Household Heterogeneity ABM Model
How do heterogeneous household consumption rules, saving buffers, and wealth positions shape the aggregate demand response to fiscal, monetary, and distributional shocks - and why does the shape of the income/wealth distribution matter more than the mean?
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 heterogeneous household ABM family.
PURPOSE: The model studies how heterogeneous household consumption, saving, and wealth accumulation interact to shape aggregate demand dynamics and the macroeconomic response to fiscal, monetary, and distributional shocks. The target output is the relationship between the cross-sectional distribution of wealth/income and the aggregate fiscal multiplier - a relationship that representative-agent models cannot produce. The target audience is fiscal policymakers evaluating transfer program design and central bankers analyzing the distributional channel of monetary policy transmission.
ENTITIES, STATE VARIABLES, AND SCALES: The model contains four entity types. Households ( to in research runs, 50 to 200 in toy demonstrations) carry state variables: income (continuous, positive), liquid wealth (continuous, non-negative in baseline), illiquid wealth (continuous, non-negative), employment status (binary), consumption rule parameters (, ), and a buffer-stock target (derived from income history). Firms ( to ) carry: output capacity, inventory, price, wage offer, profit, and a hiring/firing threshold. Banks ( to ) carry: deposits, loans, capital ratio, and interest rate spread. The government carries: tax rate, transfer schedule, and fiscal balance. The central bank carries: policy rate and Taylor rule parameters. Temporal scale is quarterly; spatial dimension is absent in the baseline. Typical simulation horizon is 80 to 200 quarters.
PROCESS OVERVIEW AND SCHEDULING: Each period proceeds in fixed order: (1) government collects proportional income tax and distributes transfers according to the fiscal rule, (2) firms set prices as a markup over unit labor cost and post vacancies based on demand relative to capacity, (3) the labor market clears through random matching of unemployed workers to vacancies with posted wages, (4) employed households receive wages plus any transfer; unemployed receive benefit plus transfer, (5) households make consumption/saving decisions using their adaptive rule, (6) the goods market clears through decentralized matching of household demand to firm supply, (7) firms record sales, update inventory, compute profits, and decide on investment, (8) banks collect deposits, process firm loan applications, and update balance sheets, (9) the central bank updates the policy rate via a Taylor-type rule, (10) aggregate statistics are recorded. The ordering ensures that household consumption is conditioned on current-period income and wealth, not on stale values.
Household 's consumption rule, written so that the MPC it implies is the MPC the model reports. Consumption equals permanent income, plus a fraction of the transitory deviation of income from permanent income, plus a drawdown (or rebuild) of the wealth gap at speed . The income MPC rises as the proportional shortfall of wealth against target rises, which is what makes the MPC-wealth gradient real rather than asserted. The first argument is the budget constraint: a household cannot spend more than wealth plus income.
Household 's wealth accumulation: end-of-period savings earn the deposit rate. This is the single form used everywhere in the file (the graph tab and the step-by-step derivation use the same expression). Substituting the consumption rule shows why it matters that enters with a NEGATIVE sign on the gap: the wealth process has autoregressive root , so wealth mean-reverts to the target instead of diverging. The earlier specification had root below target and no drawdown term above it, so the distribution was divergent and bimodal - incompatible with any stationary-Gini calibration target.
The MPC is the literal derivative of the consumption rule above, not a separate assertion. Because the permanent-income estimate is updated AFTER the consumption decision, and are predetermined within the period and the derivative is exactly . A household with zero wealth has ; one at or above target has ; one whose desired spending exceeds cash-on-hand has . This is the only MPC formula used in this file - the graph tab's equation list, this derivation, and the wealth-gap step all state the same thing.
Micro foundations of the buffer-stock rule
Carroll (1997, 2004) showed that the optimal consumption policy for a household facing uninsurable income risk and a borrowing constraint converges to a buffer-stock rule: consume a fraction of income that depends on the distance between current wealth and a target, where the target is proportional to permanent income. (Carroll 2006 is the endogenous-gridpoints method - a solution technique, not the buffer-stock theory paper, and it is frequently miscited for the theory.) The target itself depends on the income process parameters (persistence, variance) and the interest rate. The ABM simplifies this by fixing the proportionality constant quarters and calibrating the sensitivity parameter to match survey-reported MPCs, but it keeps the income-proportional target, which is what makes the MPC gradient survive: a target defined off median WEALTH rather than own INCOME would leave the derivative of consumption with respect to income equal to for everyone.
Empirical MPC distribution
Fagereng, Holm, and Natvik (2021) estimate MPCs from Norwegian lottery winners and find: bottom wealth QUARTILE MPC approximately 0.55, top quartile approximately 0.05, with a smooth decline in between. These are quartile figures and are cited as quartiles everywhere in this file; the model's own bar chart reports quintiles because that is the display convention, and the two should not be conflated. Johnson, Parker, and Souleles (2006) estimate MPCs from US tax rebates at 0.20-0.40, with higher values for low-income households. The buffer-stock consumption rule in the ABM is calibrated to match this declining-MPC-with-wealth pattern, not the average MPC from a representative-agent Euler equation. Hitting the 0.55/0.05 endpoints requires near 0.05 and near 0.50, since the model's MPC lives in ; at the toy's illustrative default of the reachable range is [0.70, 1.00] and neither endpoint is attainable.
Why sequential clearing matters for distribution
In a simultaneous-clearing (Walrasian) setup, all markets equilibrate at once and the distribution does not affect the clearing prices once you condition on aggregates. In the sequential setup, which market clears first determines which households face rationing. If the labor market clears before the goods market, unemployment is determined before consumption, and the consumption distribution reflects the employment outcome. Reversing the order changes the propagation path. This is not a bug - it reflects the real-world sequencing of payday, spending, and hiring decisions.
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.