See who responds to a US shock, not just how much.

This is a validated replication of the two-asset heterogeneous-agent New Keynesian (HANK) model of Auclert, Bardóczy, Rognlie and Straub — the model of their Econometrica 2021 paper, "Using the Sequence-Space Jacobian to Solve and Estimate Heterogeneous-Agent Models" — built directly on the authors' sequence-jacobian toolkit at the paper's production grid sizes rather than the packaged demo grids.

Where frb-us answers "how much does US output move after a shock", this model answers a question frb-us cannot: who moves. Households differ in liquid and illiquid wealth, so their marginal propensities to consume differ, and the model carries that whole distribution through the general-equilibrium response. It is a VAR-free, sequence-space HANK: no estimated reduced form, no expectations shortcut — first-order impulse responses computed from the model's own general-equilibrium Jacobian.

It is not a forecaster, and it does not score detailed tax reforms: it scores stylized shocks — monetary, fiscal spending, and productivity — around a calibrated steady state, and its distributional outputs are first-order approximations from steady-state policies.

One design point matters for what you can ask it: the labor tax is endogenous. The fiscal block balances the government budget each period, so a spending shock is implicitly tax-financed — and there is no exogenous transfer or tax-rate instrument to shock. Requests for one are refused with that explanation rather than answered with an invented mapping.

The quickest route is the hosted MCP server, where two tools reach this model: hank_shock (run a stylized shock, optionally with the distributional block) and hank_summary (the shock catalogue, units, variants, and scope limits — instant, no solve). The first call per variant pays the steady-state and Jacobian solves, roughly 18 seconds cold; both are cached, so warm calls are effectively instant. The same two are on the CLI:

pe-macro hank-summary
pe-macro hank-shock --kind monetary --size -0.0025 --persistence 0.6 --distribution

Units differ per kind. monetary is a level change in the quarterly policy rate (−0.0025 = a 25bp easing); fiscal_spending is a level change in G (steady-state output is 1, so 0.01 = 1% of GDP); productivity is a level change in TFP. A variant option selects the paper's two_asset model (the default) or a fast one_asset textbook variant — no capital, so no investment response, and monetary/productivity shocks only.

For anything beyond the hosted surface, install the package and use its Python API directly:

pip install git+https://github.com/PolicyEngine/us-hank-model

from us_hank import model, distributional

ss = model.solve_steady_state()          # ~13s, cached
G = model.solve_jacobian(ss, T=300)      # ~5s, cached

irf = model.shock("monetary", size=-0.0025, persistence=0.6, ss=ss, G_jac=G)
print(irf["Y"][:8])                      # output, % deviation from steady state
print(distributional.summary(ss))        # MPCs by quartile, hand-to-mouth share

The model's designed job is the qualitative and quantitative transmission of a stylized aggregate shock through a realistic household wealth distribution: how a monetary easing raises output, consumption, investment and inflation; how a tax-financed spending expansion raises output while the endogenous labor tax leans on consumption; and how the response concentrates among high-MPC, low-liquid-wealth households — the economic point of a HANK model, and the thing a representative-agent model cannot say.