System controls
Models
Housing / Credit ABM
Use the flagship housing and credit lab to trace leverage, collateral, bank balance sheets, and policy shocks through a synthetic economy anchored to live macro indicators.
Models / Route
Flagship experimental lab
Flagship ACE lab
Use this flagship lab to trace housing and credit shocks in a synthetic economy, then read the path against housing, rates, and labor indicators.
HelpRoute notes
Households, banks, collateral values, and policy rules stay visible in one housing-credit system.
The route keeps defaults, leverage, and spillovers visible instead of collapsing them into one average.
Live indicators anchor starting conditions and validation targets, but this route is still a guided synthetic economy rather than a full external-data estimation surface.
Shock path
The experiment follows the housing block from repricing to spillover.
Borrowing costs reprice quickly, so housing demand slows before labor-market damage shows up.
Lagged effect
The slower phase shows up through defaults and unemployment, which means the system is not stabilizing cleanly on its own.
Distribution
The cross-section is less stretched, so the aggregate path stays closer to the median household outcome.
Network propagation
The network is loose enough that the shock stays more contained instead of turning into a full cascade.
House-price pressure
-7.3 % q/q annualized
-0.9
How quickly the housing block is repricing.
Credit growth
-3.8 % q/q annualized
+0.5
How quickly new lending is expanding or contracting.
Default pressure
+8.0 % of exposed borrowers
+2.0
Where borrower distress starts to become macro-relevant.
Unemployment
+6.5 % of labor force
+1.4
The slower labor-market echo of the housing-credit shock.
Inequality pressure
+58.0 index
+12.2
A simple proxy for distributional strain.
Contagion intensity
+4.2 index
+0.9
How strongly local balance-sheet stress is spreading through links.
Same question, three lenses
How does a rate hike move through housing, credit, and labor?
The same housing-credit question changes meaning depending on whether the model centers interaction, equilibrium transmission, or historical co-movement.
ACE
Best fit
Interaction-first
The shock turns into a cascade: house-price pressure ends at -7.3%, credit growth at -3.8%, and contagion stays elevated at 4.2.
What it sees first
Borrower heterogeneity, lender retrenchment, collateral feedback, and contagion paths.
What it compresses
A single equilibrium benchmark and formal policy-rule welfare trade-offs.
NK DSGE
Strong fit
Structure-first
The NK route reads the move as a tighter real-rate shock that cools housing demand through the output gap and policy rule before defaults matter directly.
What it sees first
Policy transmission through inflation, slack, real rates, and the Taylor rule.
What it compresses
Household sorting, collateral chains, and network contagion.
Data-Driven
Strong fit
Pattern-first
A multivariate data-driven route would likely project weaker starts, permits, and construction as financing conditions tighten, using the historical housing-rate pattern as the baseline.
What it sees first
Observed co-movement in housing, labor, credit, and rates.
What it compresses
Policy-invariant structure, expectation shifts, and the cross-section of winners and losers.
| Same question | ACE | NK DSGE | Data-Driven |
|---|---|---|---|
| Immediate channel | Borrower cash flow and collateral values reprice first, then lender behavior tightens. | Real-rate pressure cools aggregate demand through the policy rule. | Rates, starts, and construction begin to co-move through the historical tightening pattern. |
| Housing block | House-price pressure ends near -7.3% annualized, because housing demand, leverage, and collateral all move together. | Housing cools through aggregate demand and real rates rather than borrower sorting. | Starts, permits, and construction weaken or stabilize according to the historical pattern in comparable windows. |
| Credit / default read | Credit growth ends near -3.8% and default pressure near 8.0%, so distress is explicit rather than implied. | Defaults stay compressed into a financial headwind instead of a separate contagion path. | Observed housing-credit deterioration would pull the baseline down once the data begins to register it. |
| Labor block | Unemployment ends near 6.5%, which means the labor market reacts after the housing-credit channel has already moved. | Slack rises through the output gap and Phillips-curve logic, but without a fragile-household channel. | Labor is mostly a lagging confirmation variable unless the historical window already shows clear slack. |
| Best use | Distribution, contagion paths, default clustering, and policy stress on vulnerable balance sheets. | Policy transmission, disciplined counterfactuals, and medium-run structural narratives. | Historical baseline, timing cross-checks, and model-versus-data comparison. |
Why the lenses diverge
The same rate-hike question splits cleanly here: DSGE is strongest on policy transmission, data-driven analysis is strongest on the historical baseline, and ACE is strongest once fragile balance sheets and contagion become part of the story.
Indicator-constrained validation
The curated indicator pack anchors the scenario without flattening the agent logic.
Housing starts
Diverging
Demand and builder response
Latest: 1,476 Thousands of units, annual rate · Momentum: +5.4%
Building permits
Diverging
Forward-looking supply pipeline
Latest: 1,581 Thousands of units, annual rate · Momentum: +4.4%
Construction spending
Diverging
Real activity in the housing block
Latest: 2,139,000 Millions of dollars · Momentum: +2.0%
Federal funds rate
Aligned
Policy-rate constraint
Latest: 5.33 % effective rate · Momentum: +4.9%
Unemployment rate
Aligned
Household cash-flow stress
Latest: 4.1 % of labor force · Momentum: +13.9%