Joseph Alois Schumpeter (1883-1950), Austrian economist and from 1932
professor at Harvard University, provided the theoretical foundation for Kondratieff's long
waves. In his main work Business Cycles (1939, two volumes, 1,095 pages) Schumpeter
linked the Kitchin, Juglar and Kondratieff cycles into a three-layer model and placed
innovation at the center.
成長の原動力としてのイノベーション
Schumpeter's core contribution: economic growth arises not from capital
accumulation or additional labor alone, but from the discontinuous appearance of
new combinations — new products, new production methods, new
markets, new sources of supply, new organizational forms. The carrier of this innovation is the
entrepreneur (Schumpeter's famous entrepreneurial hero), who challenges
the status quo of established firms.
「創造的破壊」
The mechanism central to Schumpeter is called creative destruction
— coined in Capitalism, Socialism and Democracy (1942):
"The process of creative destruction is the essential fact about capitalism.
It consists of the incessant revolutionizing of the economic structure from within,
incessantly destroying the old one, incessantly creating a new one."
創造的破壊:交差点で市場支配力が旧技術から新技術へシフトする
Every wave of new technology destroys the economic base of the previous wave —
coal is replaced by oil, mechanical typewriters by PCs, PCs by smartphones,
traditional retail by e-commerce. This destruction is not a bug but a feature:
without it, productivity gains would stagnate.
コンドラチェフとの関係
Schumpeter saw each new K-wave as a cluster of innovations that reinforce
each other. The railroad wave (2nd Kondratieff) needed steel, telegraph, railway stations,
hotels, insurance — a whole innovation network. The IT wave needed
semiconductors, software, networks, server infrastructure, mobile, cloud — again a cluster. The next
wave (AI, biotech, quantum?) will need an analogous cluster: special chips (TPU/GPU),
datacenters, energy infrastructure, sensors, robotics.
The lesson for traders is clear: never blindly bet on market leaders. The history
of disruption shows that established champions are often not the winners of the next wave —
and supposedly "safe" blue-chip investments can become penny stocks within a decade if
their core business is technologized away.
シュンペーター対ダリオ — 2つの補完的な視点
Schumpeter and Dalio explain different dimensions of the same system and do not
contradict each other:
Schumpeter explainswhy long-run growth occurs
(innovation + creative destruction) — but not when any individual cycle
is accelerating or crashing.
Dalio explains in which debt and order phase a society and capital
system currently stands — but not which industries are disrupted by which tech wave.
Combined: When Schumpeter identifies a new tech wave (AI as possible 6th wave)
and Dalio shows a debt-reset phase (Phase 5/6), the investment implications
become especially complex: tech disruption creates new growth, but Dalio's debt
unwind can fundamentally shift the timeline and capital available for it.
📋 創造的破壊の見分け方
市場リーダーの利益率低下: A sector's gross margin falls
structurally over 5+ years (not cyclically), even as the industry grows —
a sign that new technology is shifting pricing power
ベンチャーキャピタルの流入: When >20% of global VC capital flows
into one technology (currently: AI/GPU clusters), an installation
phase with frenzy risk is typically beginning
大手企業の停滞: Classic market leaders (Kodak, Nokia, Sears)
typically show 5–8 years of price stagnation before the actual crash —
the market slowly prices in disruption
🎯 トレードへの示唆:ポートフォリオにおける破壊
勝者予測よりもピックス・アンド・ショベル: NVIDIA (GPUs), TSMC
(chips), ASML (EUV lithography) benefit from the AI installation phase
regardless of which AI company ultimately dominates
ディスラプション・リスクを体系的に確認する: For every buy-and-hold
investment ask: "What technology could replace the core business in 10 years?"
(banks → embedded finance, printers → digital workflows, broadcast TV → streaming)
破壊された業種をロングで保有しない: When a sector is technologically
displaced, dividends don't help — Kodak paid dividends until 2003
3.
批判:小さなサンプル、空想のリスク
🟦
理解する
ストラウス・ハウの第四の転換
While Kondratieff and Schumpeter identified technological long waves,
William Strauss (1947-2007) and Neil Howe (born 1951)
developed a generational theory. In Generations (1991) and above all in the book
The Fourth Turning (1997) they postulate a recurring cycle of four
generational phases over a saeculum of approximately 80-100 years — roughly
the lifespan of a human being.
4つの世代的アーキタイプ
Strauss-Howe identify four recurring personality types, each spanning roughly
20-25 years and appearing in a fixed sequence:
アーキタイプ
特性
形成される環境
預言者
理想主義的、道徳的、先見の明がある
ハイ期の幼少期、覚醒期の成人期
遊牧民
現実的、個人主義的、懐疑的
覚醒期の幼少期、崩壊期の成人期
英雄
集団主義的、楽観的、制度志向
崩壊期の幼少期、危機期の成人期
芸術家
繊細、妥協志向、専門性重視
危機期の幼少期、ハイ期の成人期
近代の世代的事例
Strauss-Howe classify US generations since 1900 as follows:
GI Generation (1901-1924) — Hero. Childhood in WW1/Great Crash, adulthood in
WW2, shaped post-war institutions.
Silent Generation (1925-1942) — Artist. Childhood in Depression/War,
adulthood in the conformist boom of the 1950s.
Baby Boomers (1943-1960) — Prophet. Childhood in the Eisenhower boom,
adulthood in Vietnam/hippie movement/neoliberalism.
Generation X (1961-1981) — Nomad. Childhood in the stagflation of the 1970s,
adulthood in the tech boom of the 1990s and after 9/11.
Millennials (1982-2004) — Hero. Childhood in "soccer mom" prosperity,
adulthood in financial crisis 2008, COVID, debt crisis.
Generation Z (2005-?) — Artist. Childhood in the crisis era 2008+, adulthood
in the expected reconstruction phase from ~2030.
一サイクルにおける4つの転換
From the generational cycle follow four societal "turnings" (each ~20-25 years):
1st Turning — High (e.g. 1946-1964). Stable institutions,
collectivist values, economic boom after Crisis.
2nd Turning — Awakening (e.g. 1964-1984). Individual self-fulfillment,
cultural revolutions, questioning of High-phase institutions.
4th Turning — Crisis (~2008-2030?). Existential threat to the
order, new institutions emerge from the ruins. Ends with a "Reconstruction".
2026年現在、私たちは第四の転換の真っ只中にあります, which according to
Strauss-Howe logic began with the subprime crisis of 2008 and should end around 2030-2032. The
crisis symptoms from their perspective: debt escalation, geopolitical confrontations
(US/China, Russia/Ukraine, Middle East), institutional trust deficit, polarization
of society, COVID shock 2020.
ダリオのビッグサイクルとの関係
Strauss-Howe and Ray Dalio (Module 5.5) arrive independently at very similar diagnoses:
both see the USA in a late Crisis/Phase-5 position. While Dalio explains the mechanism
monetarily-economically (debt maximum, reserve-currency transition), Strauss-Howe explain
it generationally-psychologically (Prophet Boomers meeting Hero Millennials, old institutions
replaced by new). The convergence is remarkable — even if both theories are
methodologically disputed (see Section 5.3.5).
第四の転換モデルの批判と限界
Sample size n ≈ 3: Strauss and Howe identify ~3 completed saecula
in American history since 1584. Drawing an 80–100-year periodicity from that is
statistically untenable.
Subjective generational boundaries: Where does the Boomer generation
end and Gen X begin? The dividing lines are arbitrary — depending on the source they
vary by 5–10 years, which shifts the entire turning assignment.
USA-only — and very specifically so: The model was developed
explicitly for Anglo-American history. Even within the West it breaks down:
Japan: Post-war Japanese society ran on a completely different
cyclical trajectory — a baby boom in the late 1940s (similar to the US), then
demographic implosion (birth rate < 1.3) and a Lost Decade from 1990. The expected
"Fourth Turning" collective rebuilding never came; instead: deflationary stagnation.
Strauss-Howe has no explanatory power here.
China: The One-Child Policy (1980–2015) created a radically
different generational cohort structure. "Hero Millennials" in the American sense
simply do not exist there — the 1980s cohort is a generation of only children with
different psychological conditioning.
Europe after 1945: Germany, France and Poland experienced World
War II completely differently. No pan-European saeculum pattern is detectable;
national cycles run asynchronously.
Not falsifiable: Every social development can be assigned post-hoc
to a turning. A "High" can be extended, a "Crisis" set earlier or later — no possible
event would definitively refute the model.
Political co-optation: Steve Bannon publicly cited the Fourth
Turning as a strategic framework. This is not an argument against the theory, but a
reminder of its narrative (not scientific) character.
Adam Tooze (Yale/Columbia) as academic critic: Economic historian
Adam Tooze (Crashed, 2018; Shutdown, 2021) explicitly argues against
generational determinism models: historical crises are shaped by specific political
decisions, institutions and contingency — not by an 80–100-year generational mechanism.
Tooze calls Fourth-Turning thinking "narrative comfort that explains away historical
contingency".
Anacyclosis as precursor: The idea that societies run through
predictable cycles of rise and decline is not new. The Greek historian Polybius
(c. 200–118 BC) described Anacyclosis in his Histories — a cycle of
Monarchy → Tyranny → Aristocracy → Oligarchy → Democracy → Ochlocracy → back to Monarchy.
The Strauss-Howe model is a modern variant of a 2,200-year-old idea. That makes it
neither more nor less correct — but it relativises its novelty.
4.
トレーダーへの実践的な応用
🟦
理解する
現在地:AIウェーブは第6コンドラチェフか?
The most discussed K-wave question of 2026: is a
6th Kondratieff wave beginning right now with artificial intelligence, biotech and
quantum computing as defining general-purpose technologies — or is the AI revolution merely the
final phase of the expiring 5th (IT) wave?
賛成論:AIは独立した新しい波
Proponents argue:
Generative AI since 2022 (ChatGPT, DALL-E, GPT-4, Claude, Gemini)
fundamentally changes how knowledge work operates — not just an efficiency gain
but a new production logic.
Robotics & autonomous driving (Boston Dynamics, Tesla, Waymo)
complement software AI with a physical dimension that was absent in the IT era.
Biotechnology & CRISPR (gene therapy, mRNA platforms, synthetic
biology) revolutionize medicine — comparable in scope to industrial chemistry of the 3rd wave.
Quantum computing (IBM Quantum, Google Sycamore, IonQ) promises
fundamentally new classes of computation impossible with classical hardware.
Renewable energy & storage (solar, wind, lithium iron phosphate
batteries, hydrogen) form the energy base infrastructure every new wave
requires — analogous to coal (1st wave) and oil (4th wave).
If this thesis is correct, the 6th wave would begin around 2030, reach its peak in the 2050s,
and transition into a maturity phase around 2080.
反対論:AIはITウェーブの一部に過ぎない
Skeptics argue:
AI builds on semiconductors, cloud, data networks — the core technologies of the
5th wave. Without this infrastructure AI would not be possible.
Generative AI is more of an application layer on existing tech stacks
than a genuinely new paradigm.
Productivity data so far shows no measurable GDP
acceleration from AI — the productivity surges of real K-waves were historically
immediately visible.
Marketed as a "6th wave", AI is above all a marketing narrative for
VCs, hyperscalers and ETF providers — not necessarily a scientifically proven step.
投資家にとっての意味は?
The investment implication depends heavily on which thesis one follows:
If the pro thesis is correct: Overweight in specialist ETFs on AI (BOTZ,
ARTY, AIQ), biotech (XBI, ARKG), robotics (ROBO), quantum (QTUM), clean energy
(ICLN, TAN). Hyperscalers (NVIDIA, MSFT, GOOG, META) as picks-and-shovels plays.
If the contra thesis is correct: Caution about an AI bubble — many valuations
already price in high expectations. Risk management through broad diversification,
not AI concentration.
Robust position: A moderate overweight of innovation themes
(5-15% of the equity allocation) makes sense in both scenarios — without committing to a
specific wave.
Carlota Perez (leading K-wave researcher) herself classifies AI as a
deployment phase of the 5th wave, not as a 6th wave — a cautious academic
position prevailing in her school. The definitive answer will only come from the next
20-30 years.
歴史的なテクノロジー導入フェーズ:パターンと類似点
Every Kondratieff installation phase has a typical structure: hype bubble → crash →
consolidation → deployment. The historical parallels to the AI phase 2020–2026 are striking:
Railway shares ×10 in 3 years; "everyone" bought railway shares
Similar IPO fever: Nvidia share price ×10 in 3 years, AI startups valued at billions without revenue
3rd Wave — Radio/Auto (USA)
1920–1929: Radio stocks (RCA ×100) + auto boom (GM ×30)
1929: Great Crash −89% (Dow Jones 1929–1932)
Radio as a "revolutionary medium"; margin trading on borrowed funds
Generative AI as a "revolutionary medium"; leveraged AI investments, options speculation on AI stocks
5th Wave — Dotcom (global)
1995–2000: Internet IPO boom, Nasdaq ×5 in 5 years
2000–2002: Nasdaq −78%, 4,000+ dotcoms insolvent
Pets.com, Boo.com: billion-dollar IPOs without business models; "revenue is irrelevant"
Numerous LLM startups with multi-billion valuations and no clear monetisation path
6th Wave? — AI/Biotech
2020–?: Nvidia ×30 in 4 years, AI VC investment ×10 since 2020
未確定
"AI solves everything" — valuations based on revenues 10+ years in the future
これが現在のフェーズです
ミンスキーの視点:AIバブルはどの程度成熟しているか?
Hyman Minsky's Financial Instability Hypothesis (Module 5.4) provides a complementary
diagnosis: in the Ponzi phase, assets are held solely on the expectation of further
price gains — no longer based on cash flow or dividends. The AI sector in 2026 shows
Ponzi-like characteristics: Nvidia valuations at 35× revenue, AI startups valued in the
billions without revenue, and hyperscaler capex plans justified only by permanently rising
AI adoption. This does not automatically mean a crash — but it does mean: the margin
of safety in AI stocks is thin.
📋 AIバブルのフェーズを見極める
Installation frenzy: Nvidia P/E >50×, AI IPOs without
revenue at billion-dollar valuations, mainstream media coverage of AI exceeds
all other tech topics combined
Turning point signal: Nvidia/SMCI/AMD correction >35%
without fundamental change, first major AI startups fail publicly,
hyperscalers reduce capex guidance
Deployment phase: AI productivity gains measurable in GDP
statistics, broad diffusion into non-tech sectors (healthcare, logistics,
agriculture), tech valuations normalise to historical ranges
🎯 トレードへの示唆:AIウェーブのポジショニング
Installation phase (now): Picks-and-shovels (GPU chips,
energy infrastructure, cooling systems) over end-player bets; build tail
hedge (long put on ARKK or QQQ, 1–3% of portfolio)
Post-crash deployment: Rotate into AI beneficiaries in the
old economy: industrial automation (ROBO), healthcare AI (ARKG), energy
efficiency (ICLN) — these benefit when the tech infrastructure is in place
Independent of wave thesis: NVIDIA, TSMC, ASML as mandatory
tech exposure elements — they win in both scenarios (5th wave end
or 6th wave start)
5.
批判:周期性と疑似科学
🟧
評価する
批判:周期性と疑似科学
Long waves are the most speculative category of cycle theories. They have
significant weaknesses that every trader should know before making an application decision.
長期波動の賛成論
Historically comprehensible pattern: tech clusters genuinely characterize decades
(railways, automobiles, IT) — the qualitative finding is robust.
Innovation as an economic driver is broadly recognized, well beyond the K-wave
(Solow model, endogenous growth theory).
Provide a strategic orientation for very long investment
horizons (20〜50年): which sectors will grow structurally, which will be disrupted?
The Strauss-Howe generational model explains political and cultural turning points
(e.g. 1968, 1989, 2008) more consistently than purely economic models.
反対論・問題点
No firmly measurable periodicity: K-waves vary
between 40 and 70 years — the postulated "50-60 years" is constructed ex post,
not predictable ex ante.
Tech definition vague: What was THE defining technology of the
3rd wave — electricity, the internal combustion engine, or chemistry? The choice
determines the temporal boundaries.
Cherry-picking risk: With only 5 historical waves, almost
any theory can be "confirmed" through clever date-setting.
Strauss-Howe is partly criticized as pseudoscience
(Krugman, The Atlantic 2017): generational stereotypes are exaggerated, the
model has been retrospectively adjusted to later developments.
Sample problem: Over 240 years of industrial history there are
only ~5 K-waves and ~3 saecula — not statistically testable in a significant way.
長期波動の方法論的限界
Survivorship bias in tech selection: The "defining technologies"
are selected retrospectively from a success perspective. In 1850 nobody would have
known for certain whether railway or telegraph would be the central technology —
today we "know" it.
Political shocks dominate: The two world wars, the Great
Depression, the Cold War, COVID — these events shaped economic trajectories
more strongly than long-wave tech cycles. They are not derivable from Kondratieff mechanics.
Globalization changes the model: Kondratieff referred to
national economies (UK, France, US). Today tech waves run synchronously
globally — the original mechanism (local capex boom-bust) applies only in a
limited way.
Central bank policy (Greenspan Put, ZIRP, QE) smooths economic
fluctuations considerably. This makes the identification of long waves in
current data more difficult.
Consensus view of serious economic history 2026: Long waves are useful as
a strategic explanatory framework for tech development and generational change,
but unsuitable as a timing tool for investment decisions. Nobody can
seriously predict whether the 6th wave starts in 2028, 2032 or 2040 — or which technology
will define it.
代替理論家 — 長期波動に反対するのは誰か?
Long-wave theories have prominent academic critics who argue from economic-historical
research, not from a book to market:
Barry Eichengreen (Berkeley, Hall of Mirrors, 2015):
Economic historian comparing the Great Depression and the 2008 financial crisis.
Eichengreen shows that the decisive variables — central bank responses, international
cooperation, political capacity to act — were not determined by K-waves or generational
cycles, but by contingent decisions of specific individuals. "History doesn't rhyme
as regularly as long-wave theorists suggest."
Adam Tooze (Columbia, Crashed, 2018): Sees the 2008 financial
crisis as a product of specific transatlantic balance-sheet structures of the 2000s —
not as the deterministic culmination of an 80-year generational cycle.
Robert Gordon (Northwestern, The Rise and Fall of American Growth,
2016): Argues that the first Industrial Revolution (1870–1970) produced a one-time
productivity explosion that will not repeat. Long waves with constant periodicity ignore
this fundamental asymmetry between waves.
Mainstream economics: K-waves and Strauss-Howe are rarely cited in the
American Economic Review, Journal of Finance and related journals — not due to censorship,
but because the empirical data base (n < 6 complete cycles) does not allow
statistically sound claims about periodicity.
レンズの選択:どの理論がどの問いに答えるか?
The six cycle theories of the Cycles Group are not competitors — they illuminate different
timescales and causal mechanisms. This table helps select the right "lens" for specific
investment questions:
理論
最適投資期間
最も適した問い
不適切な用途
実証的根拠
Kitchin (3-5 yrs)
1〜5年
在庫サイクル、商品供給サイクル
世代交代、テクノロジー破壊
良好(測定可能な在庫データ)
Juglar/Kuznets (7-25 yrs)
3〜15年
設備投資サイクル、不動産サイクル
技術的突破
良好(設備投資統計)
Kondratieff/Schumpeter (50-60 yrs)
20〜50年
長期的に成長するテクノロジー・セクター
年間タイミング、金利予測
弱(n=5データポイント)
Strauss-Howe (80-100 yrs)
30〜80年
社会的気分トレンド(米国のみ)
国際市場、短期
非常に弱(n=3、反証不可能)
Minsky FIH (variable)
1〜10年
信用バブルの検出、システミック・リスク
テクノロジー破壊、世代交代
中程度(事前の実証的予測が困難)
Dalio Big Cycle (~250 yrs)
10〜50年
基軸通貨リスク、地政学的フェーズ
短期タイミング、小規模経済
Weak (n<5 complete cycles)
The "empirical basis" assessment follows academic mainstream consensus 2026.
All theories with a weak empirical basis can still be useful as a strategic orientation
framework — but not as a precise timing tool.
6.
ポートフォリオへの応用
🟩
応用する
ポートフォリオへの応用
Even without K-wave belief, a moderate innovation allocation
makes portfolio-strategic sense — because regardless of the periodicity question it is clear:
technological disruption takes place and creates most returns in the long run. The
following five-point recommendation works both for K-wave proponents and opponents.
どの期間にどのレンズを使うか?
The right cycle theory depends on the question you are answering — and your
personal investment horizon. As a rule of thumb:
投資期間
主要理論
具体的な応用
使用しない場合
<2 years (short-term)
ミンスキーのフェーズ+センチメント
Is the credit bubble in the Ponzi phase? (→ Module 5.4) Sentiment extremes? (→ 5.6)
5-15% of the equity allocation in innovation themes. Specifically: AI
(BOTZ, AIQ, ARTY), biotech (XBI, ARKG), robotics (ROBO), quantum (QTUM), clean
energy (ICLN). Maximum 3% per individual ETF to limit single-theme risk.
Picks-and-shovels rather than end players. NVIDIA, TSMC, Microsoft, ASML
benefit certainly from the AI wave because they provide the infrastructure — regardless
of which AI application ultimately prevails.
Check disruption risk in the core business. Before every buy-and-hold investment
in blue chips ask: "What technology could make this business model obsolete within 10 years?"
Examples: traditional banks by embedded finance,
insurers by InsurTech, energy utilities by self-generation boom.
Use demographic trends as long-term anchors. Strauss-Howe's
generational logic provides robust hints: health/care sector benefits from
Boomer ageing; ESG themes are strongly demanded by Hero Millennials/Artist Gen Z;
old mass-consumer brands (Coca-Cola, Kellogg's) are losing relevance with
younger cohorts.
Never "all-in" on one wave. The 1999 tech bubble and the
2021 ARK boom show what happens when investors bet too heavily on a
supposed wave — drawdowns of 70-90% are realistic. Diversification across
sectors, regions and asset classes remains the most important protective principle.