How many were simply escaping China, how many were eighteen-year-olds in uniform seeing what they could get away with, is unclear; most surrendered and did not resist arrest. One guest offered a rewording I enjoyed: another way of saying “military-age male” is “unemployed factory worker” — and those people are not doing well in the PRC these days. He looked at photographs of the boats and said a hundred-plus miles of open water is not something you would attempt in those, on the best day. The thing to chase is the intelligence and awareness end: how fast such things get picked up, and whether the people receiving the warning understand what they are looking at.

What I watch now, and where I still can’t see

If a headline saying defence spending reached 3.3% of GDP also made you exhale, the episode offers three questions instead: when does the thing that money bought actually arrive, has it been tested in a degraded environment once it did, and did the problems found get into the next version of the plan. On the roughly $29 billion of approved-but-undelivered arms, a guest was careful: a backlog is not the same as overdue, and much of this takes years to build under the best conditions. Missiles approved for sale around 2017 are arriving now; of the 66 F-16s sold, the first two landed within the past few weeks. The longest lags fall on the most expensive, genuinely symmetric systems — fighters and ships, the kit built to slug it out with an opponent head-on, which almost everyone in Washington has spent years urging Taiwan to stop buying.

The upper half shows a long arrow running from 2017 to the present for missiles that took nine years to arrive; the lower half shows sixty-six small squares with only two filled in, marking F-16 delivery progress.

One pattern I had failed to read for years got an explanation here. One guest pointed out that the KMT criticises slow American deliveries, yet the items it approved last spring out of the separate defence budget — the one appropriated outside the annual budget — were precisely the US-dependent ones, while the parts resting on joint production in Taiwan and direct commercial sales did not pass. Another guest’s version: good-faith oversight concerns among citizens are being leveraged for bad-faith political purposes. Waste, misallocation and sticky fingers in defence procurement are real worries; he even said that if a US administration proposed a bill of comparable size and length, he too would have a pile of questions. Both then turned it around — nobody’s hands are clean on dragging their feet over defence reform. One used the word sabotage about figures in the KMT including the supposedly pro-US ones, and said the DPP had plenty of chances over the last decade to choose differently, with conscription and terms of service among the changes that came very slowly.

The other signal I would have misread is “stability”. This round of cooling between Washington and Beijing looks like a small détente, but as a guest put it, it is a very unstable stability — built on narrow understandings about rare earth exports and licensing, while Beijing acts outside them: restricting sulfuric acid exports used in mining, confiscating the passports of mining and processing specialists so they cannot work for foreign firms, restricting construction equipment used to build processing plants. We may have an agreement on one small piece, he judged, while provoking systemic responses as different parts of their system kick into gear. The rationale for stability is that both sides need time — Taiwan to reform, the US to transform — but another guest doubts the interregnum has been used well, and named the cost: to push Taiwan into painful changes, America has to show it will be there, or an asymmetric posture built on resilience — avoiding the head-on fight, wearing an opponent down with cheap and numerous weapons — is a fool’s errand. On the frozen $11 billion, his larger concern is that foreign military sales have risen to the level of a presidential bargaining chip at all; the dollar figure is secondary.

So where I land. Taiwan’s trajectory is positive — all three guests agree, and so do I. What I cannot see overlaps with what they say they cannot see: whether plans have been run under realistic conditions, whether anyone owns the seam between the military and civil society, how much the legislature can actually process. None of that has a public number, and every indicator I can reach — percentages, procurement totals, exercise press releases — sits on the input side. I am willing to call this a question I currently measure badly, which leaves me nothing to say about whether it is going badly.

A wide plank labeled stability balances on a very narrow fulcrum labeled the rare earth deal, while several arrows beside the plank push up from outside that deal.

Which brings it back to those of us watching markets. You have probably had this experience: a number improves, the coverage changes tone, and you nudge your position or your mood a notch that way. The habit this episode gave me is to ask first how much of the improvement came from the thing being measured getting better, and how much from a change in how it is measured. For Taiwan’s defence share, about $10 billion of the jump is the latter. Transplant the question to a filing and it reads the same: how much of revenue growth came from an acquisition folded in, how much of the active user gain came from moving the definition from thirty days to ninety. My own practice is not clever: when I see an improved number, I go check whether its definition moved in the same period. If it did, the number goes back in the “don’t know yet” column.

Further reading

  • The War on the Rocks episode of 16 November 2026, the source for these notes
  • Taiwan Security Monitor, the George Mason University research group doing news aggregation, data analysis and map readings of PLA exercises
  • Public documents from the Lai administration’s whole-of-society resilience committee
  • The progress of Taiwan’s 2027 defence budget through legislative committee
  • Ian Easton’s reporting on mainland arrivals detained in Taiwan

The one thing worth taking with you

Inputs are not outcomes. An inventory counted, a budget appropriated, a plan written — each tells you how much you put in. Whether it works is something you learn by running it under the conditions where it breaks. I used to treat those as one thing; this episode separated them.

Here is something I tried, offered rather than assigned: take one thing you have been preparing for a while and have assumed is fine — the first aid kit, the backup of your important files, the blackout plan you agreed with your family — and test it today with one condition removed. Phone dead: can you still find that number? That person away: does the process still finish? Network down: does the backup come back? What this test most often turns up is something like the recovery password for the backup sitting on the machine the backup was meant to restore.

The top row shows four interlocking chain links standing for the hand-off working on a normal day, while the bottom row shows the same chain missing a link and broken open at a gap labeled phone dead, showing that only removing one condition reveals the break.

This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.

Comments

Loading comments…

Sign in with Google before posting. Only your name and profile picture are shown.