Eurace Large-Scale Macro ABM Model
How do goods, labor, capital-goods, credit, and financial markets interact through heterogeneous agent decisions to produce emergent business cycles, credit crunches, and policy transmission effects that no single-market model can capture?
Understand the mathematics
Follow the same equations and assumptions one section at a time or in a continuous reading.
Hover or focus a symbol defined in this model to find its exact matches. Select it to keep the highlight; Escape clears the selection.
Odd+D Protocol: Model Specification
The ODD+D (Overview, Design concepts, Details + Decision-making) protocol provides the standard documentation template for agent-based models (Grimm et al. 2006, 2010; Muller et al. 2013). The protocol is applied here to the Eurace large-scale macro ABM as specified in Dawid et al. (2012, 2018) and Deissenberg et al. (2008).
PURPOSE: The model constructs a complete macroeconomy from the bottom up - interconnected markets for goods, labor, capital goods, credit, government bonds and (in the order-book branch) equity, populated by thousands of heterogeneous agents - to study how aggregate business cycles, credit crunches, unemployment dynamics, and inflation emerge from decentralized interactions without any representative-agent aggregation. Housing is NOT among them: the published Eurace baselines have no housing market, and the housing module is a documented extension. The target audience is macro policy analysts who need to evaluate fiscal, monetary, and regulatory interventions in a setting where distributional effects and cross-market spillovers matter.
ENTITIES, STATE VARIABLES, AND SCALES: The model contains four agent populations plus two institutional agents. Households ( to in standard runs) carry: income, bank deposits, government bond holdings, equity portfolio (order-book branch only), employment status (employed/unemployed + employer id), reservation wage, consumption budget, job search list, and goods shopping list. Consumption-goods firms ( to , with as the standard ratio) carry: capital stock , technology level, labor force (list of employed households), inventory level, posted price, wage offer, outstanding bank loans, retained earnings, dividend policy, and expected demand. Capital-goods producers ( to ) carry: technology, labor force, posted machine price, and an order backlog. Banks ( to ) carry: deposit base, loan portfolio (list of active loans to firms), bond holdings, bank capital , capital ratio , interbank borrowing/lending position, and deposit rate. A note on symbols, because three different objects compete for the letter in this literature: is a FIRM's capital stock (a quantity of machines), is a BANK's capital (an equity level in money), and is the regulatory capital RATIO (dimensionless). This reference keeps them distinct throughout. The central bank carries: policy rate, inflation target, and standing facility rates, all quoted as ANNUAL rates. The government carries: tax rates, benefit replacement ratio, outstanding bond stock, and fiscal rule parameters. Temporal scale is monthly; typical runs span 200 to 600 months. Spatial scale in Eurace@Unibi is a small number of discrete REGIONS - published setups typically use two - with matching constrained by region; it is not a fine-grained 50x50 lattice, and it is absent altogether in the Genoa variant.
PROCESS OVERVIEW AND SCHEDULING: Each month proceeds in a fixed order: (1) firms form demand expectations and set production plans net of inventory, (2) labor market opens - consumption-goods firms and capital-goods producers post vacancies, unemployed workers search, matching occurs, (3) production takes place, constrained by both labor and installed capital, (4) goods market opens - firms post prices, households shop, transactions clear, (5) firms compute investment demand from capacity utilization and place orders in the capital-goods market; producers deliver and the capital stock updates, (6) firms evaluate cash positions and apply for bank credit if needed, (7) banks screen applications and allocate credit subject to capital constraints, granting a graduated fraction rather than an all-or-nothing decision, (8) firm and bank defaults are processed, losses allocated, (9) government collects taxes, pays transfers, and issues bonds to cover the deficit, which banks and households absorb, (10) in the order-book branch only, the financial market runs for rounds (typically 5-20) and households trade equity, (11) central bank updates the policy rate via a Taylor rule fed with YEAR-OVER-YEAR inflation, so the inflation measure and the annual target are on the same footing, (12) aggregate statistics recorded. This ordering is economically motivated: production decisions are made before goods market clearing, investment orders are placed after capacity utilization is known, credit is sought after the cash flow gap is known, and defaults are resolved before the next period's planning begins.
Firm 's output is the minimum of labor-based capacity (productivity times workers) and capital-based capacity (the firm's capital stock divided by the capital-output ratio ). The Leontief structure means both inputs must be present; firms cannot substitute one for the other at the margin. This constraint is part of the runnable specification, not just the narrative: the graph tab's production step applies the same and the seed data assigns with a labor ceiling deliberately just below the capital ceiling, so both arguments matter. A version that replaced the capital term with expected demand would be a different model - one in which capital is initialized and never read.
Firm adjusts its price based on the gap between target inventory () and actual inventory. When inventory is below target (excess demand), price rises; when above target, price falls. The speed is a calibrated behavioral parameter, typically 0.01 to 0.05.
The single credit rule used in all three tabs of this reference. Bank grants firm the smaller of the firm's request and a fraction of the bank's capital - both CURRENCY amounts, which is the dimensional check that matters here; taking the minimum of a loan request and a capital RATIO, as an earlier version of the graph tab did, compares a money amount to a pure number and is not a well-formed expression. The capital ratio enters separately, as the gate : it is the bank's distance above the regulatory floor scaled by a buffer width and clipped to [0,1]. Graduating the gate rather than switching it on and off matters for behavior, not just for elegance - a binary gate turns any requirement above a bank's current ratio into exactly zero lending from the first period, a corner solution that looks nothing like rationing. With the graduated form, a bank near the floor lends a reduced amount and each subsequent loss tightens it further, which is the endogenous credit contraction the model is supposed to produce.
Stock-flow consistency is non-negotiable
Every Eurace transaction debits one agent's account and credits another's. Household spending reduces household deposits and increases firm deposits. Wages reduce firm deposits and increase household deposits - in the capital-goods sector as well as the consumption-goods sector. Firm investment spending reduces the buying firm's deposits and increases a capital-goods producer's. Bank loans create deposits (credit creation). A government deficit is a bond issue that some bank or household holds as an asset. This accounting discipline means aggregate demand identities hold exactly in every period, which is not guaranteed in models that treat money as a veil - and it is also a constraint on the agent list itself: an investment rule with no capital-goods seller, or a deficit with no bond purchaser, would break the identity no matter how carefully the rest of the accounting were written.
Why search frictions matter for macro dynamics
In the goods market, households visit only firms. This means a firm with a high price does not instantly lose all customers - it just loses the fraction who happened to also visit a cheaper competitor. This generates persistent price dispersion and allows firms with market power to extract surplus. In the labor market, limited applications mean vacancies and unemployment coexist (the Beveridge curve emerges endogenously rather than being imposed).
Leontief vs. Cobb-Douglas production
Eurace uses Leontief (fixed-proportions) production rather than Cobb-Douglas. This is a deliberate choice: with Leontief, firms must have both capital and labor in the right proportion, making investment and hiring genuinely complementary. With Cobb-Douglas, a firm could substitute capital for labor smoothly, which would dampen the labor market dynamics that are central to the model's business cycle mechanism.
Transition checks
Agent update
The aggregate is computed after individual states are updated. If the rule is nonlinear or stochastic, the aggregate can change abruptly even when the plotted line connects adjacent periods.
Each point is an aggregate after one model update. The connecting line helps the eye, but the law is discrete.
Do not read the line as a continuous differential equation unless one is stated.
Cutoff and contagion
The indicator changes value discontinuously at . Aggregates inherit that discontinuity when enough agents sit near the cutoff.
A cutoff creates a kink or jump because the update rule changes at the threshold.
Do not force a smooth explanation onto a chart section driven by threshold crossing.
Variable timing
Agent state
Agent states update at dates. Abrupt aggregate movement can come from many small state changes applied at once.
Aggregate readout
The aggregate can jump when the distribution changes shape or many agents cross a boundary.
Threshold indicator
The indicator turns on at the cutoff. That creates a discontinuity in the micro rule and possibly the aggregate.