Skip to content

British Columbia · Reported library revenue · 2018–2024

Library revenue variation in British Columbia: why the weighting changes the answer

Public Libraries Branch grants dominate relative within-system variation, but adding baseline revenue weights makes local support the largest contributor. The contrast concerns a selected panel and a pinned workbook—not provincial funding risk.

Published
Share
Share this page
EmailCite & reuse this study
Key findings as text
Reusable imagesLibrary revenue variation in British Columbia: why the weighting changes the answer — 43 / 71: Metadata-consistent reporting systems retained. Seven complete years; retained revenue coverage 52.19%–57.14%, not full provincial representation.Download sharing image (JPG) ↓

Key findings

43 / 71

Metadata-consistent reporting systems retained

Seven complete years; retained revenue coverage 52.19%–57.14%, not full provincial representation.

70.95%

PLB primary pooled normalized-variance contribution

Variance-weighted signed covariance share, not its share of funding received.

49.64%

Local contribution after baseline-revenue weighting

Adding 2018 revenue weights changes the leading component; PLB becomes 38.41%.

35 / 43

Systems with PLB as largest signed contributor

No causal grant effect, future-risk estimate or named financial-risk ranking.

One funding question, two different answers to weighting

Which funding stream contributes most to fluctuations in reported public-library revenue? The answer depends on what receives weight in the calculation. In a fixed panel of 43 British Columbia reporting systems over 2018–2024, Public Libraries Branch grants account for 70.95% of a pooled decomposition of within-system normalized revenue-change variance. But adding fixed 2018 revenue weights changes the ordering: local-government support contributes 49.64%, compared with 38.41% for those grants. Neither calculation describes the variance of all British Columbia library revenue. They summarize different features of the same selected reporting panel.

This distinction matters for readers trying to interpret a funding-stability claim. A large relative movement in a small budget can matter greatly to that reporting system, while contributing little to the movement of a larger combined budget. Giving more influence to larger starting budgets therefore changes the question. The first calculation already weights systems by their observed normalized variance; the second adds baseline revenue weighting on top. Calling either result the experience of the typical library, without explaining the weights, would obscure what the numbers measure.

The analysis is also conditional on a particular official workbook. Another officially listed format contains a material historical disagreement, discussed below. Independent dated grant documents support the workbook's year labels, but do not certify every value. The retained systems represent only 52.19% to 57.14% of the revenue reported by all source rows across the seven years. These are bounded historical accounting results, not a complete provincial financial assessment, an estimate of future risk or a recommendation to replace one funding source with another.

Why the workbook and the CSV cannot be treated as interchangeable

The source is the Province of British Columbia's public-library statistics dataset, together with its published survey questions. The downloaded workbook is identified by a frozen checksum. Although the resource filename refers to version 24 and a 2002–2024 span, the workbook itself identifies version 25 and contains a later sheet. This study fixes 2018–2024 and does not append the later year. A reproducible citation must identify the actual bytes and selected sheets rather than assume that a filename is an immutable description of the contents.

The separately downloaded CSV agrees with the workbook on every paired numeric value in the four selected revenue fields for 2023 and 2024. Earlier years do not agree: only a handful of paired cells match within each year from 2018 through 2022. Several comparisons instead show the CSV carrying values from an earlier workbook year. This is a diagnostic observation, not authority to shift dates. Both formats can reconcile internally, so a successful addition check alone cannot establish which reporting period their values describe.

Independent Burnaby documents provide dated anchors. The 2019 annual report identifies provincial grants of C$476,627 in 2018 and C$470,080 in 2019. The 2021 report identifies C$512,359 in 2020 and C$470,080 in 2021, matching workbook values at the report's whole-dollar precision. The 2022 financial disclosure separately identifies a C$122,927 COVID-relief receipt, matching that year's additional grant field. These anchors support the chosen workbook dates across the disputed period. They do not validate every system's revenue or explain every format difference. No CSV values are relabelled, merged or presented as a same-period replication.

Reported revenue is broader than operating resources

The decomposition uses three source aggregates: total local-government support, total Public Libraries Branch grants, and total Other Revenue. Their sum is the source's total revenue. Local support combines municipal and regional contributions under the survey definition, which excludes in-kind contributions and costs paid directly by a municipality. The provincial-grant aggregate is specifically the Public Libraries Branch category. Other remains broad: it can include other government support, fees, fines, donations and additional sources. It is not a clean measure of private fundraising or earned commercial income.

The analysis does not turn these aggregates into recurrent operating revenue. Operating and capital detail is not consistently available over the selected years, even where column headings persist. One-off and potentially capital-related amounts therefore remain within the reported totals. Nor is a uniform cash-accounting basis demonstrated. Burnaby's 2022 disclosure explicitly distinguishes receiving restricted funds from recognizing them as revenue when related expenses occur. That document is useful as a date anchor, but its accounting total cannot simply replace the survey total, whose perimeter differs.

Transfers from reserves are reported separately and are not inserted as a fourth component. Available reserve-transfer values are incomplete: only 25 to 38 retained systems provide numeric values in a given year. Blank entries and textual unavailable or not-applicable markers remain nonnumeric, rather than becoming zeros. Any sum of the available reserve values is therefore a partial availability summary, not the panel's complete reserve activity. The revenue decomposition neither attributes unexplained movements to reserves nor substitutes expenditure for revenue when a desired funding measure is unavailable.

A consistent set of reporting records, not a verified boundary history

The workbook provides 71 system symbols in each selected year. A symbol appearing repeatedly is not enough to establish comparability. Names and system types are checked alongside it. Two symbols switch between Vancouver Public Library and Vancouver Island Regional Library in the historical metadata. They are not silently exchanged or repaired. Other short-name changes are excluded conservatively, even where a change might be cosmetic, because an undocumented assumption about identity would otherwise enter the time series.

A system must also have numeric values for all three revenue components and total revenue in every selected year, positive totals, and additive reconciliation. Missing values are not inferred from a residual that happens to be zero. These fixed checks leave 43 systems and exclude 28. Every retained component is nonnegative, and the retained components reconcile exactly as reported. The public results include the exclusion reasons so readers can distinguish metadata inconsistency from incomplete financial fields rather than treating all exclusions as equivalent evidence of poor reporting.

The retained group is described as metadata-consistent, not as a set of independently verified, unchanging service populations. Unrecorded boundary or reporting changes may remain. Its share of all source-row revenue ranges from 52.19% to 57.14%; the exclusions include large systems and cannot be dismissed as negligible. The seven-year balanced panel improves within-record comparison, but it is not a probability sample from which missing systems can be represented automatically. Province-wide claims would require a different completeness and identity solution.

How the accounting decomposition works

For each system and each of the six annual transitions, the change in each component is divided by the previous year's total reported revenue. The three resulting component contributions add to that system's total proportional revenue change. For example, a component increase of ten against a previous total of one hundred contributes ten percentage points to total revenue growth. It is not the component's own percentage growth. The common lagged-total denominator makes the contributions additive without confusing a small funding stream's high growth rate with a large contribution to the overall budget.

Across the six transitions, total normalized growth has a sample variance. Each component receives its covariance with total normalized growth. Because the component changes add to the total change, the three covariances add to total variance. Dividing each covariance by that variance produces signed shares that sum to one within a system. This is an accounting identity about co-movement. It does not identify which funding authority caused a change, whether a grant prevented a worse outcome, or what would have happened under another financing arrangement.

The primary pooled statistic sums component covariances over retained systems and divides by the sum of their variances. It therefore equals a variance-weighted average of their individual shares. Each system contributes six transitions, but equal transition counts do not make the weights equal. A separate sensitivity calculation multiplies both covariance and variance by the system's fixed 2018 revenue before pooling. Medians and quartiles provide a different distributional view. No system has zero variance here; in general, such a system would have an undefined individual covariance share rather than a share of zero.

What the primary result supports

The pooled within-system normalized-variance shares are 6.81% for local-government support, 70.95% for Public Libraries Branch grants and 22.24% for Other Revenue. The provincial-grant component is the largest signed contributor in 35 of the 43 retained systems. Local support is largest in six, and Other in two, with no ties. This is not the proportion of annual revenue supplied by each source. A component can provide a modest share of a budget while contributing substantially to its historical relative fluctuations.

The median system shares are 7.39%, 72.39% and 9.80%, respectively. The middle half of provincial-grant shares extends from 55.51% to 88.75%. The median and pooled provincial-grant figures are close in this panel, but they answer different questions and their numerical proximity should not be generalized. Component medians do not sum to one because the median for each component can come from a different system. They must not be drawn as a stacked composition or interpreted as the funding decomposition of a single representative library.

The largest individual variance weight is 16.39%. Recalculating the pooled statistic after omitting each system in turn gives a provincial-grant range of 69.66% to 75.24%. The corresponding ranges are 5.97% to 8.15% for local support and 17.13% to 23.41% for Other. These are influence diagnostics, not confidence intervals or alternative estimates from which to choose a preferred answer. The main result retains every eligible system. A large weight is not itself evidence that a system's data are wrong or that its finances are unusually badly managed.

Why budget-size weighting changes the emphasis

With fixed 2018 revenue weights added, the contributions become 49.64% local-government support, 38.41% Public Libraries Branch grants and 11.94% Other. The local-government component now exceeds the provincial-grant component. This is a substantial change in interpretation, not merely a rounding difference. It shows why an article about funding variation should disclose its weighting rule before offering a simple statement about the dominant source. A result about relative movements across reporting systems need not match one that gives larger starting budgets additional influence.

The size-weighted version still is not the variance decomposition of combined panel revenue. Its weights are products of baseline revenue and each system's normalized variance. A genuine aggregate calculation would first sum annual revenues across systems and then examine the resulting time series, incorporating co-movement between systems. Here, aggregate annual growth is supplied as a descriptive comparator, not substituted for either within-system pooled estimand. Keeping those constructions separate prevents a provincial-sounding interpretation from being attached to a statistic that was not designed to answer a provincial-total question.

Neither weighting rule is intrinsically the correct one for every reader. A service planner concerned with proportional movement within reporting units may find the normalized view useful. A reader concerned with the influence of larger budgets may want the baseline-size sensitivity beside it. All amounts remain nominal: changes in prices, purchasing power and the cost of delivering library services are outside the estimand. A financial series can vary even when service resources move differently, so the results cannot establish equivalent changes in opening hours, staffing or collections. Readers should not mistake a descriptive covariance contribution for a measure of funding adequacy, service quality or the desirability of diversification. The useful result is the contrast between explicitly defined summaries, together with the limits of the available panel.

Negative shares and annual changes need careful language

Six retained systems have a negative local-government covariance contribution, one has a negative provincial-grant contribution and ten have a negative Other contribution. Four provincial-grant shares exceed 100%. These possibilities are mathematically coherent even though the observed component revenues are nonnegative. A component that tends to move against total changes can offset another component's positive co-movement. The shares summarize those signed relationships; they are neither probabilities nor literal percentages of money received. They do not imply negative grants, accounting fraud or a causal stabilizing effect.

Annual summaries also change with aggregation. In 2023, the arithmetic mean of system-level total growth is 40.41%, while growth of the retained panel's combined revenue is 14.31%. In 2024, those figures are −21.87% and −2.19%. The arithmetic mean gives each retained system equal weight for that annual summary; combined growth effectively weights systems by their previous-year totals. These are separate from the variance weights used in the main decomposition. The contrasts demonstrate the consequences of aggregation without identifying a particular administrative or policy mechanism.

Three arithmetic mean component contributions add to the arithmetic mean total growth within each year, because they use the same systems and denominator definition. The median total growth is shown separately: medians of components would not generally add to it. Large one-off receipts followed by lower receipts can generate a pronounced increase and decrease even without a cut to recurrent support. Six historical transitions cannot establish unpredictability, a future-risk distribution or a grant programme's effect. This study therefore avoids describing every downward movement as a funding withdrawal.

What this adds, and what it leaves open

Library funding stability is already a research topic. Enami, Byington and Alm's study of Ohio libraries examines funding stabilization, diversification and institutional performance using a different policy setting and causal research design. The present analysis does not claim the first study of funding stability or attempt to reproduce those causal conclusions in British Columbia. Its contribution is narrower: a transparent decomposition of reported revenue co-movement, a comparison of weighting targets, and explicit treatment of the source-period and coverage problems that constrain the answer.

The public reproduction uses selected reporting-system inputs, not an entire multi-year workbook. It retains all 497 selected rows so that the 43-system eligibility rule and excluded records can be checked. The accompanying code runs with the Python standard library and explicit input and output paths. It reproduces the accounting identities, system-distribution summaries, alternative weights, annual comparisons and omission diagnostics. Names and identifiers support source traceability; they are not presented as an efficiency or financial-risk league table. Original file hashes and source links document which snapshot the analysis uses.

The conclusion is deliberately conditional and two-part. In this pinned, metadata-consistent panel, Public Libraries Branch grants dominate pooled within-system normalized revenue-change variance and are the largest contributor for most retained systems. Adding baseline revenue weighting changes the leading component to local-government support. Both statements must travel with the incomplete provincial coverage, broad reported-revenue perimeter, official-format conflict and short historical window. Together they offer readers a defensible way to ask what a funding-variation statistic actually weights, rather than a universal verdict on which source makes libraries financially secure.

Baseline revenue weighting changes the leading contribution

British Columbia · 43 reporting systems · 2018–2024

Full chart and source
Baseline revenue weighting changes the leading contribution. Primary weights: within-system variance. Second: variance × 2018 revenue.
Conditional on the pinned workbook; official CSV disagrees for 2018–2022.
Panel covers 52.19%–57.14% of source-row revenue, not all British Columbia.
Reported nominal revenue can include one-off/capital amounts; no causal or future-risk claim.

Scroll horizontally to inspect the chart. ↔

Chart and values

Scroll horizontally to see all columns.

Two weighting rules for signed variance contributions
ComponentPrimary pooled (%)Baseline-revenue weighted (%)System median (%)Lower quartile (%)Upper quartile (%)
Local-government support6.8149.647.391.1524.81
Public Libraries Branch grants70.9538.4172.3955.5188.75
Other Revenue22.2411.949.800.6523.39

The two pooled columns are percentages of their respective pooled variance decompositions; median and quartile columns describe individual signed system shares in percent. Primary weights are system variances; sensitivity adds fixed 2018 revenue. Neither is provincial aggregate variance. Rounded displayed percentages may not sum to 100%.

Primary weights: within-system variance. Second: variance × 2018 revenue. Conditional on the pinned workbook; official CSV disagrees for 2018–2022. Panel covers 52.19%–57.14% of source-row revenue, not all British Columbia. Reported nominal revenue can include one-off/capital amounts; no causal or future-risk claim. Source: Province of British Columbia · OGL–BC data · Analysis: Alisson Bento

Download chart (SVG) ↓ · Download chart (PNG) ↓

Medians and interquartile ranges are not a composition

British Columbia · 43 reporting systems · 2018–2024

Full chart and source
Medians and interquartile ranges are not a composition. Dots: medians. Lines: middle 50%, not confidence intervals or full ranges.
Some individual shares are negative or exceed 100%. Medians do not sum to 100%.
Conditional on the pinned workbook; official CSV disagrees for 2018–2022.
Panel covers 52.19%–57.14% of source-row revenue, not all British Columbia.
Reported nominal revenue can include one-off/capital amounts; no causal or future-risk claim.

Scroll horizontally to inspect the chart. ↔

Chart and values

Scroll horizontally to see all columns.

Two weighting rules for signed variance contributions
ComponentPrimary pooled (%)Baseline-revenue weighted (%)System median (%)Lower quartile (%)Upper quartile (%)
Local-government support6.8149.647.391.1524.81
Public Libraries Branch grants70.9538.4172.3955.5188.75
Other Revenue22.2411.949.800.6523.39

The two pooled columns are percentages of their respective pooled variance decompositions; median and quartile columns describe individual signed system shares in percent. Primary weights are system variances; sensitivity adds fixed 2018 revenue. Neither is provincial aggregate variance. Rounded displayed percentages may not sum to 100%.

Dots: medians. Lines: middle 50%, not confidence intervals or full ranges. Some individual shares are negative or exceed 100%. Medians do not sum to 100%. Conditional on the pinned workbook; official CSV disagrees for 2018–2022. Panel covers 52.19%–57.14% of source-row revenue, not all British Columbia. Reported nominal revenue can include one-off/capital amounts; no causal or future-risk claim. Source: Province of British Columbia · OGL–BC data · Analysis: Alisson Bento

Download chart (SVG) ↓ · Download chart (PNG) ↓

Data & downloads

Scroll horizontally to see all columns.

Two weighting rules for signed variance contributions
ComponentPrimary pooled (%)Baseline-revenue weighted (%)System median (%)Lower quartile (%)Upper quartile (%)
Local-government support6.8149.647.391.1524.81
Public Libraries Branch grants70.9538.4172.3955.5188.75
Other Revenue22.2411.949.800.6523.39

The two pooled columns are percentages of their respective pooled variance decompositions; median and quartile columns describe individual signed system shares in percent. Primary weights are system variances; sensitivity adds fixed 2018 revenue. Neither is provincial aggregate variance. Rounded displayed percentages may not sum to 100%.

Scroll horizontally to see all columns.

Signed contributions and influence diagnostics
ComponentLargest contributor: systemsNegative share: systemsShare above 100%: systemsOmission minimum (%)Omission maximum (%)
Local-government support6605.978.15
Public Libraries Branch grants351469.6675.24
Other Revenue210017.1323.41

Counts among 43 retained systems. Leave-one-out ranges are influence diagnostics, not confidence intervals. Largest variance weight 16.39%; no zero-variance systems or largest-contributor ties.

Scroll horizontally to see all columns.

Annual normalized growth: equal-system means versus combined revenue
YearMean local contribution (pp)Mean PLB contribution (pp)Mean Other contribution (pp)Mean total growth (%)Median total growth (%)Combined panel growth (%)
20193.910.102.406.425.193.66
2020−1.263.33−1.310.760.89−5.07
20213.98−3.181.942.751.646.98
20222.2710.661.1414.077.794.86
20233.8632.693.8640.4134.2114.31
20243.23−25.180.09−21.87−19.06−2.19

Mean component contributions are percentage points of total growth and add exactly before rounding across the same 43 systems. Displayed rounded values may not add exactly. Median total is separate; combined panel growth is not an equal-system mean.

Scroll horizontally to see all columns.

Revenue coverage and separately reported reserve availability
YearAll source-row revenue (CAD)Retained revenue (CAD)Retained share (%)Numeric reserve reports: systemsAvailable reserve sum (CAD)
2018261,436,468.00136,452,085.0052.19341,154,824.00
2019267,061,922.00141,444,221.0052.96381,407,286.00
2020254,120,071.05134,268,784.3052.8431696,904.00
2021271,004,410.15143,647,201.0053.01271,190,513.00
2022284,178,682.00150,633,339.0053.01251,927,612.00
2023301,353,579.85172,194,072.8557.14301,979,309.00
2024319,868,135.65168,426,816.7452.66322,926,980.00

71 source rows and 43 retained systems each year. Currency is nominal Canadian dollars. Available reserve sums are partial and never enter the decomposition.

Evidence & method

How we calculated it

  1. Fixed 2018–2024 workbook snapshot; CSV disagreement documented, no period shifting or data merging. Independent dated grant anchors support selected workbook periods, not universal cell validation.
  2. 43-system complete metadata-consistent panel from 71 symbols; no missing-value imputation, identity relinking or outcome-driven exclusion.
  3. Six normalized annual changes per system; exact Fraction covariance decomposition and signed shares. Primary pooling weights by within-system variance; sensitivity adds baseline 2018 revenue.
  4. Inclusive quartiles, medians, largest-contributor/sign counts, largest variance weight and all leave-one-out ranges. Component medians are not additive.
  5. Annual arithmetic mean contributions are additive; median and combined-panel total growth are separate summaries. Explicit source coverage and reserve availability.

What the data cannot tell us

  • Material official-format/year disagreement remains. Results conditional on pinned XLSX values and dates, not a repaired official series.
  • Retained systems represent 52.19%–57.14% of reported source-row revenue. Metadata continuity does not verify unchanged service boundaries; large systems excluded.
  • Broad reported nominal revenue can include capital/one-off amounts and recognition differences. Not recurrent operating resources, uniform cash flows or inflation-adjusted purchasing power.
  • Signed covariance is noncausal; six changes do not establish future risk, adequacy, efficiency or the benefit of diversification.
  • Other is broad; incomplete reserve transfers remain separate, not zero-filled or added to revenue. Alternative weighting answers a different question.

Original sources

  1. BC public-library statistics: selected official workbook

    Pinned workbook; selected 2018–2024 sheets. Filename version 24 differs from internal version 25. Analysis conditional on reported values/year labels, not uniform operating or cash-basis accounting.

    Accessed October 30, 2026

  2. BC public-library statistics: alternative CSV

    Material historical disagreement with workbook 2018–2022; selected numeric fields agree 2023–2024. Not repaired, merged or used as same-period replication.

    Accessed October 30, 2026

  3. BC annual survey questions, 2021

    Local support, PLB grants and Other add to total reported revenue; reserve transfers separate. Local support excludes in-kind and direct municipal costs. Broad categories do not isolate recurrent operating support.

    Accessed October 30, 2026

  4. Burnaby Public Library annual report 2019

    Independent 2018/2019 provincial-grant date anchors. Annual-report accounting totals differ in scope and are not substituted into survey data.

    Accessed October 30, 2026

  5. Burnaby Public Library annual report 2021

    Independent 2020/2021 provincial-grant date anchors, matching workbook at whole-dollar precision; not validation of every system or component.

    Accessed October 30, 2026

  6. Burnaby Public Library 2022 financial disclosure, board agenda 20 April 2023

    Note 4 identifies 2022 COVID-relief receipt of C$122,927 and distinguishes receipts from revenue recognition. Date anchor only; no report images reproduced.

    Accessed October 30, 2026

  7. Funding Stabilization and the Performance of Public Agencies: Evidence From Ohio Libraries

    Enami, Byington and Alm: relevant prior causal research on funding stabilization and institutional performance. Our descriptive covariance/weighting analysis does not estimate those effects or claim first funding-stability research.

    Accessed October 30, 2026

  8. BC dataset catalogue record and licence designation

    Official BC dataset with Open Government Licence–British Columbia designation, corroborated by federal catalogue index. Current question wording is applied across historical columns; headings alone do not prove identical accounting practice.

    Accessed October 30, 2026

  9. Open Government Licence–British Columbia

    Allows copying, adaptation and redistribution with attribution; excludes personal/third-party rights and official marks. Contains information licensed under the Open Government Licence–British Columbia. No endorsement implied.

    Accessed October 30, 2026

Cite & reuse this study

Alisson Bento. Library revenue variation in British Columbia: why the weighting changes the answer. Analysis of selected British Columbia public-library reporting systems, 2018–2024. Sources accessed 30 October 2026. https://alissonbento.com/research/bc-library-revenue-variation-and-weighting-2018-2024/

Contains information licensed under the Open Government Licence–British Columbia. Source: Province of British Columbia, public-library statistics. Original calculations and explanation: Alisson Bento. No official endorsement, agency marks or third-party report images reproduced.

Corrections or editorial questions

Pass it on.

Library revenue variation in British Columbia: why the weighting changes the answer | Alisson Bentoalissonbento.com

WhatsAppFacebookXTelegramRedditThreadsEmailSMS