Appendix B - License to Pour: Access & Equity in Massachusetts Liquor Licensing Report

Quantitative Data

Appendix B follows a fixed sequence: what data DLM started with (B.1), how observations were classified (B.2), what measures were built from them (B.3), what analyses those measures supported (B.4), and what limits attach to the results (B.5). Each methodological decision is stated once, in the section where it belongs, and cross-referenced elsewhere. Results appear in Findings 1, 2, 3, and 4.

B.1 Data Sources and Dataset Construct

Data Sources

Below are the data sources to support the quantitative analysis.

Dataset	Source	Reference Period	Principal Use
ABCC fiscal year 2024 annual quota reports	ABCC	Fiscal year 2024 licensing cycle. Municipality-level counts as reported by each LLA to ABCC	Allocated (allowed), issued, exception, and available counts by license category and license period. Source of every allocation, utilization, and availability figure in Findings 1, 2, and 3
ABCC fiscal year 2024 active-license extract	ABCC	Extract compiled by DLM as of December 5, 2025 (fiscal year 2024 licensing cycle)	The 8,841 license-level on-premises records: license number, type, class, status, business name and address, the special legislation flag, and the authorizing chapter and act
Municipal-level socioeconomic indicators	Division of Local Services, Department of Revenue	Fiscal year 2023, fiscal year 2026 (Department of Revenue uses 2023 population)	Population, population density, EQV
Decennial census population	US Census Bureau	1930–2020 decennial counts	Statutory formula denominator; historical allocation reconstruction
County Business Patterns	US Census Bureau	2022	Full-service restaurant counts for interstate comparison (Figure 33)
Stakeholder survey	DLM	December 2025–February 2026 (n = 260 cleaned)	Appendix C/Finding qualitative corroboration
Stakeholder interviews	DLM	2026	Process, cost, and information-environment evidence (Appendix C.4)

Data Preparation and Validation 

Municipality names were standardized to a single case and hyphenation convention before matching, which resolves the Manchester-by-the-Sea variant and the North/South/East/West compounds. Currency and percentage fields stored as text were coerced to numeric. Records without a municipality value, and county-subdivision records that are not municipalities, were dropped. Utilization values recorded as proportions in some fields and percentages in others were normalized to a single percentage scale.

Licensing records were matched to municipal characteristics on the standardized name. All 351 municipalities matched, and no imputation was required for any variable used in the analysis: Population, EQV per capita, allocated licenses, issued licenses, and the special legislation flag are complete. One entity in the extract, Devens, carries active licenses but is not a municipality; it is a regional enterprise zone administered by MassDevelopment across parts of Ayer, Harvard, and Shirley, with its own licensing authority. It is excluded from every municipal analysis, which is why recomputing a statewide figure from the raw extract will not always match the figure reported here.

No licenses were combined. Section 12 provides that when a licensee surrenders a license and receives a different form of license in its place during the same license year, the two count as one license for § 17 purposes. Because the ABCC extract records only licenses active on the extract date, a replaced license appears once, and no adjustment was needed; each active license number was counted separately, so licenses held by the same owner at separate premises are counted separately. Issued, allocated, exception, and available counts were then checked for arithmetic consistency municipality by municipality, and every value that failed the check was traced to a source record.

Analytical Samples

DLM’s dataset covers all 351 municipalities. Three analytical samples are drawn from it, and every figure in the report is reported against the sample on which it was computed.

Twenty-three municipalities are excluded from the scope of this study due to data constraints, dry license status, and license category-specific local prohibition. A municipality for which no data is available or which has voted not to allow on-premises sale has no licensing outcome to measure; carrying its zero forward would depress every per-capita and utilization figure for a reason unrelated to the quota. Retaining them as zeros was considered and rejected on that basis. The consequence is that results describe municipalities that participate in on-premises licensing and cannot be read as statewide averages across all 351 municipalities.

Classification	Number	Definition	Population246
Quota municipalities	302	Subject to state-imposed population-based quota limits on MGL c. 138, § 12, on-premises licenses	6,081,934
Local-control municipalities	26	Exempt from state-imposed population-based quota limits on MGL c. 138, § 12, on-premises licenses	872,735
Other	23	Excluded from the scope of this study	46,730

B.2 Classification of Municipalities and Licenses

Municipality Classification

Quota-bound (n = 302).Municipalities that grant § 12 on-premises licenses are subject to the § 17 population formula. This is the population the quota actually governs, and the sample used wherever the report describes how the quota operates.Boston is expressly excepted from the population formula and operates under fixed statutory numbers set in § 17 and amended only by special act. It is nonetheless retained in the quota sample because it operates under a state-imposed ceiling amendable only by the Legislature, which is the characteristic the analysis is about. Treating it as its own category was considered and rejected: A single-municipality category supports no inference, and removing the Commonwealth’s largest licensing market would understate the system it describes. Boston is an outlier in several analyses, and sensitivity checks are reported in B.4.

Local control (n = 26).Municipalities that grant § 12 on-premises licenses “in addition to, and irrespective of” the § 17 limits, having accepted §§ 17A–17C during the 1978–1984 window. They are the comparison group for Finding 2 and the basis for the hypothetical quota allocation in Finding 2.

Other (n = 23).The “Other” municipalities excluded from the scope of this study include (1) 14 municipalities for which no data was provided by ABCC: Florida, Heath, Leyden, Middlefield, Monroe, Monterey, New Braintree, Pelham, Peru, Rowe, Royalston, Tolland, Warwick, and Washington; (2) eight dry municipalities: Alford, Chilmark, Dunstable, Gosnold, Hawley, Montgomery, Mount Washington, and Westhampton; and (3) Milton, which maintains a local prohibition specifically on All-Alcohol licenses while permitting other license categories. 

License Classification

Section 17 allocates two separate categories with separate formulas and separate floors: All-Alcohol licenses and Wine-and-Malt licenses. Wine-and-Malt with a cordials permit is treated as Wine-and-Malt throughout because the cordials permit is an endorsement on a Wine-and-Malt license rather than a distinct allocation under § 17. Wine-only and Malt-only licenses, which together account for 2% of on-premises licenses, are treated the same way.

Whether the two categories are combined depends on what is being measured, and the rule is consistent throughout the report. Allocation and utilization are always computed category by category because the statute allocates that way, and a municipality can be exhausted on All-Alcohol while holding unused Wine-and-Malt capacity. Available licenses are therefore computed within each category and floored at zero before summing, so a surplus in one category cannot mask a deficit in the other. In DLM’s dataset, this is for each category: Issued without special legislation = Total Issued − Total Exception; Available = max(0, Allowed − Issued without special legislation); and Total available = All-Alcohol available + Wine-and-Malt available. This study does not compute total allowed minus total issued across categories, as that nets an All-Alcohol overage against Wine-and-Malt leftovers and understates unused capacity (for example, Webster is at 100% combined utilization but still has 3 unused Wine-and-Malt licenses). License access, by contrast, combines the categories because it measures what a resident or a prospective licensee encounters, and a customer does not experience the two tracks as separate systems. 

The combined measure is conservative. Because it treats unused Wine-and-Malt capacity as available supply, it understates the constraint facing a municipality that is fully blocked on All-Alcohol, which is the commercially decisive license for most full-service restaurants.

Seasonal § 12 licenses are granted “in addition to and irrespective of” the § 17 limitation and are therefore excluded from allocation and utilization. They are counted in license access, which includes every active on-premises license. 

Special Legislation and Exception Licenses

Special legislation licenses are identified from the flag carried on each record in the ABCC fiscal year 2024 active-license extract, together with the authorizing chapter and act. The municipality-level “total exception” field in DLM’s dataset (compiled from the ABCC fiscal year 2024 annual quota reports) is broader—it also captures seasonal and veterans’ licenses—and using it would double the municipality count. Every special legislation figure in this report comes from the record-level flag.

These licenses sit outside the § 17 count by operation of the special act that created them, and they are excluded from both the numerator and the denominator of utilization. Including them would obscure the constraint being measured: Boston has received several hundred licenses by special act, and counting those against the formula would make a heavily constrained city appear to have spare capacity. They are counted in license access, which is the all-license measure. A municipality holding more licenses than its formula allocation is therefore not an error in the data; it is the visible trace of relief already obtained.

Two exclusions are applied to the special legislation analysis specifically. Off-premises licenses are excluded because the report analyzes on-premises licensing only. Licenses held by local-control municipalities under acts passed after 1984 are excluded because those municipalities were no longer subject to the quota when the act passed, and the license cannot be read as relief from a quota. Licenses those municipalities obtained before the 1978–1984 window are retained because they were quota-bound at the time. The flag identifies 719 licenses across 101 municipalities and Devens, an enterprise zone; after these exclusions, there are 716 across 100 municipalities statewide and 706 across 97 municipalities within the outcome‑based sample.

Other statutory exceptions—war veterans’ organizations licensed under § 12, restricted airport licenses, and § 19 farmer-series pouring permits—are outside the § 17 count by statute and are treated the same way as seasonal licenses: excluded from allocation and utilization, counted in access.

License access answers how many on-premises pouring rights exist per resident—what municipalities, residents, and prospective licensees encounter—so exception licenses are counted. Utilization answers how tight the § 17 formula is, so the same exception licenses are removed from the utilization numerator (and do not expand “available” formula capacity). The two measures are deliberately paired: Finding 1 isolates formula pressure; Findings 2 and 3 measure total issued access, including relief already obtained through special legislation and other statutory exceptions.

B.3 Construction of Quotas and Analytical Measures

This section is the dictionary for the quantitative analysis. After reading it, a reader should know exactly what each measure in the findings means and how it was built.

Statutory License Allocation

Section 17 allocates one All-Alcohol license under § 12 for each population unit of 1,000 or additional fraction thereof, plus one further All-Alcohol license for each population unit of 10,000 or fraction thereof above the first 25,000, but not fewer than 14 licenses regardless of population. It separately allocates one Wine-and-Malt license for each population unit of 5,000 or a fraction thereof, but not fewer than five, regardless of population.

Two features of this language govern every allocation figure in the report and are commonly misstated. The 14 and the 5 are floors, not thresholds: The per-1,000 and per-5,000 increments run from the first resident, and the floor applies only where the formula would otherwise yield less. And the 10,000 increment applies only to the population above the first 25,000, not to the total population. A quota established under the section is never reduced because of a loss in population, which is why several municipalities that have shrunk since their peak now show utilization at or above 100%.

Boston is excepted from this formula and operates under fixed statutory numbers. Its allocation is taken from the statute as amended rather than computed.

All formulas used in the report are collected here. Unless a passage is reconstructing a historical statutory allotment, P denotes 2020 Census population when reconstructing statutory allotments and 2023 Department of Revenue/Division of Local Services municipal population estimates for per-resident rates and the local-control comparison in Finding 2; ceil denotes the smallest integer not less than its argument.

Quantity	Formula	Notes
All-Alcohol allocation, § 12	A_aa = max(ceil(P/1000) + ceil(max(P−25000,0)/10000), 14)	Floor of 14 applies regardless of population; second term applies only above 25,000
Wine-and-Malt allocation, § 12	A_wm = max(ceil(P/5000), 5)	Floor of 5
Total on-premises allocation	A = A_aa + A_wm	Boston excepted
Off-premises All-Alcohol, § 15	max(ceil(P/5000), 2)	Not analyzed in this report; stated for completeness
License utilization	U = I_f / A × 100	I_f = licenses issued on the formula categories, excluding those granted outside § 17
Available licenses	V = max (A_aa − I_aa, 0) + max (A_wm − I_wm, 0)	Computed within track, then summed, so surplus in one track cannot mask a deficit in the other
License access	L = I_total / P × 10000	I_total includes licenses granted outside the formula
Hypothetical quota allocation	A_cf = A(P) applied to a local-control municipality	Finding 2
Wealth quartile	Quartile of EQV per capita on the combined 328-municipality sample	Boundaries fixed across analyses
Epsilon-squared effect size	ε² = (H − k + 1) / (n − k)	H = Kruskal-Wallis statistic, k = groups, n = total observations
Rank-biserial correlation	r = 2U/(n₁n₂) − 1	Effect size for the Mann-Whitney comparison

License Supply and Access Variables

Variable	Construction
Allocated licenses	ceil(P/1000) + ceil(max(P−25000,0)/10000), floored at 14, plus ceil(P/5000) floored at 5; P denotes 2020 population when reconstructing statutory allotments and 2023 Department of Revenue/Division of Local Services population for per-resident rates and hypothetical quota allocations.
Issued (formula track)	Total on-premises issued minus licenses granted outside the § 17 formula
License utilization	Issued (formula track) ÷ allocated, expressed as a percentage
Available licenses	Allocated minus issued (formula track), floored at zero within each beverage track
License access	All on-premises licenses issued, including those granted outside the formula, per 10,000 residents
Wealth quartile	Quartile of EQV per capita computed on the combined 328-municipality sample
Hypothetical quota allocation	The § 17 formula applied to a local-control municipality’s population
Special legislation count	Count of active on-premises licenses flagged as authorized by special legislation in ABCC records

License Supply and Access Variables

Variable	Construction
Allocated licenses	ceil(P/1000) + ceil(max(P−25000,0)/10000), floored at 14, plus ceil(P/5000) floored at 5; P denotes 2020 population when reconstructing statutory allotments and 2023 Department of Revenue/Division of Local Services population for per-resident rates and hypothetical quota allocations.
Issued (formula track)	Total on-premises issued minus licenses granted outside the § 17 formula
License utilization	Issued (formula track) ÷ allocated, expressed as a percentage
Available licenses	Allocated minus issued (formula track), floored at zero within each beverage track
License access	All on-premises licenses issued, including those granted outside the formula, per 10,000 residents
Wealth quartile	Quartile of EQV per capita computed on the combined 328-municipality sample
Hypothetical quota allocation	The § 17 formula applied to a local-control municipality’s population
Special legislation count	Count of active on-premises licenses flagged as authorized by special legislation in ABCC records

License Supply and Access Variables

Variable	Construction
Allocated licenses	ceil(P/1000) + ceil(max(P−25000,0)/10000), floored at 14, plus ceil(P/5000) floored at 5; P denotes 2020 population when reconstructing statutory allotments and 2023 Department of Revenue/Division of Local Services population for per-resident rates and hypothetical quota allocations.
Issued (formula track)	Total on-premises issued minus licenses granted outside the § 17 formula
License utilization	Issued (formula track) ÷ allocated, expressed as a percentage
Available licenses	Allocated minus issued (formula track), floored at zero within each beverage track
License access	All on-premises licenses issued, including those granted outside the formula, per 10,000 residents
Wealth quartile	Quartile of EQV per capita computed on the combined 328-municipality sample
Hypothetical quota allocation	The § 17 formula applied to a local-control municipality’s population
Special legislation count	Count of active on-premises licenses flagged as authorized by special legislation in ABCC records

License counts are expressed per 10,000 residents throughout. Per 1,000—the unit the statute uses—was considered and rejected for presentation because it produces values below three for most municipalities and obscures the differences the report is describing. The choice affects presentation only. No population weighting is applied to municipality-level averages: Each municipality is one observation because the unit of policy is the municipality.

Quota Utilization and Bands

Utilization is issued licenses on the formula track divided by allocated licenses, expressed as a percentage; available capacity is the remainder, computed within the category and floored at zero. Formally, for All-Alcohol (variable aa) and Wine-And-Malt (variable wm) separately:

Issued_wo = Issued - Exception; Available_track = max(0,Allowed - Issued_wo)

Total available = Available_aa + Available_wm; Utilization (combined) = (Issued_wo_aa + Issued_wo_wm) ÷ (Allowed_aa + Allowed_wm).
 

These match DLM dataset fields All_alcohol_Total Available, WM/WMC_Total Available, and total_on_prem_util_w/o_exception. 

A municipality is classified as constrained when utilization is at or above 80%. The 64 constrained municipalities have a combined population of exactly 2,531,376 (of 6,081,934 residents in the 302 quota municipalities, or 41.6%). Of the 263 track-floored available licenses in constrained municipalities, 255 are in the high band (80–99%), and 8 are in the at/over band (≥100%). The at/over unused licenses are Wine-and-Malt leftovers in municipalities that are at or over on the combined (or All-Alcohol) measure.

The 80% threshold is a judgment, not a statutory category, and the report states it wherever the constrained count appears. It was chosen because, at that point, a municipality has a small enough remaining margin that an ordinary sequence of applications can exhaust it within a licensing cycle. Thresholds of 70%, 75%, 85%, and 90% were computed as well; the sensitivity is reported in B.4. The substantive conclusions of Finding 1 hold at every threshold tested. The headline count of 64 does not and should always be quoted with its threshold.

Municipal Wealth

Municipal wealth is measured as EQV per capita. EQV is determined biennially by the Department of Revenue on a common basis across municipalities, is the standard comparative measure of municipal fiscal capacity in Massachusetts, and is the measure the Commonwealth itself uses to distribute aid and apportion assessments. Income per capita was considered and rejected as the primary measure because it describes residents rather than the municipal government that operates the licensing process and files home rule petitions.

Municipalities are grouped into wealth quartiles constructed on the combined 328‑municipality sample, with the boundaries held fixed across every analysis so the quota and local-control results are comparable. Quartiles were used for the primary analysis because they produce strata large enough—82 municipalities each in the combined sample—to support non-parametric comparison. Because the boundaries are held fixed, the quota and local-control subsamples do not divide evenly across them.

Deciles were computed as a supplemental check and are reported in B.4. They are the reason the report does not claim a smooth wealth gradient: Mean access does not rise monotonically across deciles, and the effect is concentrated in the top decile. A continuous specification was also run and is consistent with the grouped results.

Wealth Quartile	EQV per Capita Range	Combined	Quota	Local Control
Q1 (lowest)	$76,427– $177,434	82	79	3
Q2	$177,434–$231,944	82	78	4
Q3	$231,944–$328,158	82	75	7
Q4 (highest)	$328,158–$2,837,676	82	70	12

Other Derived Measures

Population bands group municipalities by the statutory breakpoints—under 14,000, 14,000 to 25,000, and above 25,000—so that the effect of the floors and the over-25,000 increment can be seen separately. The hypothetical quota allocation applies the § 17 formula to each local-control municipality and compares the result with the licenses it actually issues; it is constructed to isolate the effect of the exemption by asking what the same municipality would face under the same statute and is reported in Finding 2. Historical allocation figures apply the same formula to decennial census counts back to 1930.

B.4 Analytical Strategy and Statistical Methods

The analyses are organized by the questions the report asks, not by the tests used. Each test appears with the question it was chosen to answer.

Why Rank-Based Methods

Every principal variable departs materially from normality, and Shapiro-Wilk rejects normality for each at p < 0.001. License access is strongly right-skewed, driven by small municipalities where the statutory floors produce very high per-resident values and by resort communities for which the service population far exceeds their residents.

Variable	n	Mean	SD	Median	Skew	Excess Kurtosis	Shapiro-Wilk W	p
Access combined	328	14.39	12.42	11.30	3.47	16.40	0.679	<0.001
Access quota	302	13.28	10.70	10.91	3.59	19.27	0.698	<0.001
Access local	26	27.39	21.22	21.48	2.02	3.60	0.746	<0.001
Utilization quota	302	51.18	30.73	53.74	-0.11	-1.23	0.946	<0.001
EQV combined	328	316,020.46	316,666.83	231,943.50	4.89	29.93	0.512	<0.001
Population combined	328	21,203.26	42,584.14	11,850.00	10.60	148.27	0.351	<0.001

The analysis therefore uses rank-based methods throughout: Kruskal-Wallis for comparisons across more than two groups, Mann-Whitney for pairwise contrasts, Spearman for correlation, and Wilcoxon signed-rank for the paired hypothetical quota allocation. Chi-square goodness-of-fit is used for the categorical question of how local-control municipalities are distributed across wealth quartiles. Medians are reported alongside means wherever the two differ materially, and the significance tests rest on the median pattern rather than the means. Confidence intervals on means are bias-corrected percentile bootstrap intervals from 10,000 resamples, because the normal-theory interval is not trustworthy at this level of skew. Effect sizes are reported as epsilon-squared for Kruskal-Wallis and rank-biserial correlation for Mann-Whitney, because a significant result on 328 observations can still describe a small difference. Welch’s t-test appears only as a parametric sensitivity check.

Finding	Null Hypothesis	Test
3, Analysis A	License access per 10,000 residents is distributed identically across wealth quartiles in the combined 328-municipality sample.	Kruskal-Wallis; Mann-Whitney for the Q4–Q1 contrast
3, Analysis B	License access is distributed identically across wealth quartiles within the 302 quota municipalities.	Kruskal-Wallis; Mann-Whitney for Q4–Q1
3, Analysis C	Local-control municipalities are distributed uniformly across wealth quartiles.	Chi-square goodness of fit
2	Local-control and quota municipalities issue the same number of licenses per 10,000 residents.	Mann-Whitney, one-sided
2	Local-control municipalities issue no more licenses than the § 17 formula would allocate to them.	Wilcoxon signed-rank on paired hypothetical quota allocation and actual
4	Licenses obtained by special legislation are distributed identically across wealth quartiles.	Kruskal-Wallis; chi-square on the proportion holding any
3, mechanism	License utilization is unrelated to municipal wealth.	Spearman rank correlation; Kruskal-Wallis across wealth quartiles

Population, Quota, and Utilization

This analysis asks whether the formula allocates licenses where they are wanted. The § 17 allocation was reconstructed for every municipality and compared with licenses issued, then examined by population band to isolate the effect of the statutory floors. The distribution of unused licenses was computed across the utilization bands and cross-tabulated against population, density, and wealth. The comparison that matters is between where available licenses sit and where utilization indicates they are wanted; results are in Finding 1. Constrained municipalities (n = 64 at the ≥80% threshold) account for 2,531,376 residents in the 2023 population; available licenses use the category-floored formulas in B.3.

Local Control Against Quota

This analysis asks whether the exemption changed outcomes. Local-control and quota municipalities are compared on licenses issued per 10,000 residents using a one-sided Mann-Whitney test, and the comparison is repeated at every decile of the distribution to establish whether the gap is general or driven by a few high values. A population-weighted comparison is reported alongside the unweighted one. The § 17 formula is then applied to each local-control municipality and compared against its actual issuance with a Wilcoxon signed-rank test on the paired values. A matched-pair comparison against quota municipalities of similar population and wealth was run as an independent check. Results are in Finding 2.

Wealth and Equity

This analysis asks whether license access varies with municipal wealth and, if so, through which mechanism. Access is compared across wealth quartiles by Kruskal-Wallis, first on the combined 328-municipality sample and then within the 302 quota municipalities alone, with the Q4–Q1 contrast tested by Mann-Whitney. The distribution of local-control municipalities across wealth quartiles is tested by chi-square. Wealth is then related to access continuously by Spearman correlation and re-examined across deciles.

Having established a gradient, the analysis tests the three candidate mechanisms in turn. Allocation is ruled out by construction, since the formula counts only residents. Utilization is tested against wealth by Spearman correlation across the quota municipalities. Special legislation is tested in two ways: whether wealthier municipalities are more likely to hold any such license, by chi-square on the proportion, and whether they hold more of them, by Kruskal-Wallis on the volume. Distinguishing those two questions is what allows the report to say that wealth does not predict whether a municipality seeks relief but does correspond to how much it obtains. Results are in Findings 3 and 4.

Sensitivity and Robustness

Each principal result was re-run under the alternative specifications in the following table. The table reports the checks that bear on whether a finding survives.

Check	Result
Excluding Boston from the quota sample	Spearman (EQV, access) 0.114 against 0.118 with Boston; Kruskal-Wallis H = 5.918, p = 0.1157; Q4–Q1 difference +37.0%. Conclusions unchanged.
Wealth deciles instead of quartiles	Spearman (decile rank, access) rho = 0.176, ρ = 0.176, p = 0.001. The gradient is not monotonic across deciles; the effect is concentrated in the top decile. This is why the report does not describe a smooth wealth gradient.
Constrained threshold at 70%, 75%, 85%, and 90%	Counts range from 101 to 33 municipalities. Substantive conclusions of Finding 1 hold throughout; the headline count does not, and the threshold is stated wherever the count appears.
Excluding the three highest local-control municipalities	Regime difference remains highly significant (p < 0.001); median ratio moves from 1.97 to 1.84.
Continuous wealth measure instead of grouped	Spearman ρ = 0.180 (p = 0.001) combined; ρ = 0.118 (p = 0.040) within the quota framework. Consistent with the grouped analyses.

Multiple Comparisons and Statistical Power

The report performs a small number of pre-specified tests corresponding to its findings rather than an exploratory search, so no multiplicity correction is applied as the primary analysis. As a check, a Bonferroni correction across the seven principal tests gives an adjusted threshold of 0.0071. The combined-sample wealth comparison and the regime comparison survive it; the within-quota omnibus test and the local‑control distribution test do not, and the report describes those as patterns rather than as tested results. No comparison within the local-control group across wealth quartiles is reported as a finding, because the strata are too small to report one, with three local-control municipalities in Q1 and four in Q2.

B.5 Quantitative Limitations

1. Measurement. Utilization proxies for constraint but does not measure demand. A municipality with spare capacity may face no demand rather than no barrier.
2. Classification assumptions. Combining Wine-and-Malt with cordials, and treating Wine-only and Malt-only as Wine-and-Malt, are analytical choices. Both are conservative in the sense that they add capacity to the denominator.
3. Population-based normalization. The report’s per-10,000 rates use 2023 US Census Bureau Population Estimates (via the Department of Revenue/Division of Local Standards). The resident population is not the population that generates demand in seasonal, commuter, or student communities. Per-resident measures overstate access in resort municipalities and understate it in employment centers.
4. Wealth measurement. EQV is a property-value measure. It does not capture income, employment, or fiscal effort, and it is inflated in communities with large second-home bases relative to their year-round population.
5. Comparability. Local-control municipalities are not a random sample. Selection into the 1978–1984 window turned on prior seasonal licensing, and the hypothetical quota allocation assumes issuance behavior would have been unchanged under a binding ceiling, which cannot be tested.
6. Special legislation. The flag captures acts for which licenses remain active. It does not capture petitions that failed or acts for which licenses have since lapsed, so it understates the number of petitions filed and cannot support any statement about success rates.
7. Causal inference. Every relationship reported is an association measured at one point in time across municipalities that differ in many unmeasured ways.

B.6 Municipality Reference Lists

The full membership of each classification is listed in the following table so that any figure in the report can be traced to the municipalities it covers. These lists are reference material and are placed at the end of this appendix to keep the methodological sequence in B.1 through B.5 continuous.

Local-Control Municipalities (n = 26)

Municipalities exempt from MGL c. 138, § 17, quota limits on § 12 on-premises licenses

#	Municipality	County	Region	Population (2023)	EQV per Capita (Fiscal Year 2026)
1	Barnstable	Barnstable	Cape Cod	49,709	$528,992
2	Bourne	Barnstable	Cape Cod	20,422	$392,582
3	Cambridge	Middlesex	Greater Boston	118,214	$722,243
4	Dartmouth	Bristol	South Coast	33,419	$254,835
5	Dennis	Barnstable	Cape Cod	14,903	$876,782
6	Falmouth	Barnstable	Cape Cod	33,069	$650,356
7	Franklin	Norfolk	MetroWest	33,125	$257,761
8	Great Barrington	Berkshire	Berkshires	7,194	$306,123
9	Haverhill	Essex	Merrimack Valley	67,415	$168,516
10	Kingston	Plymouth	South Shore	13,948	$246,562
11	Lee	Berkshire	Berkshires	5,694	$254,555
12	Lenox	Berkshire	Berkshires	5,064	$379,411
13	Marlborough	Middlesex	MetroWest	41,179	$221,787
14	Mashpee	Barnstable	Cape Cod	15,396	$651,648
15	Middleborough	Plymouth	South Shore	24,504	$180,413
16	Nantucket	Nantucket	Islands	14,444	$2,817,820
17	Newburyport	Essex	North Shore	18,731	$395,601
18	North Adams	Berkshire	Berkshires	12,483	$83,573
19	Orleans	Barnstable	Cape Cod	6,385	$1,052,573
20	Pembroke	Plymouth	South Shore	18,286	$237,537
21	Plymouth	Plymouth	South Shore	65,405	$278,788
22	Provincetown	Barnstable	Cape Cod	3,709	$1,574,301
23	Sturbridge	Worcester	Central MA	9,928	$197,168
24	Williamstown	Berkshire	Berkshires	7,385	$193,022
25	Worcester	Worcester	Central MA	207,621	$115,659
26	Yarmouth	Barnstable	Cape Cod	25,103	$432,766

Excluded Municipalities (n = 23)

Municipalities outside the scope of this study

#	Municipality	County	Population (2023)
1	Florida	Berkshire	675
2	Heath	Franklin	720
3	Leyden	Franklin	731
4	Middlefield	Hampshire	392
5	Monroe	Franklin	115
6	Monterey	Berkshire	1,080
7	New Braintree	Worcester	993
8	Pelham	Hampshire	1,265
9	Peru	Berkshire	800
10	Rowe	Franklin	421
11	Royalston	Worcester	1,276
12	Tolland	Hampden	467
13	Warwick	Franklin	776
14	Washington	Berkshire	496
15	Alford	Berkshire	481
16	Chilmark	Dukes	1,232
17	Dunstable	Middlesex	3,408
18	Gosnold	Dukes	67
19	Hawley	Franklin	353
20	Montgomery	Hampden	825
21	Mount Washington	Berkshire	157
22	Westhampton	Hampshire	1,626
23	Milton	Norfolk	28,374

From these excluded municipalities, Alford, Chilmark, Dunstable, Gosnold, Hawley, Montgomery, Mount Washington, and West Hampton voted to prohibit the sale of alcoholic beverages. Milton prohibits All-Alcohol licenses while permitting Wine-and-Malt licenses.

Quota Bound Municipalities (n=302)

Municipalities bound to MGL c. 138, § 17 quota limits on § 12 on-premises licenses

#	Municipality	County	Population (2023)	EQV per Capita (Fiscal Year 2026)
1	Abington	Plymouth	16,970	$193,195
2	Acton	Middlesex	24,029	$284,449
3	Acushnet	Bristol	10,602	$186,591
4	Adams	Berkshire	7,995	$98,794
5	Agawam	Hampden	28,406	$154,538
6	Amesbury	Essex	17,303	$227,235
7	Amherst	Hampshire	40,277	$91,085
8	Andover	Essex	36,485	$356,875
9	Aquinnah	Dukes	446	$2,211,660
10	Arlington	Middlesex	46,111	$344,666
11	Ashburnham	Worcester	6,415	$178,729
12	Ashby	Middlesex	3,198	$170,705
13	Ashfield	Franklin	1,690	$237,063
14	Ashland	Middlesex	18,591	$246,229
15	Athol	Worcester	11,937	$117,821
16	Attleboro	Bristol	46,654	$161,797
17	Auburn	Worcester	16,820	$196,912
18	Avon	Norfolk	4,725	$262,414
19	Ayer	Middlesex	8,546	$223,597
20	Barre	Worcester	5,536	$141,462
21	Becket	Berkshire	1,921	$391,223
22	Bedford	Middlesex	14,394	$394,263
23	Belchertown	Hampshire	15,339	$154,797
24	Bellingham	Norfolk	17,556	$232,194
25	Belmont	Middlesex	26,886	$446,087
26	Berkley	Bristol	6,822	$196,743
27	Berlin	Worcester	3,425	$308,544
28	Bernardston	Franklin	2,113	$159,016
29	Beverly	Essex	42,318	$254,620
30	Billerica	Middlesex	41,619	$270,033
31	Blackstone	Worcester	9,210	$161,343
32	Blandford	Hampden	1,204	$256,742
33	Bolton	Worcester	5,738	$299,009
34	Boston	Suffolk	653,833	$369,761
35	Boxborough	Middlesex	5,451	$321,573
36	Boxford	Essex	8,149	$354,379
37	Boylston	Worcester	4,972	$282,034
38	Braintree	Norfolk	38,490	$268,651
39	Brewster	Barnstable	10,381	$689,214
40	Bridgewater	Plymouth	28,818	$183,892
41	Brimfield	Hampden	3,699	$171,301
42	Brockton	Plymouth	104,890	$127,532
43	Brookfield	Worcester	3,448	$130,574
44	Brookline	Norfolk	62,962	$536,333
45	Buckland	Franklin	1,809	$166,714
46	Burlington	Middlesex	26,527	$389,341
47	Canton	Norfolk	24,635	$335,585
48	Carlisle	Middlesex	5,192	$482,194
49	Carver	Plymouth	11,675	$189,253
50	Charlemont	Franklin	1,177	$164,728
51	Charlton	Worcester	13,391	$230,005
52	Chatham	Barnstable	6,685	$1,893,646
53	Chelmsford	Middlesex	36,180	$256,214
54	Chelsea	Suffolk	38,319	$165,668
55	Cheshire	Berkshire	3,195	$140,202
56	Chester	Hampden	1,211	$146,000
57	Chesterfield	Hampshire	1,173	$219,667
58	Chicopee	Hampden	54,838	$106,162
59	Clarksburg	Berkshire	1,640	$98,605
60	Clinton	Worcester	15,530	$151,736
61	Cohasset	Norfolk	8,392	$504,316
62	Colrain	Franklin	1,629	$138,944
63	Concord	Middlesex	18,086	$553,232
64	Conway	Franklin	1,762	$205,156
65	Cummington	Hampshire	819	$214,923
66	Dalton	Berkshire	6,200	$152,883
67	Danvers	Essex	27,896	$276,620
68	Dedham	Norfolk	24,968	$318,727
69	Deerfield	Franklin	5,176	$207,261
70	Dighton	Bristol	8,182	$190,074
71	Douglas	Worcester	9,241	$188,868
72	Dover	Norfolk	5,864	$614,821
73	Dracut	Middlesex	32,291	$184,677
74	Dudley	Worcester	11,849	$141,133
75	Duxbury	Plymouth	16,185	$461,802
76	East Bridgewater	Plymouth	14,408	$189,866
77	East Brookfield	Worcester	2,202	$177,645
78	East Longmeadow	Hampden	16,378	$174,686
79	Eastham	Barnstable	5,803	$904,336
80	Easthampton	Hampshire	16,031	$154,206
81	Easton	Bristol	25,364	$214,457
82	Edgartown	Dukes	5,268	$2,837,676
83	Egremont	Berkshire	1,371	$635,428
84	Erving	Franklin	1,663	$534,328
85	Essex	Essex	3,694	$375,138
86	Everett	Middlesex	50,318	$210,241
87	Fairhaven	Bristol	15,878	$220,405
88	Fall River	Bristol	93,840	$108,751
89	Fitchburg	Worcester	41,579	$116,669
90	Foxborough	Norfolk	18,484	$245,484
91	Framingham	Middlesex	71,875	$214,117
92	Freetown	Bristol	9,290	$248,533
93	Gardner	Worcester	20,974	$110,885
94	Georgetown	Essex	8,444	$238,225
95	Gill	Franklin	1,552	$141,042
96	Gloucester	Essex	29,959	$374,847
97	Goshen	Hampshire	944	$233,280
98	Grafton	Worcester	19,975	$207,208
99	Granby	Hampshire	6,060	$171,025
100	Granville	Hampden	1,525	$196,547
101	Greenfield	Franklin	17,628	$126,293
102	Groton	Middlesex	11,277	$258,444
103	Groveland	Essex	6,743	$226,901
104	Hadley	Hampshire	5,277	$266,433
105	Halifax	Plymouth	7,718	$194,278
106	Hamilton	Essex	7,586	$310,721
107	Hampden	Hampden	4,913	$186,096
108	Hancock	Berkshire	739	$545,567
109	Hanover	Plymouth	14,753	$300,580
110	Hanson	Plymouth	10,586	$200,023
111	Hardwick	Worcester	2,664	$146,815
112	Harvard	Worcester	6,928	$277,270
113	Harwich	Barnstable	13,595	$758,256
114	Hatfield	Hampshire	3,319	$229,729
115	Hingham	Plymouth	24,189	$437,453
116	Hinsdale	Berkshire	1,896	$262,247
117	Holbrook	Norfolk	11,284	$186,118
118	Holden	Worcester	19,934	$191,193
119	Holland	Hampden	2,565	$199,581
120	Holliston	Middlesex	15,015	$265,606
121	Holyoke	Hampden	37,628	$76,427
122	Hopedale	Worcester	6,028	$191,671
123	Hopkinton	Middlesex	19,540	$346,262
124	Hubbardston	Worcester	4,346	$180,725
125	Hudson	Middlesex	19,904	$215,357
126	Hull	Plymouth	10,108	$332,017
127	Huntington	Hampshire	2,071	$147,233
128	Ipswich	Essex	13,903	$348,415
129	Lakeville	Plymouth	11,968	$253,076
130	Lancaster	Worcester	8,802	$170,123
131	Lanesborough	Berkshire	3,007	$201,003
132	Lawrence	Essex	88,172	$94,364
133	Leicester	Worcester	11,066	$158,734
134	Leominster	Worcester	43,627	$146,548
135	Leverett	Franklin	1,861	$226,291
136	Lexington	Middlesex	33,882	$552,286
137	Lincoln	Middlesex	6,889	$443,322
138	Littleton	Middlesex	10,251	$312,249
139	Longmeadow	Hampden	15,621	$197,930
140	Lowell	Middlesex	114,296	$120,478
141	Ludlow	Hampden	20,845	$146,219
142	Lunenburg	Worcester	11,851	$200,412
143	Lynn	Essex	101,241	$147,334
144	Lynnfield	Essex	13,034	$390,665
145	Malden	Middlesex	65,133	$170,965
146	Manchester By The Sea	Essex	5,396	$666,826
147	Mansfield	Bristol	23,862	$242,963
148	Marblehead	Essex	20,296	$468,251
149	Marion	Plymouth	5,292	$501,379
150	Marshfield	Plymouth	25,765	$324,098
151	Mattapoisett	Plymouth	6,652	$426,015
152	Maynard	Middlesex	10,604	$209,982
153	Medfield	Norfolk	13,088	$313,637
154	Medford	Middlesex	58,744	$280,246
155	Medway	Norfolk	13,560	$268,322
156	Melrose	Middlesex	29,357	$277,522
157	Mendon	Worcester	6,331	$247,699
158	Merrimac	Essex	6,718	$212,869
159	Methuen	Essex	53,455	$179,738
160	Middleton	Essex	9,849	$346,584
161	Milford	Worcester	30,257	$186,655
162	Millbury	Worcester	14,002	$197,262
163	Millis	Norfolk	9,018	$247,291
164	Millville	Worcester	3,146	$151,366
165	Monson	Hampden	8,104	$144,083
166	Montague	Franklin	8,444	$151,599
167	Nahant	Essex	3,300	$443,893
168	Natick	Middlesex	36,518	$346,172
169	Needham	Norfolk	32,157	$443,945
170	New Ashford	Berkshire	247	$240,054
171	New Bedford	Bristol	100,695	$110,535
172	New Marlborough	Berkshire	1,503	$548,280
173	New Salem	Franklin	1,014	$175,063
174	Newbury	Essex	6,766	$413,685
175	Newton	Middlesex	88,415	$503,981
176	Norfolk	Norfolk	11,594	$245,656
177	North Andover	Essex	31,615	$273,452
178	North Attleborough	Bristol	30,943	$196,664
179	North Brookfield	Worcester	4,724	$142,897
180	North Reading	Middlesex	16,054	$311,648
181	Northampton	Hampshire	29,370	$182,422
182	Northborough	Worcester	15,689	$289,444
183	Northbridge	Worcester	16,455	$168,036
184	Northfield	Franklin	2,863	$241,000
185	Norton	Bristol	19,146	$204,929
186	Norwell	Plymouth	11,299	$379,389
187	Norwood	Norfolk	31,230	$263,990
188	Oak Bluffs	Dukes	5,355	$1,201,689
189	Oakham	Worcester	1,886	$179,536
190	Orange	Franklin	7,558	$115,967
191	Otis	Berkshire	1,619	$545,243
192	Oxford	Worcester	13,306	$170,079
193	Palmer	Hampden	12,315	$114,675
194	Paxton	Worcester	5,016	$166,995
195	Peabody	Essex	54,056	$222,205
196	Pepperell	Middlesex	11,710	$180,573
197	Petersham	Worcester	1,181	$215,805
198	Phillipston	Worcester	1,724	$205,657
199	Pittsfield	Berkshire	43,076	$120,524
200	Plainfield	Hampshire	637	$198,875
201	Plainville	Norfolk	9,881	$246,835
202	Plympton	Plymouth	2,932	$268,079
203	Princeton	Worcester	3,508	$231,882
204	Quincy	Norfolk	101,597	$239,737
205	Randolph	Norfolk	34,487	$180,721
206	Raynham	Bristol	15,649	$215,986
207	Reading	Middlesex	25,428	$313,923
208	Rehoboth	Bristol	13,313	$221,288
209	Revere	Suffolk	57,954	$207,408
210	Richmond	Berkshire	1,401	$470,486
211	Rochester	Plymouth	5,871	$270,829
212	Rockland	Plymouth	17,594	$197,006
213	Rockport	Essex	6,988	$494,187
214	Rowley	Essex	6,358	$286,381
215	Russell	Hampden	1,627	$147,731
216	Rutland	Worcester	9,422	$164,683
217	Salem	Essex	44,744	$209,519
218	Salisbury	Essex	9,259	$294,925
219	Sandisfield	Berkshire	981	$351,915
220	Sandwich	Barnstable	20,469	$370,304
221	Saugus	Essex	28,630	$260,292
222	Savoy	Berkshire	647	$153,961
223	Scituate	Plymouth	19,297	$430,873
224	Seekonk	Bristol	15,725	$225,063
225	Sharon	Norfolk	18,442	$287,843
226	Sheffield	Berkshire	3,317	$288,037
227	Shelburne	Franklin	1,878	$189,409
228	Sherborn	Middlesex	4,409	$428,335
229	Shirley	Middlesex	6,851	$158,216
230	Shrewsbury	Worcester	38,889	$235,454
231	Shutesbury	Franklin	1,721	$189,098
232	Somerset	Bristol	18,209	$193,989
233	Somerville	Middlesex	80,407	$332,880
234	South Hadley	Hampshire	17,992	$133,421
235	Southampton	Hampshire	6,195	$182,466
236	Southborough	Worcester	10,429	$364,670
237	Southbridge	Worcester	17,880	$98,783
238	Southwick	Hampden	9,195	$171,525
239	Spencer	Worcester	11,920	$147,388
240	Springfield	Hampden	153,672	$89,393
241	Sterling	Worcester	8,204	$224,810
242	Stockbridge	Berkshire	1,994	$657,374
243	Stoneham	Middlesex	22,854	$269,393
244	Stoughton	Norfolk	28,962	$216,959
245	Stow	Middlesex	7,088	$296,882
246	Sudbury	Middlesex	19,394	$392,068
247	Sunderland	Franklin	3,636	$178,457
248	Sutton	Worcester	9,436	$256,113
249	Swampscott	Essex	15,487	$320,402
250	Swansea	Bristol	17,375	$207,353
251	Taunton	Bristol	60,412	$159,328
252	Templeton	Worcester	8,261	$140,104
253	Tewksbury	Middlesex	31,243	$246,488
254	Tisbury	Dukes	4,865	$1,099,786
255	Topsfield	Essex	6,544	$321,611
256	Townsend	Middlesex	9,004	$166,485
257	Truro	Barnstable	2,479	$1,637,935
258	Tyngsborough	Middlesex	12,475	$229,936
259	Tyringham	Berkshire	420	$592,215
260	Upton	Worcester	8,311	$235,507
261	Uxbridge	Worcester	14,526	$227,017
262	Wakefield	Middlesex	27,810	$279,787
263	Wales	Hampden	1,802	$151,980
264	Walpole	Norfolk	25,964	$284,704
265	Waltham	Middlesex	64,477	$316,593
266	Ware	Hampshire	10,067	$113,078
267	Wareham	Plymouth	23,198	$256,620
268	Warren	Worcester	5,000	$123,127
269	Watertown	Middlesex	35,256	$372,794
270	Wayland	Middlesex	13,752	$419,066
271	Webster	Worcester	17,593	$155,601
272	Wellesley	Norfolk	30,733	$586,124
273	Wellfleet	Barnstable	3,627	$1,176,938
274	Wendell	Franklin	921	$148,832
275	Wenham	Essex	4,957	$279,874
276	West Boylston	Worcester	7,790	$201,904
277	West Bridgewater	Plymouth	7,681	$234,703
278	West Brookfield	Worcester	3,844	$168,533
279	West Newbury	Essex	4,598	$336,921
280	West Springfield	Hampden	28,424	$140,007
281	West Stockbridge	Berkshire	1,325	$495,246
282	West Tisbury	Dukes	3,586	$1,380,377
283	Westborough	Worcester	22,012	$290,887
284	Westfield	Hampden	40,509	$124,243
285	Westford	Middlesex	24,543	$291,343
286	Westminster	Worcester	8,377	$209,581
287	Weston	Middlesex	11,645	$767,431
288	Westport	Bristol	16,461	$316,754
289	Westwood	Norfolk	16,244	$448,374
290	Weymouth	Norfolk	59,114	$226,922
291	Whately	Franklin	1,625	$231,456
292	Whitman	Plymouth	15,316	$175,390
293	Wilbraham	Hampden	14,518	$176,800
294	Williamsburg	Hampshire	2,469	$173,171
295	Wilmington	Middlesex	23,148	$326,871
296	Winchendon	Worcester	10,441	$122,042
297	Winchester	Middlesex	22,837	$509,454
298	Windsor	Berkshire	815	$209,288
299	Winthrop	Suffolk	18,319	$211,444
300	Woburn	Middlesex	41,647	$298,866
301	Worthington	Hampshire	1,179	$232,005
302	Wrentham	Norfolk	12,543	$280,149

246.      Massachusetts Department of Revenue, Division of Local Services, Data Analytics and Resources Bureau, “Population By Municipality and Year,” Municipal Databank, 2023 column, accessed September 13, 2026, https://dls-gw.dor.state.ma.us/reports/rdPage.aspx?rdReport=Socioeconomic.Population.population_main, which republishes US Census Bureau, Vintage 2023 Population Estimates: City and Town Population Totals, Massachusetts, released May 2024, https://www.census.gov/newsroom/press-kits/2024/subcounty-population-estimates.html. Population figures are the 2023 estimates for each of the 351 Commonwealth municipalities, summed within each classification; the three categories total 7,001,399. These 2023 estimates are used for all per-resident rates in this report.

Date published:	October 13, 2026