White paper · Part 2

Aprize Metrics: From theory to data

Part 2 operationalizes the framework: a multi-sector, regionally adaptive normalization method with sector-specific Local Spend Share benchmarks across 20 top-level NAICS codes and a piecewise-linear transformation to a Silver-tier baseline.

Download PDF ← Part 1: The economic rationale
Aprize
Published · 7 min read

Where Part 1 Left Off

Part 1 of this series, The Economic Rationale, established why Aprize measures local economic impact by directly observing transactions rather than estimating them. We introduced two core metrics drawn from observed spending data: Local Spend Share (LSS), the share of an initial outlay that stays in the defined local economy on the first round; and Observed Local Multiplier (OLM), the cumulative additional retention from observed downstream re-spending. We then proposed a Composite Local Impact Score (CLIS) that combines the two into a single 0-to-100 rating mapped to a Diamond / Platinum / Gold / Silver / Bronze / Unrated tier.

Part 2 turns from theory to data. It addresses three operational questions Part 1 deferred: (1) how do we make LSS and OLM comparable across very different industries and across very different definitions of local; (2) what mathematical transformation produces the CLIS; and (3) what data-collection rules govern measurement, eligibility, and re-rating in production.

A Multi-Sector, Regionally Adaptive Method

Aprize's CLIS is rooted in real spending data and is built from the two metrics introduced in Part 1:

  • Local Spend Share (LSS): the share of the initial outlay that remains in the local economy.
  • Observed Local Multiplier (OLM): the cumulative additional retention as further rounds of spending stay local.

To enable fair comparisons across both diverse industries (top-level NAICS) and competing geographic definitions of local (county vs. state), we normalize LSS by anchoring each observation to a sector- and geography-specific benchmark average. This addresses what we call the cold-start problem: at the moment a firm is first measured we typically lack granular data on its peers' purchasing patterns, so the benchmark substitutes a defensible prior until enough real transactions accumulate to refine it.

Below we present a reference table of scalar average LSS values - one for a typical county-level definition of local (e.g., Grant County, Indiana) and one for a state-level definition (Indiana). These benchmarks align with published regional input-output studies where available, and are otherwise derived via Aprize's structured-reasoning workflow when public data are sparse. They are starting points, not final values: every benchmark is designed to be replaced by observed peer means as the panel of measured firms grows.

Local Spend Share (LSS) Benchmarks by Top-Level NAICS

All values are expressed as percentages (0-100). County and state columns are independent benchmarks, not a hierarchy.

NAICS Sector County avg LSS (%) State avg LSS (%) Reasoning / reference
11 Agriculture, Forestry, Fishing & Hunting 25 45 Most off-farm inputs (fertilizer, equipment, fuel) are sourced regionally or nationally; land, labor, and direct-to-buyer sales retain locally. State-level retention rises with in-state processing.
21 Mining, Quarrying, Oil & Gas Extraction 20 40 Capital-intensive with concentrated national vendors; royalties, severance taxes, and labor are the main local channels.
22 Utilities 30 55 Generation fuel and bulk equipment are typically out-of-area; transmission, distribution labor, and franchise fees stay local.
23 Construction 55 75 Labor-intensive with a heavy local-subcontractor footprint; structural materials are the principal leakage driver. Civic Economics & ILSR field studies.
31-33 Manufacturing 25 45 National/global supply chains for raw materials and components dominate; wages and a fraction of MRO services retain locally.
42 Wholesale Trade 20 35 Pass-through margins to upstream national producers; warehousing labor and transport contracts are the main local share.
44-45 Retail Trade 35 60 Cost-of-goods-sold flows out to national distributors; store-level wages, rent, and services retain locally. AMIBA Local Premium studies place local-independent retail meaningfully above chains.
48-49 Transportation & Warehousing 35 55 Fuel and equipment leakage offset by local payroll, terminal services, and maintenance.
51 Information 20 35 Software, content licensing, and cloud infrastructure are predominantly extra-local; editorial and support payroll retain.
52 Finance & Insurance 25 45 Interest income and reinsurance flow to national balance sheets; branch payroll, local underwriting, and small-business lending retain.
53 Real Estate, Rental & Leasing 50 70 Rents, property taxes, and on-site services are inherently local; institutional ownership is the primary leakage vector.
54 Professional, Scientific & Technical Services 55 70 Largely labor; software tooling and national subcontractors are the main leakage.
55 Management of Companies & Enterprises 20 35 Corporate-headquarters function; profits frequently routed to parent jurisdictions.
56 Administrative & Support / Waste Management 55 70 Janitorial, security, landscaping, and waste hauling are predominantly locally staffed and locally sub-contracted.
61 Educational Services 65 80 Wage-heavy with regional vendor base; textbooks, technology, and capital construction are the main outflows.
62 Health Care & Social Assistance 60 75 Provider payroll dominates spend; pharmaceuticals, medical devices, and major equipment are leakage. IBRC regional input-output models.
71 Arts, Entertainment & Recreation 55 70 Performer fees and venue payroll retain; touring acts and licensing flow out.
72 Accommodation & Food Services 50 70 Wage-intensive with local food sourcing where available; franchise fees and national distributor procurement are the main leakage drivers.
81 Other Services (except Public Administration) 65 80 Personal services, repair, and civic/religious organizations are deeply local in their cost structure.
92 Public Administration 55 70 Government payroll retains locally; state/federal program contracts and capital projects can leak materially.

Note: rows without a direct citation reflect Aprize's structured reasoning based on patterns observed in adjacent, better-studied sectors. Every benchmark is replaced by the observed peer mean as soon as the measured panel for that NAICS / geography pair reaches the minimum-sample threshold (currently n = 12 verified firms).

Transformation Formula

Our target rating for an average firm with no downstream observed spending - that is, a firm whose LSS equals the sector benchmark and whose OLM equals its LSS - is 65, the lower bound of the Silver tier. Either being above the sector benchmark, or enabling additional downstream observed spending, is required to reach Gold or above.

To achieve this, we define a piecewise-linear transformation. Let x be the observed metric (LSS or OLM, expressed on a 0-100 scale) for a firm, and let mu be the scalar benchmark average from the NAICS / geography table above. We define the normalized score f(x) as:

f(x) = 65 · (x / mu), when x ≤ mu

f(x) = 65 + 35 · (x − mu) / (100 − mu), when x > mu

This produces a continuous mapping such that LSS values between 0 and mu are linearly mapped to scores between 0 and 65, and LSS values between mu and 100 are linearly mapped to scores between 65 and 100. The function is exact at the two anchor points (f(0) = 0, f(mu) = 65, f(100) = 100) and is continuous at x = mu.

To start, we apply the same transformation to OLM, and we weight OLM and LSS equally in the CLIS formula:

CLIS = 0.5 · f(OLM) + 0.5 · f(LSS)

This guarantees that an average firm with no downstream observed spending - one whose LSS equals the sector benchmark and whose OLM equals its LSS - receives a CLIS of 65 and lands exactly at the Silver threshold. Firms that out-perform the benchmark on either dimension, or whose downstream partners are themselves Aprize-rated and contribute to OLM, move into Gold and above.

Conclusion

By combining:

  • sector-specific average LSS values for both a county and a state definition of local, and
  • a piecewise-linear mapping anchored at the benchmark mean,

we establish a fair, adaptive normalization for LSS and OLM. This approach tackles the cold-start challenge - where we rely on tabled benchmarks until real transaction data accumulates - and ensures that a typical performer (one whose LSS equals the sector average) with no downstream observed spending maps to a CLIS of 65, or the Silver tier.

Over time, as organizations submit actual data, these initial references (whether published or model-derived) are continuously refined toward observed peer means. The end result is a transparent, consistent scoring system that rewards higher retention of local dollars and guides local governments, nonprofits, and businesses toward greater local economic impact.

Other Data-Processing & Measurement Details

Self-declaration and verification

  • Firms self-declare their NAICS code; Aprize verifies it as part of onboarding.
  • Firms self-declare any related entities; declared relationships may be audited.
  • Transactions with related entities are excluded from the measurement set, unless the related entity is also Aprize-verified, in which case the related entities are treated as a single synthetic firm: intercompany transfers are netted out and the combined external transaction footprint is what is measured.
  • Worked example: Company A has Shell B. If Shell B is also Aprize-rated, transfers between A and B are netted and the score is computed against the consolidated external expenditure of A + B.

Measurement window

  • By default, we measure over a trailing twelve-month period.
  • Firms are eligible to be re-rated quarterly when (a) their contract is current and (b) updated transaction data has been provided for the new measurement window.

Numerator and denominator

  • Numerator: the share of the denominator that falls within the specified geography.
  • Denominator: all observed transactions within the measurement window.

Things to note

  • Profits, unless reported as an observable line item, are excluded from the measurement set.
  • For grant-monitoring engagements we wait until the contracted period is complete and use the total grant amount as the denominator.
  • The LSS averages assume wage bills are observed and mapped to the worker's place of residence.

Selected References