Baseline Macro ABM Model
How do heterogeneous firms competing through innovation, adaptive pricing, and credit-constrained investment generate endogenous business cycles, a heavy right-tailed firm-size distribution, and long-run growth without any exogenous shock process?
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 is the standard documentation format for agent-based models, introduced by Grimm et al. (2006, 2010) and extended by Muller et al. (2013) for decision-making. The protocol is applied here to the generic benchmark macro ABM documented on this page - a composite in the K+S / stock-flow-consistent tradition, not a transcription of K+S Mark I as published. For the published K+S specification, see the sibling K+S growth model page.
PURPOSE: The model generates endogenous business cycles, long-run growth, and realistic cross-sectional distributions (firm sizes, growth rates) from the decentralized interaction of heterogeneous firms with boundedly rational behavioral rules and credit constraints. The target audience is macroeconomists studying the micro origins of aggregate fluctuations, and policymakers evaluating fiscal, industrial, or competition policy where the firm-size distribution is a first-order concern.
ENTITIES, STATE VARIABLES, AND SCALES: The model contains three entity types plus a banking sector. Capital-goods firms (F1, typically 50) carry: technology frontier (continuous, positive), R&D spending (fraction of revenue), client list (set of F2 firm ids), price (continuous, positive), labor force (F1 firms build machines with labor, so they hire in the same labor market as F2 firms and their wage bill is part of aggregate wage income), and accumulated profits. Consumption-goods firms (F2, typically 200) carry: machine stock (vector of vintages with heterogeneous productivities), expected demand (continuous), market share (continuous, 0 to 1), price (continuous), inventory (continuous, non-negative), retained-earnings stock (continuous, may be negative), period profit flow (continuous, may be negative - a distinct object from the stock), and bank debt (continuous, non-negative). Workers ( to ) carry: employment status (binary, employed in either sector) and wage (shared, economy-wide). The bank carries a full balance sheet: loans to F2 firms and reserves on the asset side, household deposits and equity on the liability side, plus the credit multiplier. The deposit liability is what allows the loan book to exceed equity and therefore what allows the credit multiplier to bind at all; with equity as the only liability the balance sheet would force loans to equal equity and the constraint would be vacuous. The temporal scale is quarterly periods over 200 to 500 quarters, with a transient burn-in of 50 to 100 periods discarded before analysis.
PROCESS OVERVIEW AND SCHEDULING: Each period proceeds in 13 steps grouped into three blocks. Schumpeterian block (steps 1-2): F1 firms invest in R&D, draw innovation outcomes, update technology frontiers, and set machine prices. Keynesian block (steps 3-7): F2 firms form demand expectations, plan production, compute investment demand, apply for credit, and order machines. Market-clearing block (steps 8-13): the labor market clears, F2 firms produce, the goods market updates market shares via the replicator dynamic, wages adjust, profits are computed, and entry/exit is processed. The ordering ensures investment conditions on financial positions and production conditions on actual labor availability.
Innovation process for F1 firm . The search hazard is concave in the firm's R&D budget relative to the sector average and is bounded strictly below 1 by construction - which is the point of the exponential form. A linear specification such as is not a probability: it exceeds 1 for any firm spending more than times the average budget. Conditional on success, the firm draws a productivity increment from a scaled Beta distribution; the Beta shape parameters control the mean and variance of innovations and scales the support.
The single market-share replicator used everywhere in this reference. Every symbol: is firm 's price and the sector average price; is firm 's unfilled demand (orders it could not serve from output plus inventory) and the sector average; and are non-negative weights on price and non-price competition; is the reallocation speed. Two details are load-bearing. First, is a sum of penalties and is therefore NEGATIVE by construction, so the deviation is divided by ; dividing by itself flips the sign and would hand every better-than-average firm a negative growth factor, the exact opposite of selection. Second, the benchmark is the SHARE-WEIGHTED mean, and the shares are renormalized explicitly, because the growth factors do not sum to one on their own and unnormalized shares would break the accounting identity that the shares of total demand add to 1.
Aggregate wage adjustment. Wages rise with average productivity growth (the channel) and with labor market tightness measured as the GAP between the target unemployment rate and the actual rate (the channel). The gap form matters: writing the tightness term as leaves a residual of at - 0.88 at the 6% target used here - so wages would keep rising at per period even with the labor market exactly at target. With the tightness contribution is zero at target, positive when unemployment is below it, and negative when above. This is the Phillips curve analog in the ABM: not a structural equation but a behavioral rule calibrated to wage-setting evidence.
The bank's balance sheet, stated so the credit multiplier constraint can actually bind. Assets are loans to F2 firms plus reserves; liabilities are household deposits plus bank equity. Without the deposit liability the identity would collapse to loans equal to equity, pinning the loan book at exactly one times equity and making a multiplier of unreachable and therefore never binding. With deposits present, the loan book can sit anywhere below CR times equity, and firm defaults that write down equity pull the ceiling toward the current loan book - which is the financial accelerator.
The replicator dynamic comes from evolutionary game theory
The market share update rule is a discrete-time replicator dynamic from evolutionary game theory (Taylor and Jonker 1978), adapted to industrial dynamics by Silverberg et al. (1988) and Dosi et al. (1995). It replaces the Walrasian auctioneer: instead of a price that clears the market instantaneously, market share flows gradually from less fit to more fit firms. The speed of reallocation () is a structural parameter that captures both search frictions and switching costs in the goods market.
Why two sectors, not one
A single-sector model (firms produce a homogeneous good using labor only) cannot generate the investment channel that is central to the K+S business cycle mechanism. The two-sector structure separates the innovation source (F1 firms producing better machines) from the demand source (F2 firms buying machines and hiring workers). This separation is what allows Schumpeterian supply shocks and Keynesian demand shocks to interact rather than collapse into a single channel.
Burn-in is not optional
The model's initial conditions (equal market shares, identical machines, no accumulated profits) are far from the ergodic distribution. The first 50-100 periods are dominated by convergence dynamics: innovation draws break the initial symmetry, the replicator concentrates share, and retained earnings build up. Aggregate statistics computed during the burn-in are artifacts of initial conditions, not properties of the model. All published K+S results discard the burn-in.
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.