AIG and the Geometry of Systemic Risk
The Functional Proximity Law is denied for the 2008 financial crisis — r = 0.042, p = 0.824 between derivatives counterparty and direct credit exposure. This is not a failure. It is the finding. The denial has a named mechanism: institutional regime mismatch.
The Result First
The Functional Proximity Law predicts that hub-importance scores should correlate across structurally similar layers.
For the 2008 financial crisis network — 16 institutions, 3 layers — the result between derivatives counterparty exposure (d1) and direct credit exposure (d2) is:
r = 0.042, p = 0.824
The law is denied. This is not a failure. This is the structural picture of 2008.
Why the Denial Has a Name
- The two layers encode different institutional operating models:
- Derivatives counterparty (d1): Who is on the other side of your OTC derivatives contracts? AIG, Goldman Sachs, Lehman Brothers, Deutsche Bank. These are trading-desk institutions. Their systemic risk is in the contracts they write.
- Direct credit exposure (d2): Who holds loans and mortgage-backed securities backed by your balance sheet? Fannie Mae, Freddie Mac, WaMu, Citigroup (commercial arm). These are lending institutions. Their systemic risk is in the loans they originate and hold.
A hub in the trading network and a hub in the lending network are not functionally proximate. The two layers measure structurally different risk channels. The law's prerequisite — both layers encoding the same relational regime — is not satisfied. r ≈ 0 is the correct result.
This is boundary condition #3 on the Functional Proximity Law. The other two are psychiatry (surface co-occurrence vs mechanistic cascade) and mathematics (formal containment vs proof usage). All three denials have the same root: resolution mismatch between the layers.
AIG: The One Exception
AIG is the exception that makes the rule visible.
AIG is #1 in derivatives (largest counterparty exposure across the trading network) and #2 in direct credit (second largest balance-sheet credit hub, after Fannie Mae).
It is the only institution that operated simultaneously in both structural regimes. AIG Financial Products wrote credit default swaps — derivatives instruments — that were exposed to mortgage credit risk. It sat at the intersection of the two layers.
When AIG failed, the contagion cascaded through both channels. The derivatives counterparty network froze because AIG's counterparties could not replace their CDS protection. The credit exposure network deteriorated because the underlying mortgages defaulted.
IRDME classifies AIG as a universal_hub: top-ranked in multiple independently defined layers. In 2008, that classification was the systemic risk label regulators needed.
Fannie Mae: The Hub Shadow
Fannie Mae is #13 in derivatives and #1 in direct credit.
Rank gap: 12 positions.
This is the largest hub shadow in the dataset. In IRDME terminology, a hub shadow is a node that is structurally central in one layer and structurally peripheral in another — meaning it is invisible to any single-layer risk analysis.
The Federal Reserve's standard post-LTCM practice was to monitor derivatives counterparty exposure. Fannie Mae was rank #13 in that network. A single-layer analysis would classify it as a mid-tier institution.
But Fannie Mae held or guaranteed approximately $5 trillion in mortgage assets. In the credit exposure layer, it is the largest hub. No derivatives-only risk framework could see it.
This is the structural argument for multi-layer systemic risk monitoring: not every institution's risk lives in the same layer.
The Numbers
| Institution | Derivatives rank | Credit rank | Gap | IRDME archetype | |---|---|---|---|---| | AIG | #1 | #2 | 1 | universal_hub | | Fannie Mae | #13 | #1 | 12 | hub_shadow | | Freddie Mac | #14 | #4 | 10 | hub_shadow | | WaMu | #16 | #7 | 9 | hub_shadow | | Goldman Sachs | #2 | #7 | 5 | chameleon | | Lehman | #3 | #8 | 5 | hub_shadow |
10 of 16 institutions have a rank gap ≥ 4. The network of 2008 counterparty relationships is structurally incoherent at the institution level. Each layer reveals a different risk picture.
h3 (≥ 5 institutions with rank gap ≥ 4) is CONFIRMED.
What the Denial Tells Us
The IRDME denial for finance is not a boundary of the method. It is a boundary of the domain. The 2008 financial system was structurally fragmented by design — derivatives and lending operated in separate regulatory regimes, separate business models, separate risk vocabularies.
The law works correctly. It finds correlation where the layers encode the same relational regime. When they do not, it reports no correlation and explains why. Both outputs are informative.
Full pre-registered case study: /case-studies/finance-2008