The Regulatory Instrument That Does Not Exist Yet

Plain-language explainer for doi:10.5281/zenodo.20701683 (#426)

The central idea in one sentence

The Basel framework regulates solvency (capital) and liquidity (LCR/NSFR) but has no instrument for a third, deeper category — systemic topology risk, the risk that funding-gap resolutions across institutions conflict at the network level, making the system irresolvable by any bilateral or multilateral netting — and the 2008 crisis was exactly this kind of failure.

What Basel actually measures

Basel III gives regulators three main instruments:

  • Capital requirements (CET1, Tier 1): can the institution pay its debts? — solvency risk
  • LCR / NSFR ratios: can it fund itself short-term? — liquidity risk
  • G-SIB buffer: is it systemically important? — a size-and-proxy measure

Lehman Brothers had an 11% Tier 1 capital ratio at end-2007. It appeared liquid in August 2008. Every Basel instrument said it was fine. What failed was the network — cascading collateral conflicts that no bilateral resolution could untangle. The failure was invisible to every Basel instrument because no regulator was computing the topology of the exposure network.

The three layers of financial risk

The paper identifies three risk categories, each requiring a deeper instrument than the last:

Layer 0 — Solvency risk (bilateral, H⁰): Can institution $i$ pay its debts? Measured by capital ratios. Basel III is broadly adequate here.

Layer 1 — Liquidity risk (triangular, H¹): Can the network fund itself? LCR and NSFR are institution-level instruments — they cannot detect network-level funding loops. Three banks each satisfying LCR individually can still form an irresolvable funding loop collectively. The number of independent funding loops is $\beta_1 = m - n + c$, where $m$ = bilateral exposures, $n$ = institutions, $c$ = connected components. This is computable in $O(n+m)$ time from existing regulatory data.

Layer 2 — Systemic topology risk (surface, H²): Can the network resolve itself? This is not addressed by any existing instrument. The relevant quantity is $\beta_2 = \dim(\ker B_2 / \operatorname{im} B_1)$ — the number of irresolvable conflict cycles in the three-party exposure network. The G-SIB buffer does not measure this; it gives no actionable restructuring guidance.

The marginal topology capital charge

The paper proposes a new regulatory instrument: a capital charge based on each institution’s marginal contribution to systemic irresolvability:

\[K^{\mathrm{top}}_i = \kappa \cdot \Delta\beta_2(i) \cdot \mathrm{EAD}_i\]

where $\Delta\beta_2(i) = \beta_2(\text{full network}) - \beta_2(\text{network without } i)$ is institution $i$’s marginal contribution to the count of irresolvable conflict cycles, and $\kappa$ is a regulatory calibration constant.

Three properties make this instrument attractive:

  1. Actionable: it tells each institution exactly which bilateral exposures to restructure, novate to a CCP, or reduce. An institution with $\Delta\beta_2(i) = 0$ pays zero regardless of size.
  2. Computable: $\beta_1$ in $O(n+m)$; $\beta_2$ in $O(n^3)$; marginal $\Delta\beta_2$ in $O(n^2)$ per exposure via rank-one matrix update. The data is largely already collected (EMIR, FR 2052a, AnaCredit).
  3. Incentive-compatible: CCP clearing, trade compression, and netting set restructuring all reduce the charge directly.

The critical threshold and Bagehot’s rule

The paper identifies a critical threshold

\[\beta^*(\rho) = \frac{3}{8}\ln\!\left(\frac{1}{1-\rho}\right)\]

where $\rho$ is the density of active bilateral exposures. This is the inverse temperature at which the exposure network transitions from self-resolving (below $\beta^$) to cascade-prone (above $\beta^$). The Countercyclical Capital Buffer should activate automatically when $\rho$ crosses $\rho^(\beta^) = 1 - e^{-8/3} \approx 0.931$.

Bagehot’s rule — “lend freely at a high rate against good collateral” — is the Layer 1 intervention: the penalty rate corresponds to $\beta^*(\rho)$, the critical temperature at which self-resolution is maximally incentivised while cascade is prevented. Mehrling’s extension (dealer of last resort) addresses the same Layer 1 problem for the collateral/shadow banking network.

Why centralisation is mathematically necessary

Systemic topology risk is a global property of the network. Computing $\beta_2 = \dim(\ker B_2 / \operatorname{im} B_1)$ requires both the bilateral exposure matrix $B_1$ and the three-party conflict matrix $B_2$ in full. No individual institution can see both. The network topology regulator must be centralised — not as a policy choice, but as a mathematical necessity: the Betti number is a global invariant that cannot be decomposed into bilateral components.

The 2008 crisis as a topology failure

The standard narrative attributes 2008 to excess leverage, lax underwriting, and regulatory arbitrage. These are correct but incomplete. The deeper failure was topological: the repo/collateral network had accumulated enough three-party conflict cycles ($\beta_2 > 0$) that when Lehman failed, no resolution authority could find a consistent assignment of collateral across the network. The system was irresolvable by construction — and no Basel instrument could have detected this, because none of them measure $\beta_2$.

The paper states this as a theorem: a network with $\beta_2 > 0$ cannot be resolved by any bilateral or multilateral netting, regardless of the solvency or liquidity position of individual institutions.

Companion papers

  • The Unhedgeability Theorem (#396) — the mathematical foundation: a financial risk is hedgeable iff its H¹ class is trivial; convexity, basis risk, and XVA as H¹ classes
  • Pacioli Homology (#291) — double-entry accounting as a gauge theory; the Pacioli identity as a conservation law; the mathematical structure underlying the three-layer architecture
  • XVA as Gauge Curvature (#299) — CVA/DVA/FVA as curvature of a connection on the Pacioli manifold

For the full technical treatment, see doi:10.5281/zenodo.20701683