ZIThe Zhong Institute

Policymakers need better tools—tools that are timely, explainable, and built to the standards the moment demands. This page outlines the challenges that motivate our work.

The Post-COVID Macro-Financial Landscape

The years since 2020 have fundamentally reshaped the macro-financial environment.

Compressed Policy Cycles

Central banks worldwide have executed some of the most aggressive rate cycles in decades. The rapid shift from near-zero rates to restrictive policy exposed vulnerabilities accumulated during the low-rate era.

Duration and Funding Stress

Banks and nonbank financial institutions faced severe interest-rate risk as asset values declined sharply. The 2023 banking turmoil demonstrated how quickly duration mismatches can become existential threats.

Nonbank Vulnerabilities

The growing role of nonbank financial intermediaries—money market funds, hedge funds, private credit—has created new channels of systemic risk that are less visible and less regulated.

Market Disorder Episodes

From Treasury market stress to crypto contagion, recent years have seen multiple episodes of market dysfunction that threatened broader stability.

What Official Assessments Show

U.S. financial authorities have documented these vulnerabilities in detail.

The Federal Reserve Financial Stability Report (April 2025) identifies ongoing risks from valuation pressures, borrowing levels, financial-sector leverage, and funding fragilities.

— Federal Reserve

The FSOC Annual Report (December 2024) emphasizes nonbank vulnerabilities, funding market stress, and the need for improved risk monitoring across the financial system.

— Financial Stability Oversight Council

The Office of Financial Research Annual Report (2024) highlights systemic-risk channels, data gaps, and the critical importance of high-frequency monitoring.

— Office of Financial Research

These official assessments make clear: the risks are known, but the tools to address them remain inadequate.

The Skills Gap

The Hybrid-Skills Problem

Effective AI/ML for financial stability requires a rare combination of capabilities. Most institutions have specialists in one or two domains—but very few have professionals who combine all five at production quality.

Skill DomainWhat's Required
Macro-FinanceDeep understanding of sovereign, banking, currency, and corporate risk dynamics
Machine LearningProficiency with modern ML methods—gradient boosting, neural networks, ensemble techniques
Software EngineeringProduction-grade development: CI/CD, testing, version control, secure deployment
InterpretabilityAbility to make models explainable: SHAP values, model cards, governance documentation
Policy CommunicationTranslating technical outputs into actionable insights for senior decision-makers

In Central Banks & Regulators

  • AI/ML is viewed as promising but risky to adopt
  • Pilot projects stall because outputs cannot be explained
  • Dependencies on external vendors create concerns

In Academia

  • Research produces models that never reach production
  • Papers optimize for publication, not utility
  • Methods documented for researchers, not practitioners

In Private Sector

  • Cutting-edge tools remain proprietary
  • Commercial incentives don't align with stability objectives
  • Solutions designed for revenue, not public goods

Policy Imperatives

U.S. financial regulators have been explicit about what's needed. Their priorities align directly with our work.

Risk Identification

Early Warning Capabilities

Policymakers need indicators that update frequently enough to catch turning points, cover multiple risk dimensions, and are transparent enough to be trusted and acted upon.

Funding Runs

Liquidity Stress Detection

Institutions need real-time funding stress indicators, tools to detect shifts in depositor behavior, and early warning of liquidity fragilities before they become runs.

Supervision

Transparency Requirements

Supervisors need explainable models that can be interrogated and audited, governance artifacts that meet regulatory standards, and drift monitoring for reliability.

Methodology

Rigorous Analytics

The field needs open methodologies for community review, gap-aware cross-validation for missing data, and reproducible research that others can build upon.

Why Existing Solutions Fall Short

Academic Research

Strengths

Methodological innovation, peer review, theoretical rigor

Gaps

Rarely production-ready; documentation for researchers, not practitioners

Private Vendors

Strengths

Engineering quality, commercial support, regular updates

Gaps

Proprietary "black boxes"; expensive licenses; misaligned incentives

In-House Government

Strengths

Tailored to institutional needs; full control

Gaps

Hybrid-skills scarcity; competing priorities; difficulty attracting talent

Existing Think Tanks

Strengths

Policy expertise, credibility, independence

Gaps

Limited ML/engineering capability; focus on narrative over operational tools

What's Missing

An organization that combines think-tank independence and policy expertise, production-grade engineering capabilities, state-of-the-art ML with built-in explainability, and commitment to open, public-good outputs.

This is the gap we fill.

Learn About Our Work →