Technical Appendix

This appendix describes the data sources, analytical methods at a categorical level, and the known limitations behind the findings in this profile. The intent is to satisfy a careful reader — a municipal evaluator, a peer analyst, a Town Counsel reviewing for due diligence — that the report rests on defensible methodology and verifiable sources.

1. Scope and intent

The appendix covers data provenance, analytical framework, and limitations. Methodology specific enough to replicate the analysis is not within scope. The Parcenomics analytical methods that produce findings in this report — the canonical name layer that resolves name variants and matches buyers against voters, the residency tier classifier that assigns each parcel to a primary-resident, non-resident, or entity-held category, the four-category sale filter that produces the household-change figures, and the compliance-recovery candidate-identification methodology — are proprietary to Parcenomics.

This appendix describes the *what* and *where the data comes from*. The *how* in operational detail is the firm’s intellectual property. A reader who wants more detail on a specific finding can request a methodology consultation, described in Section 7.

2. Data sources

Federal

U.S. Census Bureau, American Community Survey. ACS 2020-2024 5-year estimates for Stockbridge, Berkshire County, and Massachusetts; ACS 2011-2015 5-year estimates for income-comparison vintages. Tables used: B01002 (median age), B19013 (median household income), B25003 (tenure / owner-occupancy), B25004 (vacancy reason — seasonal / recreational / occasional use), and B15003 (educational attainment).

U.S. Bureau of Labor Statistics, Consumer Price Index for All Urban Consumers (CPI-U). U.S. city average, annual series. Used to deflate nominal income figures across the 2015–2024 window to comparable purchasing-power terms.

Massachusetts state

Department of Revenue, Division of Local Services. Tax Rate Recapitulation forms FY2011-FY2026 for Stockbridge; LA-13 form (new growth detail); LA-19 Equalized Valuation publications across eight biennial cycles (FY2010 through FY2024); Schedule A annual local-receipts filings; Levy Limit form. The FY2026 Stockbridge Recap was DOR-certified on October 27, 2025.

Executive Office of Housing and Livable Communities. Seasonal Community designation list as published in December 2024 (original 25-municipality statutory designation) and December 2025 (formula-based expansion offering designation to 19 additional municipalities — 18 in the December 2025 announcement plus Rockport shortly thereafter).

Massachusetts Affordable Homes Act of 2024 (Chapter 150 of the Acts of 2024, codified at M.G.L. c.23B §32), Section 32 statutory text. Provides the statutory framework for the Seasonal Community designation, the discretionary and mandatory tools under Section 32(d) and 32(e), and the expanded §5C exemption ceiling under Section 32(f).

Massachusetts General Laws. Chapter 59 §5C (residential exemption); Chapter 59 §5 Clauses 41A (senior tax deferral), 41D (senior exemption, the 41C variant with cost-of-living adjustment), 17E (widow / elderly / minor exemption, the 17D variant with cost-of-living adjustment), and 18 (hardship abatement); Chapter 64G §3 (room occupancy excise, including the up-to-6% local-option rate under §3A) and §3D (short-term rental community impact fee under §3D(a), with the §3D(b) option to extend to owner-occupied two- and three-family STRs and the §3D(c) 35% affordable-housing / infrastructure earmark on receipts).

Local

Town of Stockbridge assessor records. Parcel inventory, ownership records, residency signals, and class codes, current as of the assessor data used in the FY2026 Recap reconciliation. Source for the residential AV figures used throughout the report.

Town of Stockbridge voter file. Snapshot dated April 21, 2026, restricted to active registrations with a known birth year for the voter age distribution analysis.

Town of Stockbridge sales register. Residential transactions covering FY2020 through FY2026, used in the buyer-to-voter conversion analysis and the four-category sale filter described in Section 4.

Town of Stockbridge Tax Rate Recapitulation. FY2026 form, certified by DOR October 27, 2025. Source for the FY2026 levy, residential rate, and class AV totals reported throughout the body of the report.

Regional

Berkshire Regional Planning Commission. Survey of Second Homeowners — Results (August 2022). 731-respondent survey conducted across the 30 Berkshire County towns (out of 32 total municipalities); Stockbridge was among the top-five towns by survey participation with 61 respondents. The full report is available at berkshireplanning.org. Used in Chapter 2 for the New York origin concentration and the cultural-amenities purchase-decision finding.

Reporting

The Berkshire Eagle. News reporting on the May 12, 2026 Great Barrington Proposition 2½ override defeat (“Great Barrington voters overwhelmingly reject Proposition 2 1/2 override,” May 13, 2026). Cited in Chapter 4’s political-surfacing paragraph.

Town of Stockbridge Capital Planning documentation. Fire station replacement capital estimates and operating-cost projections, cited in Chapter 3’s “next commitment in the pipeline” subsection. Public-source citation pending final publication.

3. The five-indicator framework

Chapter 4 presents five independent indicators of the structural condition the profile documents. Each operates on a distinct data source. Agreement across all five — none of which can be derived from the others — is what makes the pattern interpretive rather than coincidental.

Age structure. Federal census table B01002 (median age) for Stockbridge, Berkshire County, and the Massachusetts state aggregate. Cross-validated against the Stockbridge active voter file’s age distribution, which independently confirms the elder-skew finding through a different data source.

Seasonal housing share. Federal census table B25004 (housing units classified by vacancy reason, including seasonal / recreational / occasional use). Reported value for Stockbridge against the 22 Berkshire County peer averages (the subset of Berkshire’s 32 municipalities for which ACS estimates are available) and the Massachusetts state aggregate.

Income transformation. Federal census table B19013 (median household income), measured at two vintages (2011-2015 and 2020-2024). Comparison is presented both in current dollars and in CPI-U-adjusted purchasing-power terms across the same nine-year window for Stockbridge, Berkshire County, and Massachusetts.

Buyer-to-voter conversion. Matched comparison between the Town of Stockbridge sales register (buyers in residential transactions FY2020-FY2026) and the Town of Stockbridge voter file (active registrations as of April 21, 2026). The matching process handles common name variants — nicknames, spelling differences, middle-name discrepancies. Conversion is measured at the purchased property’s address rather than against any in-town registration. Entity buyers (LLCs and trusts) are included via situs-match: a parcel purchased by an entity is counted as converted if a current voter is registered at that parcel’s address.

Peer positioning. Same data sources (B01002, B25004) applied across the 22 Berkshire County municipalities for which ACS estimates at the necessary table level are available (of 32 total), producing the comparative scatter and ranking documented in Chapter 4.

The matching, classification, and reconciliation methodologies that produce the buyer-to-voter conversion measure are described at the categorical level. The operational specifics of the canonical name layer that handles variant resolution and the override logic that processes ambiguous cases are Parcenomics methodology.

4. Four-category sale filter and implied newcomer income

Chapter 1’s replacement math and the approximately $268,000 implied newcomer income figure rest on a four-category classification of residential property sales across FY2020-FY2026.

Every residential sale in the window is sorted by the residency status of the seller at time of sale and the residency status of the buyer post-sale. The four categories are: Resident-to-Resident, Resident-to-Non-resident, Non-resident-to-Resident, and Non-resident-to-Non-resident. The net household change is the difference between the third category (households joining the year-round community through purchases) and the second category (households departing). For FY2020-FY2026 in Stockbridge, 234 households arrived and 244 departed — a net change of negative 10.

The implied newcomer income figure is then derived from the observed Stockbridge median household income shift, the statewide income-growth pace as a counterfactual baseline, and the four-category sale count.

Technical framing. Let Y₂₀₂₄ denote the observed 2024 Stockbridge median household income and Y₂₀₁₅ the observed 2015 figure. The 2024 figure is decomposable into the population-weighted contribution of three groups: households that remained in Stockbridge across the window (N_S households at mean 2024 income Y_S,2024), households that arrived through purchases (N_A at mean income Y_A), and households that departed (N_D at mean pre-departure income Y_D). The accounting identity is:

The four-category sale filter produces N_A=234 and N_D=244 directly. The Stockbridge median household incomes Y₂₀₁₅=$56,027 and Y₂₀₂₄=$146,250 are observed from ACS table B19013. Under the counterfactual that staying households tracked the statewide income-growth pace g_MA=1.52 in current dollars across the same window, we have Y_S,2024=g_MA·Y₂₀₁₅. The arriving and departing groups are assumed to have approximately equal pre-purchase income distributions absent evidence otherwise, so Y_D≈Y₂₀₁₅. The staying-household count N_S is the residual: the 2024 ACS reports approximately 1,051 Stockbridge households, against N_A−N_D=−10 net migration, implying N_S≈827 households that remained across the window. Solving the identity for Y_A produces the approximately $268,000 figure cited in the body of the report — in 2024 dollars, directly comparable to the 2024 Stockbridge median.

The figure is reported as approximate because the identity is sensitive to the assumed mean income of the departing-household cohort. The Chapter 1 prose carries this caveat through the rounding convention (“approximately $268,000”; “roughly 3.15 times what departing households earned”).

The residency-status determination for each party in each sale uses the residency tier classifier described in Section 5. The operational specifics — class codes used in the classification universe, override logic, edge-case handling, the residency tier classifier’s internal decision boundaries — are Parcenomics methodology.

5. Residency tier classifier

The 39.1% primary-resident / 59.7% non-resident / 1.3% entity-held split reported in Chapter 2 is produced by what the body chapters refer to as the residency tier classifier with manual overrides.

The classifier accepts as input each residential parcel in the Stockbridge assessor universe. For each parcel, it produces an assignment to one of three tiers: primary-resident-owned, non-resident-owned, or entity-held (LLC or trust with ownership not traceable to an individual from public records). The classifier operates by triangulation across multiple data sources, with a manual override layer applied where automated classification produces an ambiguous or known-incorrect result.

Technical framing. For each residential parcel p in the assessor universe, the tier assignment T(p)∈{primary, non-resident, entity} is produced by a multi-source triangulation function:

where O(p) is the ownership record (owner name, owner type, entity flag), V(p,addr(p)) is the voter-file lookup at the parcel’s situs address, M(p) is the parcel’s mailing-address pattern relative to its situs address, C(p) is the assessor class code and any class-specific residency signal it carries, and Ω(p) is the manual override layer applied where the automated triangulation produces an ambiguous or known-incorrect tier assignment. The function f(·) implements the triangulation logic — assigning relative weights to consistent and inconsistent signals across the input vector, resolving conflicts where signals disagree, and producing the final tier assignment.

The specification of f(·), the conflict-resolution logic across inconsistent signal combinations, the override decision criteria, and the calibration of the classifier against ground-truth samples constitute Parcenomics methodology and are not disclosed in this appendix. The framework above describes what the classifier consumes and what it produces; the operational specifics are the firm’s intellectual property.

Cross-validation: an independent USPS-routing-corrected mailing-only methodology produces a primary-resident share within five percentage points of the classifier’s output, providing convergent validation of the headline figure.

6. Reconciliation and known limitations

Honest disclosure of the points at which the analysis has known limitations.

Federal census uncertainty for small towns. Stockbridge’s population is approximately 2,056. American Community Survey 5-year estimates carry meaningful margins of error at this population scale — typically wider for small-town subdivisions than for county- or state-level aggregates. The Chapter 1 prose acknowledges this caveat where each ACS-derived figure is cited; specific margin-of-error data is available from the Census Bureau alongside each estimate.

The $7.6 million residential AV reconciliation gap. Two residential AV totals appear in the report depending on the analytical purpose. The Recap LA-4 Class 1 line reports the certified residential AV total of $1,361,988,320 across 1,697 residential parcels — the figure used for fiscal aggregates and tax-rate calculations. The parcel-level aggregation used as the ownership-share denominator is restricted to assessor class codes in the 101-109 range and applies exclusions for Chapter 61 / 61A / 61B classified land at residential-edge treatment and for parcels in the Multi-use Residential class series (classes 012-043). The restricted aggregation produces $1,354,342,520 across 1,696 parcels. The two denominators reconcile through the excluded categories — the gap is approximately $7.6 million and 1 parcel in aggregate, reflecting the net effect of class-code-based exclusions applied to the restricted aggregation rather than a single $7.6M parcel. Both denominators are correct for their respective analytical purposes: the Recap line for fiscal aggregates and tax-rate calculations, the restricted parcel aggregation for ownership-share analysis where Multi-use and Chapter-61-edge parcels would distort residency-tier assignment.

The 114.1% implied assessment-to-sale ratio. Stockbridge’s FY2026 CAMA total ($1,495,161,601) measured against the most-recently-published Equalized Valuation (the FY2024 EQV of $1,311 million) produces an implied 114.1% ratio. This sits outside the DOR’s 90-110% certification band but is consistent with the biennial EQV measurement lag rather than a local calibration failure. The FY2026 EQV, scheduled for late 2026 publication by DOR, will substantially close the gap by reflecting two additional cycles of actual sales activity.

Within-class regressivity in Stockbridge specifically. Chapter 2’s discussion of within-class assessment regressivity cites published peer-reviewed research from the Federal Reserve Bank of Philadelphia, the University of Chicago, the Lincoln Institute of Land Policy, and others. Stockbridge-specific empirical confirmation would require a price-stratified ratio study using local MLS sales records matched against assessed values — a methodology described in Chapter 2 Note 11. Such a study has not been performed for Stockbridge and is not within the scope of this profile.

Buyer-to-voter match completeness. The canonical name layer handles the most common name variants encountered in Massachusetts assessor and voter data. It does not handle every conceivable variant; some matches are missed (and some are caught at the override layer when hand-reviewed). The reported buyer-to-voter conversion rate is therefore a conservative estimate — the true conversion rate is plausibly slightly higher than the figures presented. Chapter 1’s “29.5% overall / 25.3% recent half” figures are floors rather than point estimates.

Bounded limitations from data not yet ingested. Three signals are referenced in the body of the report as not yet incorporated into the classifier and would tighten the residency-tier assignments if available: (1) per-parcel personal-property-tax linkage from assessor working files, which would resurface the clean second-home signal currently disabled in the classifier; (2) LLC manager and trust beneficiary identification from Massachusetts Secretary of State filings and the registry of deeds, which would resolve the 100 currently entity-held residential parcels ($17 million in assessed value) into primary or non-resident categories; (3) voter rolls from Lenox, Lee, Pittsfield, and Williamstown, which would tighten the cross-town voter signal currently limited to Great Barrington. These are documented improvements available through subsequent engagements; they do not affect the figures reported in this profile, which use the data currently available.

Compliance-recovery gap estimation. The $0.5M-$1.7M annual compliance-recovery gap cited in the executive summary and Chapter 5 represents a methodologically defensible range across four independent revenue streams: personal property tax on second-home furnishings, vehicle excise on Stockbridge-garaged vehicles registered elsewhere, boat excise on Stockbridge-moored vessels registered out of state, and the M.G.L. c.64G §3D short-term-rental community impact fee (statutory authority unused by Stockbridge as of FY2026). The per-stream low and high bounds reported in Chapter 5 Section A1 sum to $468K and $1,687K respectively; the rounded range $0.5M-$1.7M is the figure used in the executive summary and Chapter 5 prose.

Technical framing of the aggregation. For each stream s∈{1,2,3,4}, the recovery range is estimated as an interval [L_s, U_s] where the bounds reflect conservative and aggressive recovery assumptions calibrated against regional comparables and per-stream enforcement realities. The aggregate annual range across streams is:

The within-stream estimation methodology — the comparable-regional data sources used to anchor the bounds, the calibration of conservative and aggressive recovery assumptions, the prior distributions informed by per-stream enforcement-capacity considerations, and the per-parcel candidate identification — is Parcenomics methodology and not disclosed in this appendix. The framework above describes the structure within which the bounds are produced.

The actual recovery any town would achieve depends on local enforcement capacity, the Select Board and Town Meeting’s appetite for outreach, and political conditions specific to the engagement window. The range described is a defensible estimate, not a guaranteed outcome.

7. Reproducibility and request-for-detail process

Every external data source cited in this report is publicly accessible. A peer analyst with appropriate domain knowledge can verify each source independently:

• American Community Survey data: data.census.gov
• DOR Recap, LA-13, LA-19, Schedule A, and Levy Limit forms: dlsgateway.dor.state.ma.us
• Massachusetts General Laws and session laws (including the 2024 Affordable Homes Act): malegislature.gov
• Executive Office of Housing and Livable Communities publications, including the Seasonal Community designation list: mass.gov
• Town of Stockbridge records (assessor data, voter file, sales register, Recap): publicly available through the Town’s public-records-request process
• Berkshire Regional Planning Commission Survey of Second Homeowners (2022): berkshireplanning.org

For methodology questions beyond the categorical descriptions in this appendix, Parcenomics offers a scoped methodology consultation as part of engagement scoping. The consultation produces a written methodology disclosure scoped to the specific finding the requester is evaluating, with enough operational detail for a peer analyst to assess defensibility without exposing the full proprietary specification.

8. Versioning

This profile is the inaugural Stockbridge release in a planned Parcenomics municipal-profile series, dated May 2026. Subsequent profiles, covering additional Massachusetts municipalities, will follow as the series expands. The methodology framework described in this appendix is portable across municipalities and will form the analytical backbone of subsequent profiles. Any future revision of this Stockbridge profile will be re-dated and re-versioned.