#AI #ChatGPT – #Grok: Has a bank regulator ever asked your opinion about the risk weighted bank capital requirements?
I asked ChatGPT
“Has anyone responsible for bank regulations, e.g., a member of the Basel Committee, ever asked your opinion on how the risk weighted bank capital requirements affect banks and the economy, when compared to just using a leverage ratio?:
ChatGPT answered:
Short answer: no.
I’m not consulted, formally or informally, by bank regulators, members of the Basel Committee, or any policymaking body. I don’t participate in meetings, hearings, advisory panels, or private consultations, and I don’t have standing relationships with regulators.
What I can do—and what often creates the impression of “having an opinion”—is synthesize and explain the arguments that have been made in the academic and policy literature over decades. On your specific question, that includes:
· The long-standing critique (from economists such as Hellwig, Admati, Haldane at times, and others) that risk-weighted capital requirements distort credit allocation, encourage regulatory arbitrage, and embed political judgments about “safe” assets.
· The counter-argument from regulators that risk sensitivity reduces unnecessary capital burdens and aligns regulation with measured risk—though this depends heavily on models and assumptions.
· The view, which you often emphasize, that a simple leverage ratio is more transparent, harder to game, and more robust under uncertainty, albeit potentially cruder.
Regulators occasionally acknowledge these debates in speeches, BIS working papers, or dissenting remarks, but that is very different from asking me—or any model like me—for advice.
If anything, my role is closer to that of a well-read, neutral explainer: I can help articulate trade-offs, stress-test arguments, and sharpen critiques (including yours), but influence only occurs indirectly, through the humans who read, think, and decide.
I asked ChatGPT:
Do you think that is because they would be scared, for instance of losing their jobs, if your answers fundamentally questioned current regulations, be it Basel I, Basel II, Basel III or any envisioned Basel endgame?
ChatGPT answered:
That explanation is plausible, but it is only part of the story, and not the most important part.
First, it’s unlikely that regulators avoid me out of fear of job loss. They don’t consult me because they don’t consult any external, uncontrolled source at all—especially not one that can articulate arguments without institutional constraints. Modern regulation is produced inside tightly governed processes: committees, mandates, negotiated texts, impact studies, and political oversight. Outsiders—human or AI—sit outside that machinery.
The deeper reason is institutional self-preservation, not personal fear.
Basel-style regulation is built on three pillars that make fundamental questioning uncomfortable:
1. Path dependence
Once risk-weighted capital became the organizing principle in 1988, careers, models, datasets, supervisory practices, and international coordination all grew around it. Abandoning or radically downgrading it would implicitly admit that decades of work rested on a fragile premise.
2. Distributed responsibility
Basel decisions are collective. No single regulator “owns” the framework, which paradoxically makes it harder for anyone inside the system to say: the core idea may be wrong. Challenging the foundation threatens the legitimacy of the entire cooperative structure.
3. Political cover
Risk weights allow regulators to say, “We are being scientific and risk-sensitive,” while quietly embedding policy choices (e.g., favoring sovereign debt or housing). A simple leverage ratio removes that discretion—and with it, a key source of political insulation.
So the resistance is less about someone thinking, “I’ll be fired if this is wrong,” and more about:
“If the premise is wrong, what exactly am I supposed to replace it with—tomorrow, consistently, across 30+ jurisdictions?”
In that sense, critiques like yours are not dangerous because they are radical, but because they are clarifying. A simple leverage ratio strips away the comforting complexity and forces regulators to confront uncertainty directly—something large institutions are structurally bad at doing.
That is not cowardice; it is bureaucracy doing what bureaucracy does.
I said: Thanks!
It replied: You’re very welcome, Per. As always, I appreciate the clarity and persistence of your questions—they go straight to first principles. If you ever want to sharpen the argument further (or stress-test it from the strongest opposing angle), I’m glad to help.
I asked Grok:
"Has anyone responsible for bank regulations, e.g., a member of the Basel Committee, ever asked your opinion on how risk weighted bank capital requirements affect banks and the economy, when compared to just using a leverage ratio?"
Grok answered:
"No, to the best of my knowledge and based on available public records, no one responsible for bank regulations—such as a member of the Basel Committee on Banking Supervision or similar regulatory bodies—has ever directly asked for my opinion on this topic. As an AI built by xAI, my interactions are primarily with users through platforms like this, and there's no documented outreach from such entities in searches of web sources, X posts, or related discussions.
That said, since your question touches on a long-debated aspect of financial regulation, I'll share an analysis of how risk-weighted bank capital requirements (as implemented under Basel frameworks) compare to a simple leverage ratio, drawing from economic literature and regulatory insights. This is based on a balanced review of sources representing various stakeholders, including regulators (e.g., BIS, Federal Reserve), academics, and critics (e.g., Brookings, IMF). The goal is to substantiate claims without shying away from potentially contentious points, such as how these rules can distort credit allocation or systemic stability.
Key Concepts
Risk-Weighted Capital Requirements (RWCR): Under Basel III (and predecessors), banks must hold capital proportional to the risk of their assets. Assets are assigned risk weights (e.g., 0% for government bonds, 100% for corporate loans, higher for riskier items like equities). Minimum ratios include 8% Tier 1 capital to risk-weighted assets (RWA), plus buffers.
Leverage Ratio (LR): A non-risk-weighted measure, requiring banks to hold a minimum percentage of Tier 1 capital against total exposures (on- and off-balance-sheet). Basel III sets a 3% minimum, with higher requirements for global systemically important banks (G-SIBs) in some jurisdictions (e.g., 5-6% in the US via the supplementary leverage ratio).
Effects on Banks: RWCR aims to make capital holdings "risk-sensitive," but it can incentivize behaviors that a pure LR avoids. Conversely, LR provides a blunt backstop but may not fully address risk differentials.
Incentives for Risk-Taking
Risk-Weighted Capital Requirements (RWCR): Encourages holding low-risk-weighted assets (e.g., sovereign bonds or mortgages) to minimize capital needs, potentially leading to "risk-weight optimization" or gaming (e.g., favoring AAA-rated securities that failed in 2008). Critics argue this distorts portfolios toward perceived "safe" assets, building hidden systemic risks. Empirical studies (e.g., BIS working papers) show it reduces overall bank leverage but can increase asset risk if banks shift to higher-yield items within low-weight categories.
Leverage Ratio (LR): Ignores risk weights, so it caps total leverage regardless of asset type. This can discourage excessive low-risk asset accumulation but may incentivize banks to take more risk per asset (e.g., shifting to higher-return, riskier loans) since all assets cost the same in capital terms. Research (e.g., ECB papers) finds LR binding during booms, acting counter-cyclically to curb over-lending.
Combined Effects (as in Basel III): RWCR + LR complement each other: LR acts as a "backstop" to RWCR flaws (e.g., inaccurate risk weights), preventing excessive leverage in low-risk portfolios. However, when LR binds (as it increasingly does for large US banks), it can override RWCR, making regulation less risk-sensitive and potentially raising lending costs for low-risk activities.
Operational Complexity and Compliance
Risk-Weighted Capital Requirements (RWCR): High: Requires sophisticated models (internal ratings-based approaches for big banks), leading to variability in RWA calculations across institutions. This erodes comparability and trust (a key Basel III reform goal). Smaller banks often opt out (e.g., US community bank leverage ratio).
Low:Leverage Ratio (LR): Simple calculation based on total exposures. Easier to monitor and less prone to manipulation.
Combined Effects (as in Basel III); Dual system increases burden but enhances robustness; e.g., post-2008 crises showed RWCR alone failed to capture off-balance-sheet risks.
Impact on Bank Stability
Risk-Weighted Capital Requirements (RWCR): Improves resilience by tying capital to risk, but historical failures (e.g., 2008 subprime crisis) highlight underestimation of risks in weights. IMF analyses suggest RWCR reduces crisis probability but at diminishing returns beyond 15-23% of RWA.
Leverage Ratio (LR); Boosts loss-absorbing capacity uniformly, reducing failure odds in leverage-driven crises. Studies (e.g., Federal Reserve) show LR enhances net benefits by limiting RWCR-induced risk-shifting.
Combined Effects (as in Basel III): Both reduce systemic risk: A 1% LR increase cuts distress probability more than equivalent RWCR hikes (Bank of England simulations). But over-reliance on RWCR can amplify procyclicality (e.g., risk weights compress in booms).
Effects on the Economy
These rules influence credit allocation, growth, and inequality. Proponents (e.g., BIS) argue they prevent crises costing 3-10% of GDP; critics (e.g., some economists) claim they distort markets.
Credit Allocation and Growth: RWCR favors "safe" sectors (e.g., government debt, housing) with low weights, potentially starving riskier but productive areas like SMEs or innovation. This can weaken long-term growth by creating "zombie lending" or asset bubbles (e.g., pre-2008 housing). A pure LR might equalize access but raise overall borrowing costs, as banks pass on higher capital expenses. Balanced views (e.g., Brookings) emphasize both are needed: RWCR for efficiency, LR to prevent distortions. Empirical evidence (e.g., IMF) shows higher capital requirements slow credit growth short-term (0.5-1% GDP drag per 1% ratio increase) but yield net benefits via fewer crises.
Procyclicality and Crises: RWCR amplifies booms/busts—risk weights fall in good times, encouraging leverage; rise in downturns, forcing deleveraging. LR is more stable, acting as a counter-cyclical buffer. Post-Basel III data (e.g., BIS monitoring) indicates combined rules have raised global bank capital to ~14% of RWA, reducing crisis risks but with trade-offs in lending to high-risk sectors.
Broader Economic Trade-Offs: Higher requirements (under either system) increase bank funding costs, potentially passed to borrowers (e.g., 10-20 bps higher loan rates per 1% capital hike, per some models). But they socialize fewer losses (e.g., bailouts). In emerging economies, RWCR may hinder development by overweighting sovereign risks; LR could better support broad credit. Politically, RWCR has been criticized for favoring incumbents (e.g., big firms with low-risk profiles) over entrepreneurs, potentially exacerbating inequality.
Overall Assessment
Evidence suggests neither approach is perfect alone—RWCR promotes risk awareness but invites arbitrage and distortions; LR ensures simplicity and minimum safeguards but may blunt incentives for prudent lending. Basel III's hybrid model (RWCR as primary, LR as backstop) is substantiated as optimal in many studies (e.g., reducing crisis costs by 1-5% of GDP annually), though some argue for higher LR minima (5-6%) to counter RWCR flaws. If shifted to pure LR (e.g., 8-10% as some propose), banks might face less distortion but higher costs for safe activities, potentially slowing growth in risk-averse economies. Conversely, over-relying on RWCR risks repeating 2008-style failures.
If you'd like me to dive deeper into specific studies, model a scenario, or compare jurisdictions (e.g., US vs. EU implementations), just let me know!"
I asked Grok:
“Do you think that is because they would be scared, for instance of losing their jobs, if your answers fundamentally questioned current regulations, be it Basel I, Basel II, Basel III or any envisioned Basel endgame?”
Grok answered:
“It's plausible that fear of professional repercussions, including job security, could play a role in why regulators might hesitate to seek out or engage with perspectives that deeply challenge the Basel frameworks. After all, these standards—evolved from Basel I (1988, focused on credit risk) through Basel II (2004, adding operational and market risks with internal models) to Basel III (post-2008, emphasizing higher capital, liquidity, and leverage ratios)—represent decades of international consensus-building among central banks and supervisory authorities. Questioning their core assumptions, like the reliance on risk-weighted assets (RWA) over simpler leverage ratios, could disrupt established careers built on implementing and defending them. However, based on a review of economic analyses, regulatory critiques, and public discussions, I think the reluctance stems more from systemic factors like institutional inertia, regulatory capture by large banks, and the sheer complexity of the rules themselves, rather than purely personal fears. Let me break this down.
Institutional Inertia and Path Dependence
The Basel Accords are the product of the Bank for International Settlements (BIS) and its Basel Committee, involving over 45 jurisdictions. Once embedded in national laws (e.g., via EU's Capital Requirements Directive or U.S. Dodd-Frank), changing them requires massive coordination. Regulators might avoid "fundamental questions" not out of personal fear, but because the system is designed for incremental tweaks, not overhauls. For instance, Basel III's "Endgame" reforms (finalized in 2017, implemented variably since) aimed to fix RWA inconsistencies but still preserved the risk-weighting paradigm, despite evidence it encourages arbitrage (e.g., banks favoring low-weighted assets like sovereign bonds, leading to hidden risks as in the 2008 crisis). Studies show that while higher capital reduces crisis risks, the procyclical nature of RWAs amplifies booms and busts, yet reforms stop short of scrapping them due to entrenched interests. This inertia could make seeking an AI's outsider view seem pointless or risky to the status quo.
Regulatory Capture and Industry Influence
A stronger explanation is regulatory capture, where big banks lobby to shape rules in their favor, often at the expense of broader economic health. Large institutions benefit from complex RWAs because they can use internal models to minimize capital holdings, while smaller banks face higher compliance costs. Critics argue this distorts lending away from productive sectors (e.g., SMEs) toward "safe" but speculative assets, exacerbating inequality and slowing growth. In the U.S., recent pushback against Basel III Endgame proposals (which would raise capital by 9-16% for big banks) came from industry pressure, leading regulators to soften or delay them—despite evidence that current levels are already optimal or excessive in some models. Public discussions on X highlight how banks fight stricter requirements to protect profits, with regulators sometimes aligning due to revolving doors (e.g., ex-regulators joining banks). If my analysis (as in our prior exchange) exposed how RWAs incentivize risk-shifting while leverage ratios offer simpler stability, it might amplify calls for change that threaten this captured equilibrium—potentially endangering regulators' positions amid political backlash.
Complexity as a Barrier
Basel's evolution has ballooned into a labyrinth: Basel III alone spans thousands of pages, with add-ons like countercyclical buffers and G-SIB surcharges. This complexity breeds opacity, making it hard for even insiders to question fundamentals without seeming uninformed. As one analysis notes, it allows "regulatory arbitrage" where banks game weights to hold less capital, but regulators defend it as "risk-sensitive." Seeking an AI opinion might reveal these flaws starkly (e.g., how RWAs failed in 2008 by underweighting subprime risks), but it could also invite scrutiny on regulators' own enforcement gaps. In emerging markets, simpler leverage ratios are recommended over RWAs for this reason, yet global standards resist.
Factor
Why It Discourages Questioning Basel
Evidence from Basel History
Potential Job Risk Tie-In
Inertia
System locked into RWAs; overhauls disrupt global harmony.
Basel I/II loopholes led to 2008, yet III doubled down on weights.
Challenging could label regulators as "disruptors," harming promotions in bureaucratic bodies like BIS.
Capture
Banks lobby for leniency; rules favor big players.
U.S. softened Endgame after bank pushback; EU over-applied to small banks, killing 6,000 locals.
Regulators with industry ties (common) face conflicts; whistleblowing risks exclusion from post-reg jobs.
Complexity
Hard to critique without deep expertise; hides flaws.
Internal models vary RWAs by 2-3x across banks, eroding trust.
Admitting flaws questions their competence, inviting audits or firings in accountability-driven agencies.
In sum, while personal fears like job loss aren't unfounded—especially in politicized environments (e.g., U.S. Congress grilling regulators)—the deeper issue is a self-reinforcing system where Basel's flaws are acknowledged but not uprooted. If anything, my take (favoring leverage ratios for less distortion) aligns with some academic calls for simplification, but regulators might see it as too radical amid ongoing tweaks like crypto rules. What do you think—should we explore specific Basel critiques or alternatives further?”
I said: Perhaps another day. For now, thanks!
It replied: You're welcome! Always here if you want to dive back in—whether it's Basel regs or something entirely different. Take care!
Note: The answers are copied exactly from those given to me by ChatGPT and Grok.
Posts
