Distributional Incidence Matrix¶
Template — instantiates Welfare Analysis and Distributional Effects Assessment
Tabulates each affected group against each benefit, cost, risk, right, or service dimension so incidence can be reviewed instead of inferred from aggregate totals.
The Distributional Incidence Matrix is a fixed tabular artifact — affected groups down the rows, welfare dimensions across the columns (benefit, cost, risk, right, service level) — whose whole discipline is that every cell must carry an entry. Its defining idea is that a blank cell is a finding: once "night-shift riders" is a row and "span of service hours" is a column, you cannot leave the intersection unanswered. The matrix is a structure for holding an entire allocation in one place where it can be read at a glance, not a method that computes a comparison and not a test that returns a verdict. It makes incidence legible; deciding whether that incidence is acceptable is another mechanism's job.
Example¶
A city redesigns its bus network from a coverage web into a high-frequency grid. The headline is clean: mean wait time falls. The planning team builds an incidence matrix before anyone celebrates. Rows: downtown commuters, low-income cul-de-sac neighborhoods, wheelchair users, night-shift workers, students, and one outer suburb that loses its single one-seat ride. Columns: wait time, walk-to-stop distance, one-transfer trips, span-of-service hours, fare. They fill every cell with a signed, sized entry — "+", "−", or a magnitude — using the best evidence each cell can afford.
Filling forces the awkward intersections into view. The outer suburb × walk distance cell turns strongly negative; the night-shift × span hours cell shows a service the average never noticed getting cut. Two rows are flagged as priority groups — wheelchair users and low-income neighborhoods — so their cells can't be washed out by the mean. The output is not a decision. It is a single legible picture showing that a real average improvement conceals two concentrated losses, now handed on to the guardrail and compensation steps that will judge and remedy them.
How it works¶
The template is defined by a few structural choices, not by an analytical procedure:
- Rows are the affected-party enumeration. Deciding who gets their own row is the map; a group with no row is invisible, so under-represented parties are added deliberately.
- Columns are the welfare dimensions. Benefit, cost, risk, right, and service-level are kept as separate columns so a gain on one cannot silently net out a loss on another.
- Fill-every-cell rule. Each intersection carries a sign and a magnitude (or an explicit "unknown"). An empty cell is treated as an open question, never as a zero.
- Priority rows are flagged. Vulnerable or protected groups get marked rows so they read as distinct lines, never folded into an aggregate.
- It is a shared, reviewable artifact. The value is a picture the whole room reads together, with per-cell provenance noted so a guess is not mistaken for a measurement.
Tuning parameters¶
- Row granularity — how finely groups are split. Finer rows surface concentrated harm but multiply effort and raise privacy risk on small groups.
- Column set — which welfare dimensions get their own column. More columns catch non-monetary effects (dignity, access) but dilute focus.
- Cell content — a qualitative sign versus a quantified estimate. Numbers look authoritative but invite false precision where evidence is thin.
- Priority flagging — which rows are marked protected. Marking more rows guards more groups but can blunt the signal if everyone is "priority."
- Authorship — analyst-filled versus co-filled with affected groups. Co-filling catches missing rows but slows the artifact down.
When it helps, and when it misleads¶
Its strength is that it forces the blank cell into the open and defeats aggregate masking on the first pass — you cannot table every group against every dimension and still honestly say "on average it improves." It is cheap, shared, and legible, which is exactly why it travels well into a review meeting.
Its failure mode is false completeness: a tidy, fully-populated matrix implies its rows and columns are the right ones, when the group that most needed a row may simply be absent, and a confidently-filled cell can be a guess in a fact's clothing. Its classic misuse is presenting the matrix as the whole assessment — "we ran a distributional analysis" — when it is only the ledger, with no judgment attached. The guarding discipline is to treat absent rows as findings the way empty cells are, and to keep per-cell provenance so economic incidence, not nominal incidence,[n1] is what the matrix records.
How it implements the components¶
The matrix realizes the mapping-and-legibility side of the archetype — the parts a static artifact can hold, not the parts that judge:
affected_party_map— the rows are the map; the template forces every party, including the usually-unlisted ones, onto a named line.benefit_burden_incidence_model— the cells allocate each welfare dimension to each party, recording where an effect lands rather than where it formally originates.priority_group_lens— flagged priority rows keep vulnerable groups visible as distinct lines that cannot be averaged away.
It does not test those allocations against a floor or proportionality limit (equity_guardrail_set, minimum_floor_constraint) — that verdict is Equity Guardrail Test — nor evaluate whether a remedy repairs a loss (compensation_and_mitigation_pathway), which is Compensation Adequacy Review.
Related¶
- Instantiates: Welfare Analysis and Distributional Effects Assessment — the matrix is the shared ledger the rest of the assessment reads from.
- Sibling mechanisms: Counterfactual Welfare Comparison · Subgroup Disaggregation Audit · Value-Weight Sensitivity Analysis · Compensation Adequacy Review · Externality and Spillover Inventory · Equity Guardrail Test · Public Reason Disclosure Protocol
Editorial Notes¶
Form Classification¶
Form family: Representation, Specification & Plan
Rationale: Distributional Incidence Matrix operates as a non-executable information artifact that externalizes static or prospective structure because it tabulates each affected group against each benefit, cost, risk, right, or service dimension so incidence can be reviewed instead of inferred from aggregate totals.
Independent corroboration: The frozen evidence defines Distributional Incidence Matrix as 'Tabulates each affected group against each benefit, cost, risk, right, or service dimension so incidence can be reviewed instead of inferred from aggregate totals', so its operative form is Representation, Specification & Plan.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Economics & Finance
Origin pattern: Single lineage
Present-day reach: Multi-domain
Rationale: Public economics cohered incidence analysis as tracing who actually bears costs and receives benefits rather than accepting their nominal assignment.
Related originating lineages:
- Public Administration & Policy — Distributional policy analysis operationalizes incidence across populations, rights, services, and risk dimensions.
Review resolution: Both current reviews place distributional_incidence_matrix primarily in economics_finance; the reconciled classification retains only lineages that materially shaped the mechanism and keeps breadth of origin separate from reach.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
The matrix tabulates groups you already know and gives each a row up front. Drilling into an aggregate to discover an unsuspected subgroup pattern is a different move — that is Subgroup Disaggregation Audit. The matrix is the frame; the audit is what you do when you suspect the frame is hiding something inside a row.
[n1] Economic incidence — who actually bears a cost or captures a benefit — often diverges from statutory or nominal incidence, who is formally assigned it. A charge levied on sellers can land on buyers through price; the matrix's cells are meant to record the economic landing point, not the nominal one. ↩