Macro by Mark

Models / Route

Interaction-first route

Agent-based computational macroeconomics

Agent-based computational macro models begin where averages stop being enough. The route builds the economy from heterogeneous households, firms, banks, and policymakers whose local rules and interactions generate aggregate outcomes over time.

Section Notes
All ModelsAgent-BasedHousing / CreditDSGEForecasting
Route notes

ACE opens from heterogeneity, interaction, and emergence, with aggregate outcomes arriving from local rules and encounters.

Indicator data enters later as constraints, target moments, and validation targets that shape the synthetic economy.

The flagship template is housing and credit because leverage, defaults, inequality, and macroprudential rules stay visible in one intuitive system.

Why ACE exists

The doorway is heterogeneity, interaction, and emergence.

Agent-based computational macro models build the economy from the bottom up. Instead of starting from one representative agent or a compact equilibrium system, this route simulates many households, firms, banks, and policymakers whose local rules and interactions generate aggregate outcomes over time.

That makes ACE the natural route when averages hide the mechanism, when networks propagate the shock, or when nonlinear thresholds turn a modest disturbance into a boom, bust, or cascade. The point is not a single baseline point forecast. The point is to see who gets hit, how the shock travels, and what emerges from the interaction.

Empirical forecasting

Pattern first. Best when the question is what the historical data implies next.

DSGE

Structure first. Best when policy rules, expectations, and equilibrium conditions need to stay explicit.

ACE

Interaction first. Best when distribution, local rules, adaptation, and network propagation matter more than a representative average.

ACE routes

The route moves from a synthetic economy to constrained validation and policy stress.

Conceptual doorway

Why ACE?

ACE starts where averages stop being enough. The route begins with heterogeneity, local rules, and interaction before it ever asks what the aggregate series looks like.

This is the doorway when representative averages hide the mechanism or the distributional split.

Open route
Reference economy

Build a Demo Economy

The first lab is synthetic on purpose: households, firms, banks, government, and the central bank are assembled as a bottom-up economy before any validation target enters.

The early route teaches assembly and propagation, not blind calibration.

Open route
Interaction core

Add Heterogeneity and Networks

Income, wealth, leverage, productivity, and network structure change how shocks move. This is the section where emergence becomes visible.

Changes in distribution or topology need to change the macro result for understandable reasons.

Open route
Indicator-constrained

Calibrate and Validate

Indicator data enters here as initial conditions, target moments, and validation targets tied to the simulation itself.

The goal is disciplined confrontation with macro evidence, not pretending every agent is directly estimated from one series list.

Open route
Scenario lab

Run Policy Experiments

Rate hikes, LTV caps, transfers, hiring subsidies, supply shocks, and default cascades can be compared through distributions, network paths, and sector winners and losers.

ACE pays off when the question is how the shock propagates and who bears it.

Open route

Build a demo economy

The first lab is synthetic so the assembly stays legible.

A reference economy is easier to understand than a giant ingestion workflow. The model begins with households, firms, banks, government, and the central bank, then layers in income rules, borrowing constraints, pricing rules, and policy rules.

HouseholdsFirmsBanksGovernmentCentral bank

The aim is to watch the system take shape before any external constraint tells the user whether the aggregates look plausible.

Housing / Credit ABM

Flagship templateFlagship template
The strongest V1 template because heterogeneous households, banks, leverage, house prices, and macroprudential rules all stay visible in one intuitive system. LTV and LTI caps, rate moves, transfers, default stress, housing booms and busts.

Labor Market ABM

Distributional routePlanned extension
Worker search, firm hiring, wage adjustment, and unemployment duration become visible once firms and workers follow their own rules. Hiring subsidies, wage support, unemployment insurance, sector mismatch.

Firm-Bank Macro ABM

Contagion routePlanned extension
Firm balance sheets and bank lending links make amplification and contagion legible when credit channels matter as much as the aggregate shock. Credit support, capital buffers, lending standards, default cascades.

Supply-Chain / Network ABM

Propagation routePlanned extension
Network structure changes how shortages, energy shocks, and supplier failures travel through production, prices, and inventories. Input disruption stress tests, reshoring, bottleneck mapping, resilience planning.

Financial Market Microstructure ABM

Market dynamicsPlanned extension
Agent rules in trading, liquidity provision, and leverage make market dislocations, circuit breakers, and nonlinear stress more interpretable. Circuit breakers, leverage shocks, liquidity stress, fire-sale containment.

Heterogeneity and networks

The aggregate result changes because the distribution and the topology change.

This is the core ACE promise: local variation and local connection structure alter the macro path in ways a representative average cannot show.

Heterogeneity levers

Income dispersionWealth dispersionCredit constraintsFirm productivityBank balance sheetsAdaptive rules

These levers determine who cuts spending first, who defaults sooner, and which agents amplify the shock while others absorb more of it.

Network levers

Credit linksBank-firm chainsSupply linksMarket liquidity linksRegional spilloversContagion paths

Topology decides whether the same disturbance fades locally or becomes a system-wide cascade.

Calibrate and validate

Indicators constrain the economy and test it. They do not replace the agent logic.

Indicator data enters as initial conditions, target moments, and validation targets. That keeps ACE distinct from ARIMA-style raw series fitting.

Demo Economy

The first route teaches the model with a synthetic population and a reference economy before external targets constrain anything.

Synthetic householdsRule-based firmsBank balance sheetsPolicy block

Indicator-Constrained Economy

Observed indicators anchor the simulation through macro constraints and validation targets with a clear model role.

Initial conditionsTarget momentsValidation targetsScenario benchmarks

Advanced Data Mode Later

Survey and microdata layers can arrive later, extending the constraint logic without flattening the route into a generic fitting exercise.

Household survey dataFirm-level panelsCredit distributionsNetwork surveys

Policy experiments

Rate hike versus credit-constrained households

LTV cap versus house-price and leverage cycle

Fiscal transfer versus household cash-flow stress

Hiring subsidy versus labor-market slack

Default shock versus bank-firm contagion

Supply disruption versus network bottlenecks

Outputs that matter

Aggregate outputInflationUnemploymentCreditDefaultsInequalityNetwork contagionSector winners and losers

The payoff shows up in the distribution of outcomes, the propagation path, and the thresholds that make the scenario nonlinear.