Macroeconomic Agent-Based Model

• Macroeconomic ABMs are computational frameworks that model diverse agents interacting in labor, goods, credit, and financial markets to generate emergent macroeconomic phenomena.
• They incorporate detailed micro-behavioral rules for households, firms, banks, and governments, ensuring stock-flow consistency and capturing effects like bank runs and policy shifts.
• ABMs facilitate counterfactual policy experiments and risk assessments by simulating scenarios such as CBDC uptake and financial contagion, offering actionable insights for macro stability.

A macroeconomic agent-based model (ABM) is a computational framework in which heterogeneous agents—households, firms, banks, governments, and central banks—interact across multiple economic markets, with aggregate macroeconomic phenomena emerging endogenously from their localized, adaptive behaviors. This modeling paradigm offers a highly granular, bottom-up alternative to representative-agent or equilibrium-based approaches, explicitly accommodating bounded rationality, financial fragility, and market frictions. Macroeconomic ABMs are particularly suited to studying complex phenomena such as bank runs, endogenous crises, financial contagion, expectation formation, monetary policy transmission, and the effects of regulatory constraints.

1. Core Structural Elements and Market Interactions

In macroeconomic ABMs, the economy is constituted by populations of agents, each endowed with explicit balance sheets and specific behavioral rules. Canonical models, such as in (Barucci et al., 24 Oct 2025), operationalize at least four core agent types:

• Households: Provide labor, earn income, make consumption/saving decisions, allocate wealth across deposits, assets, or digital currencies (e.g., CBDC).
• Firms: Produce goods/services, hire/fire workers, set prices/output, seek external financing from banks, and may face liquidity or solvency constraints.
• Banks: Intermediate liquidity, accept deposits, extend loans, participate in interbank markets, manage capital adequacy, and face default risk.
• Government/Central Bank: Impose fiscal policy, manage taxes and transfers, issue bonds (government), conduct monetary policy, issue currency or digital tokens (central bank), and set regulatory constraints.

Agent interactions span multiple interconnected markets:

• Labor Market: Demand-driven hiring/firing, wage-setting (e.g., by unions), frictions in matching unemployed workers to vacancies.
• Goods Market: Decentralized consumption based on permanent/perceived income, search frictions and adaptive firm-level price-setting.
• Credit Market: Firms' credit demands are accepted by banks subject to risk-based lending constraints, regulatory ratios (Basel-style), and capital costs.
• Interbank and Financial Markets: Banks adjust liquidity via interbank borrowing/lending, with interest rates reflecting counterparty risk; agents invest or seek liquidity.
• Public and Money Markets: Implementation of taxation, bond issuance, public spending, and instruments such as central bank digital currency (CBDC) as direct liabilities of the central bank.

Agents' balance sheets are tracked individually, ensuring full stock-flow consistency and global wealth conservation (excluding government net position).

2. Micro-Behavioral Rules and Endogenous Macro Dynamics

A defining feature is the explicit modeling of agent micro-behavior. Decisions are based on observed or perceived state variables, expectations, and regulatory or market constraints, often far from the fully rational, globally optimizing benchmark.

• Household liquidity allocation: Households shift deposits from banks to safe assets (such as CBDC) in response to rising perceived bank risk, operationalized by leverage or other risk metrics (Barucci et al., 24 Oct 2025). The allocation is mathematically parameterized, e.g.,

$$CBDC^h_{i,t} = \psi(RM_{h,t}) \cdot w_i^h \cdot nw^H_{i,t}$$

where $\psi(RM_{h,t})$ is a behavioral function of riskiness.

• Firm production and demand: Output and price targets are set adaptively based on past demand, inventory, and expected profits.
• Bank risk management: Bank lending is constrained by capital adequacy, leverage, and liquidity ratios, with sensitivity to loan risk as determined by borrower balance sheets and internal or regulatory models.

These micro-rules are critical in generating non-trivial aggregate outcomes: unemployment, output, interest rates, or systemic stability are not imposed top-down, but emerge from the distributed agent interactions and adaptation.

3. Phase Transitions, Instability, and Policy Transmission

Macroeconomic ABMs are uniquely capable of generating discontinuous regime shifts, tipping points, and endogenous instability, owing to the feedback-rich, nonlinear nature of agent decision rules (Gualdi et al., 2013, Gualdi et al., 2015).

• Bank-run phenomena and flight-to-quality: Introduction of a fully risk-free asset (CBDC) produces state-contingent withdrawal dynamics, with risk-driven deposit flight endogenously escalating into full-scale bank runs if not curtailed (e.g., through limits on CBDC holdings) (Barucci et al., 24 Oct 2025).
• Phase transitions: Models such as Mark-0 (Gualdi et al., 2013) exhibit sharp transitions between high and low unemployment (“good” and “bad” phases), often governed by parameters like the hiring/firing asymmetry or credit constraints.
• Monetary and regulatory policy: Transmission of CB’s policy rate (via Taylor rules) and the bank sector’s regulatory response can, depending on the model region, promote either stability or crisis (so-called “dark corners” (Gualdi et al., 2015)), often contradicting DSGE policy implications.

The non-equilibrium, path-dependent nature of these models allows for amplification of shocks, regimes of persistent unemployment, inflation/deflation cycles, and endogenous oscillations, all absent in standard rational-expectations frameworks.

4. Quantitative Macroeconomic Outcomes and Welfare Analysis

Quantitative evaluation in macroeconomic ABMs is rooted in Monte Carlo simulation of model trajectories, often spanning thousands of agent-lives and decades of macro time. Key findings from high-fidelity simulations include:

• Banking sector fragility: Unconstrained CBDC uptake (up to 80% of deposits reallocated when risk is high) markedly raises bank default rates (from ~7% baseline to ~12%), triggers large asset liquidations, and elevates tail risk (GDP can drop to 50% of baseline in worst-case percentiles) (Barucci et al., 24 Oct 2025).
• Policy design effects: Imposing a moderate CBDC cap (≤ 40% of deposits) or coupling adoption with deposit insurance mitigates run risk, reduces bank failures, and largely preserves macro stability and welfare.
• Redistribution and welfare: Households accumulate wealth and are insulated from bank failures as CBDC uptake increases, whereas banks and firms lose out; real welfare impact, assessed via Atkinson or Mean-Variance social welfare functions, is negative under unconstrained digital runs but mildly positive with capped adoption.
• Interest rates and credit: Flight-to-quality to CBDC reduces bank funding, raising loan rates to firms (by up to 11bps) and minimally contracting credit supply, with effect size closely tied to the degree of CBDC uptake.
• Aggregate macro impact: Despite financial system volatility, aggregate variables (GDP, unemployment) move little on average under bounded CBDC, but experience high variance and downside risk otherwise.

A summary table consolidates the comparative effects across scenarios:

CBDC Scenario	Bank Run Risk	Financial Stability	Macro Effects	Welfare
10% Fixed (CBDC0)	Very Low	High	Negligible	Mild ↑
Risk-based up to 80% (CBDC1)	High	Low	Sharply worse crises	↓
30% Cap, Risk-based (CBDC2)	Low	High	Minimal	Mild ↑
Insurance-based (CBDC4)	Low	High	Minimal	Neutral
No Cap	Severe	Unstable	Severe downside	Marked ↓

5. Advances, Methodological Innovations, and Implications

Macroeconomic ABMs embody several methodological advances:

• Behavioral realism: Agent behaviors are encoded as rules or functions of observable (individual or market) states, capturing a broad behavioral spectrum from memory-based adaptation to risk-triggered switches.
• Stock-flow consistency: All financial flows are explicitly modeled, ensuring system-wide consistency and conservation.
• Calibration and validation: Modern ABMs can now be calibrated on real data (via synthetic population generation, sectoral detail, and Bayesian inference), allowing quantitative forecast validation (Wiese et al., 27 Sep 2024).
• Policy experimentation: The agent modularity enables counterfactual experiments with alternative monetary regimes (e.g., sovereign money (Peters et al., 2022)), regulatory rules, or financial innovations, by modifying agent rules or asset space.

The policy implications are substantial: fears of systemic disintermediation from CBDC introduction are unwarranted provided policy design tightly constrains uptake. More generally, ABMs highlight the possibility of design features—such as holding limits, risk-dependent conversion rules, or insurance caps—that mitigate macro-financial externalities.

6. Future Directions and Open Challenges

Contemporary research continues to extend macroeconomic ABMs in the following directions:

• Parameter space exploration: Efficient algorithms identify “stiff” parameter directions responsible for regime shifts, allowing for tractable phase diagram mapping even in high dimensions (Naumann-Woleske et al., 2021).
• Expectation formation: LLM-powered ABMs simulate heterogeneous, micro-founded expectations with unprecedented realism, exceeding traditional rule-based agents in capturing distributional and semantic nuance (Lin et al., 23 May 2025, Li et al., 2023).
• Ecological realism and urban dynamics: Integrated frameworks with spatial, sectoral, or urban expansion modules (e.g., SimCity (Feng et al., 1 Oct 2025)) reproduce stylized macro-urban phenomena and their feedbacks.
• Networked financial risk: Explicit modeling of credit networks, interbank liabilities, and systemic risk quantification (e.g., DebtRank (Terry-Doyle et al., 31 Jan 2025)) improves the realism of crisis propagation and risk contagion analysis.

A key challenge remains reconciling the rich heterogeneity and path dependence of ABMs with efficient calibration, robust forecast evaluation, and transparent exposition of causal mechanisms. Nevertheless, macroeconomic agent-based models now represent a robust and empirically-grounded platform for the analysis of stability, welfare, and policy in complex and evolving monetary economies.