Historical bubble analogues, the post-2008 deregulation chain, and market-manipulation enforcement

Built from research/macro-history-dereg-manipulation.json (web-verified: CNBC, DOJ/CFTC/SEC press, Fortune, Roosevelt Institute, The Hill, Marketplace) + the FDIC time series in macro-fdic.

Thesis. Every bubble cycle moves risk into a less-regulated venue and uses an accounting/financing device that books revenue or hides leverage until a discrete unwind. AI 2025–26 rhymes precisely with dotcom (vendor financing) and 2008 (off-balance-sheet leverage) — enabled by post-2008 deregulation and a documented market-manipulation record.

1. The analogues — same devices, larger scale

Era	Device	What happened	Modern parallel
Dotcom 1999–2002	Vendor financing	Lucent/Nortel/Cisco extended billions of credit to customers to buy the vendors' own gear, booking financed sales as revenue (Lucent ~$8B of commitments; loans soured as carriers failed in 2001). Nasdaq −78% peak-to-trough.	NVIDIA → OpenAI/CoreWeave/xAI/neocloud equity + take-or-pay backstops that fund customers' GPU purchases. Same structure, larger scale (NVIDIA self-funding ratio ~18–56%, `data/graph.json`).
GFC 2008	Off-balance-sheet leverage (SIVs, CDOs, ABCP)	Risk warehoused in vehicles invisible until 2007–08. FDIC failures spiked to 30 (2008), 148 (2009), 157 (2010). Glass-Steagall already repealed (1999).	AI datacenters via SPVs/JVs (Stargate, Meta-Hyperion/Blue Owl ~$27B) and private credit kept off the hyperscalers' balance sheets (`macro-cre-privatecredit`; First Brands the canary, `macro-firstbrands-ubs`).
2022–23 rate-shock + crypto	HTM accounting + uninsured-deposit runs + crypto leverage	Fed hiking created huge unrealized securities losses (FDIC peak −$517B, 2024Q1); SVB (~$209–220B), Signature (~$110B), First Republic (~$229B) failed; Silvergate wound down; FTX collapsed (Nov 2022). 2023 failed-bank assets ~$532B.	The HTM/AFS overhang persists (−$325B, 2026Q1; HTM −$214.5B not marked to AOCI — the SVB failure mode), re-widened by rising long rates from the AI-debt issuance wave.

All three devices are present simultaneously now — the novelty is scale and the speed/abstraction of the financing, not the mechanism.

2. The deregulation chain (1999 → 2018 → 2023)

1999 — Gramm-Leach-Bliley repeals Glass-Steagall → merges commercial + investment banking; sets up scale/concentration.
2010 — Dodd-Frank → enhanced prudential standards for banks >$50B (stress tests, liquidity, resolution plans).
2018 — EGRRCPA raises the enhanced-supervision/SIFI threshold $50B → $250B, exempting SVB and dozens of mid-size banks from regular stress testing and enhanced liquidity/resolution. SVB CEO Greg Becker had lobbied for this since 2015.
2023 — consequence: SVB (~$209–220B, below the $250B line) failed having escaped enhanced oversight — the direct dereg→failure causal chain. Critics (Warren/Porter, Roosevelt Institute) tie the rollback to the collapse. (Causal weight is contested; the threshold change is fact — cross-ref influence-congress-funding-compromise S.2155.)

2b. Which post-2008 dereg enabled which 2025-26 structure (added 2026-06-16, #77)

The chain above is bank-tailoring; the fuller map ties each specific deregulatory (or deliberate non-regulatory) action to the modern structure it enabled and the block that documents it. Actions/dates are fact (cross-checked); the "enabled" mapping is labeled interpretation. The dominant signature is deregulation by gap — declining to extend prudential/disclosure rules to nonbanks — not repeal.

Enabler	The action (date)	2025-26 structure	Block
Private-credit / NDFI gap	Dodd-Frank capital/leverage/stress regime applies to banks, not BDCs/direct lenders; no consolidated leverage limit; Fed only began reporting NDFI exposure (2026)	the ~$1.7T+ private-credit market + bank→NDFI→datacenter-SPV chain	`macro-cre-privatecredit`, `macro-private-credit-bdc-stress`
Retail access to alts	Aug 7 2025 EO on 401(k) alternatives + DOL rule; accredited-investor expansion (eff Dec 8 2020), never inflation-indexed	private credit reaching 401(k)/retail	`macro-private-credit-bdc-stress`
GENIUS Act	signed Jul 18 2025 (P.L. 119-27); 1:1 HQLA reserves, monthly disclosure	stablecoins as a captive short-Treasury buyer	`blockchain-leg`, `macro-us-fiscal-trap`, `macro-official-data-integrity`
Volcker 2.0	covered-funds amendments final Jul 31 2020 — VC/credit-fund exclusions	banks back inside VC/private-credit funds	`macro-cre-privatecredit`
SLR relief	2020 COVID Treasury/reserve exclusion (lapsed Mar 2021); eSLR cut final Nov 25 2025 (buffer = 50% of Method-1 surcharge)	dealer balance sheet for the ~$1T Treasury basis trade	`macro-treasury-basis-trade`
AOCI opt-out + HTM	Basel III US final rule (Jul 2013): non-advanced banks opt out of AFS-through-capital; HTM unmarked	the hidden HTM hole (−$325B 2026Q1) / SVB failure mode	`macro-bank-htm-marks`, `macro-fdic`
Basel III Endgame retreat	proposed Jul 27 2023 (~16–20% more capital) → rescinded + weaker re-proposal Mar 19 2026	thin big-bank buffers as exposure grows	`macro-bank-htm-marks`
ILC loophole	BHC-Act carve-out (CEBA 1987); Dodd-Frank moratorium lapsed; 23 charters, ~$247.7B	commercial/fintech firms owning FDIC banks outside Fed supervision	`macro-ilcs`
Family-office exemption	SEC Rule 202(a)(11)(G)-1 (Jun 2011), no leverage disqualifier; the Archegos fix Rule 10B-1 proposed Dec 2021, never finalized	hidden swap leverage (Archegos 2021, ~$10B+ losses)	`macro-family-offices`
EGRRCPA 2018	May 24 2018, SIFI $50B→$250B	the mid-size-bank tail outside enhanced oversight (SVB)	`macro-fdic`

Two signatures. (1) Deregulation by gap, not repeal — the heaviest items (private-credit carve-out, family-office exemption) are non-actions; the most relevant Archegos fix (Rule 10B-1) was proposed and abandoned. (2) A 2025-26 coordinated capital-relief wave — eSLR cut (Nov 2025), Basel III Endgame rescinded + re-proposed weaker (Mar 2026), the GENIUS Treasury-reserve rail — all loosen capital/leverage and expand Treasury intermediation at once, thinning buffers exactly as the AI-debt wave, the HTM hole, and private-credit exposure all grow. (A primary-document pass should lock the EO number and CFR rule numbers verbatim before any identifier is relied on; magnitudes ~$1.7T / ~$1T are approximate.)

3. The market-manipulation record (venues move, the pattern repeats)

JPMorgan precious-metals & Treasuries spoofing (2008–2016): hundreds of thousands of spoof orders in gold/silver/platinum/palladium/Treasury futures. Record $920.2M settlement (Sep 29 2020, DOJ+CFTC+SEC); traders Nowak & Smith convicted (2022). Venue: COMEX/CME, US Treasuries. Fact (adjudicated).
LME nickel / Tsingshan short squeeze (Mar 2022): Xiang Guangda's Tsingshan held a massive nickel short; a squeeze spiked prices and the LME suspended trading and cancelled ~$12B of trades (highly controversial; litigation followed). Separately, JPMorgan held LME nickel warrants that turned out to be "bags of rocks" (fake, 2023). Venue: London Metal Exchange. Fact.
SHFE / Chinese futures manipulation: recurring crackdowns; the LME/SHFE/COMEX arbitrage is a repeated stress point (see the 2025–26 COMEX-LME gold/copper dislocations in commodities-metals). Fact (pattern).

4. Synthesis

The AI bubble is not novel in mechanism, only in scale and in the speed/abstraction of its financing. Vendor financing (dotcom) + off-balance-sheet leverage (2008) + an HTM/AFS rate-shock overhang (2023) are all present simultaneously now, layered on a banking system deregulated at the margin (2018) and markets with a documented manipulation record (JPM/LME/SHFE). The formal cascade model (models/tla) treats these as the channels through which an AI-core shock propagates into the banking/credit system.

Sources: Dot-com bubble, FDIC Quarterly Banking Profile, EGRRCPA / SVB, Roosevelt Institute — 2018 rollback & SVB, CFTC — JPM $920M spoofing, DOJ — JPM traders sentenced, Fortune — LME "bag of rocks".

← Research index · structured data: macro-history-dereg-manipulation.json · macro-history-dereg-manipulation.md