Skip to content

Stock-Flow Diagnostic Map

Diagnostic model — instantiates Stock-First Control Restoration

Diagrams which underlying stock a flow lever actually acts through, and the range over which that stock still responds, so a dead lever can be traced to a stock outside its window rather than to insufficient force.

Stock-Flow Diagnostic Map is the causal picture that separates a lever from the stock it acts through. It draws the chain — lever → mediating stock → outcome — and annotates the stock with the range over which a marginal lever change still moves the outcome (its responsive window) and the boundary past which the coupling goes to zero (the dead zone). Its whole reason to exist is to reframe the archetype's central error: it turns "our lever is too weak, push harder" into "the stock is outside the window where the lever works." Everything downstream — where to set a floor, when to stop escalating, what to repair — is located on this map, so it is the shared reference frame rather than an action.

Example

A central bank's policy interest rate (the flow lever) does not touch the economy directly; it acts through a stock — the health of bank and household balance sheets and the credit intermediation they support. In normal conditions a rate cut flows through that stock into more lending and demand: the stock is inside its responsive window. The map's payoff appears in a balance-sheet-impaired economy at the zero bound, where it shows the transmission stock sitting outside its window — the classic liquidity trap — so further "cuts," or the wish to cut below zero, produce nothing.[1] With the mediating stock made explicit, the near-zero response reads correctly: not a lever that needs more force, but a stock that must be repaired (or a different tool that acts on it) before the rate lever can transmit again.

How it works

The map is built by tracing the lever's causal path to the single stock that mediates it — not the lever, not the final outcome, but the reservoir in between — and then characterizing that stock's behaviour: over what range does a marginal lever change move the outcome, and where does the coupling collapse. The distinctive act is that separation of lever from mediating stock; once the stock is named, the responsive window and its boundary can be drawn against evidence. The result is a static structure, refreshed when the system's wiring changes, not a running signal.

Tuning parameters

  • Grain of the causal chain — a single lever→stock→outcome hop or a multi-stage chain of intermediate stocks. Finer grain locates the true mediating stock but risks a sprawling, unusable map.
  • Window characterization method — derive the responsive range from theory, from historical response data, or from a live probe. Data-fit windows are sharper but need history that actually spans the boundary.
  • Boundary sharpness — model the dead zone as a hard cliff or a gradual roll-off; over-sharpening invites false confidence about exactly where control dies.
  • Refresh trigger — re-map on a schedule versus only when a structural shock is suspected of having moved the window.

When it helps, and when it misleads

Its strength is that it is the direct antidote to "pull harder": by showing the stock off its window, it makes escalation visibly the wrong move, and it gives every other mechanism a common picture on which to hang a floor, a stop rule, and a repair target. But a map is a model, and naming the wrong mediating stock — or mis-drawing the window — sends the whole response at the wrong reservoir; you then repair something that was never the constraint. Maps also silently go stale when structure shifts and no one re-draws them. The classic misuse is drawing the map to rationalize a tool already chosen, picking the mediating stock that flatters the intervention you wanted. The discipline is to fix the mediating stock from causal evidence and check the window against observed response, not against the lever you would prefer to pull.

How it implements the components

Stock-Flow Diagnostic Map realizes the diagnosis side of the machinery — the structural understanding the rest of the restoration is built on:

  • control_mediating_stock — its central act: naming the specific stock the lever actually acts through.
  • responsive_window_model — it characterizes the range over which that stock still transmits a lever change.
  • disabled_control_boundary — it marks the edge past which the lever-to-outcome coupling collapses.

It does not continuously watch the stock's level (stock_state_observability → Minimum Stock Floor Alert) or carry out the repair it points to (stock_repair_plan → Substrate Repair Protocol); the map is a diagnosis, not a sensor and not a cure.

References

[1] A liquidity trap is the standard name for the state in which the policy-rate lever stops transmitting because the mediating stock has left its responsive range — the textbook case of a lever that is not too weak but acting on a stock outside its window. ↩