This study analyses English and Welsh charity financial data from the Charity Commission's Public Register of Charities, specifically submissions from five years of recent annual return cycles (2020 to 2024). The data extract was downloaded on XX April XXXX.

Of the public register, three data extracts are used: Charity Annual Return History, Charity Annual Return Part A, and Charity Annual Return Part B. From the Charity Annual Return History extract 'organisation_number', 'registered_charity_number', and 'charity_name' were selected. From the Charity Annual Return Part A extract 'organisation_number', 'ar_cycle_reference', 'total_gross_income', 'total_gross_expenditure', 'income_from_government_contracts', and 'income_from_government_grants' were selected. From the Charity Annual Return Part B extract 'organisation_number', 'ar_cycle_reference', 'income_donations_and_legacies', 'income_other_trading_activities', 'income_charitable_activities', 'income_investments', 'income_other', 'income_endowments', 'reserves', 'funds_endowment', 'funds_unrestricted', 'funds_restricted', and 'funds_total' were selected.

Charity Annual Return Part A and Charity Annual Return Part B were inner joined on 'organisation_number' and 'ar_cycle_reference' before inner joining the merged dataset to Charity Annual Return History on 'organisation_number'. The combination of Part A and Part B data is methodologically significant beyond its role in data validation. Part A records income as classified by statutory funding relationships — government contracts and grants as distinct instruments — while Part B records income as classified by the organisation according to its operational purposes under SORP. Where these classifications differ for the same organisation, the gap captures the institutional hybridity that single-source studies cannot observe. Specifically, the degree to which government-commissioned income has been absorbed into an organisation's core operational identity as charitable activities income rather than recorded as a distinct statutory funding relationship. The government axis therefore measures statutory dependence as funders classify it, while the market axis captures delivery income as organisations classify it. The structural gap between these two representations, visible only through the Part A and Part B combination, is the empirical marker of what Billis (2010) calls entrenched hybridity.

Any charity with less than £500,000 'total_gross_income' in any annual return cycle was excluded from the dataset, producing an initial sample of 21,479 charities. The £500k threshold was chosen for both regulatory and theoretical reasons. In regulatory terms, it aligns with Charity Commission requirements for full SORP compliance. In theoretical terms, it identifies organisations with sufficient strategic capacity to make deliberate funding portfolio decisions rather than relying on opportunistic resource acquisition. This threshold introduces a risk of selection bias; however, it is defensible insofar as organisations below this level typically lack the professional infrastructure, dedicated fundraising staff, and strategic planning processes required for portfolio management behaviour. The threshold also ensures institutional visibility as charities above £500k income are more likely to interact with similar donors, face comparable regulatory demands, and compete in overlapping resource markets, thereby occupying a shared institutional environment suitable for meaningful archetype comparison.

To validate the accuracy of income data between Part A and Part B, any charities with a variance between 'total_gross_income' in Part A and the sum of 'income_donations_and_legacies', 'income_other_trading_activities', 'income_charitable_activities', 'income_investments', and 'income_other' was excluded. Similarly, to validate accuracy of fund composition filings within Part B, any charity with a variance between 'funds_total' and the sum of 'funds_restricted', 'funds_unrestricted', and 'funds_endowment' was excluded. This produced a refined dataset of 20,695 charities (3.6% exclusion rate, n=784).

After filtering the refined dataset to include only submissions for the annual return cycles between 2020 and 2024, including only those with a submission in each of the five cycles, the final dataset was reduced to 5,503 charities.

A four axis model was built for Donation, Government, Investment, and Market income. For each axis, the component variables were divided by 'total_gross_income' to produce a proportional income measure.

Donation was calculated by subtracting 'income_endowments' from 'income_donations_and_legacies'. Government was calculated as the sum of 'income_from_government_contracts' and 'income_from_government_grants'. Investments used 'income_investments'. Market was calculated as the sum of 'income_other_trading_activities' and 'income_charitable_activities'.

Endowments were excluded from the donations axis because they represent capital transfers rather than ongoing income streams. Unlike regular donations that can be immediately used for programme activities, endowments are typically restricted to capital preservation with only investment returns available for operations. Organisations receiving large endowments may appear donation-dependent in accounting terms but face fundamentally different strategic constraints than those relying on annual giving campaigns or foundation grants.

A related classification issue concerns the boundary between the government and market axes within the Charity Commission’s income reporting framework. The Charity Commission’s data architecture records 'income_from_government_contracts' and 'income_from_government_grants' as distinct fields in Part A of the annual return, while 'income_from_charitable_activities' is recorded in Part B as part of the Statement of Financial Activities. Income items are recorded only once within a charity’s accounts, but the same income stream may be represented across Part A and Part B under different classificatory logics. A commissioned service is recorded once for accounting purposes, but may simultaneously contribute to multiple analytical dimensions when Part A and Part B data are combined. In particular, where public funding is structured and interpreted as a contract for services (including many service‑level agreements) the resulting income is commonly classified as income from charitable activities under SORP.

Whether the same income is recorded in the government contracts field depends on the extent to which the counterparty is recognised as “government” by the filing charity. In the case of arm’s‑length public bodies and QUANGOs, there may be inconsistencies in how this classification standard is applied across charities. The government axis is therefore likely to understate statutory dependence among contracted service providers. This asymmetry runs predominantly in one direction and reflects a structural feature of UK third‑sector funding rather than a flaw in data quality or reporting compliance. As Carmel and Harlock (2008) document, procurement mechanisms have progressively blurred the boundary between statutory and market income at the level of organisational practice, and this blurring is reproduced in how charities classify their financial returns.

A series of yearly mean proportional income measures were calculated for each axis across the five annual return cycles, producing 20 year‑specific classification thresholds. Each submission (n = 27,515) was scored relative to the corresponding annual mean for each axis. Submissions with axis values strictly greater than the mean were classified as "H", while those at or below the mean were classified as "L", reflecting the analytical choice to reserve high classifications for organisations exhibiting above‑average income concentration on a given axis.

Measuring strategic capacity

TBC

Considerations

The analysis compares positions rather than organisations, identifying systematic patterns in how funding structure shapes strategic capacity. The analytical logic follows directly from the strategic selectivity framing: if funding environments are selectively structured, we should observe that strategic capacity varies not randomly across organisations but in patterns that correspond to structural position. The test is whether organisations in similar funding positions face similar capacity constraints regardless of their size, sector, or management quality — and whether those constraints are severe enough, and widespread enough, to constitute a structural feature of the funding landscape rather than a collection of individual organisational situations.

 

OLD

Organisations were classified using a binary high/low system on each axis, with thresholds set at the sample mean: donations (32.5%), government (20.6%), investments (6.2%), and market activities (57.8%). Organisations scoring above the mean on an axis were coded as "H" (high relative dependence), those below as "L" (low relative dependence). The decision to use sample means rather than sector-specific thresholds reflects the focus on cross-sector institutional environments: the strategic implications of being "high" or "low" relative to the overall field of large charities remain consistent across sectors, enabling meaningful comparison between organisations facing similar constraint patterns regardless of cause area.

The binary classification reflects a theoretical commitment consistent with the strategic-relational framework. What matters for strategic selectivity is not the precise degree of dependence on a funding stream but whether that stream is dominant enough to impose its characteristic accountability pressures and stakeholder demands. An organisation deriving 35% of income from government contracts faces qualitatively different strategic pressures than one deriving 5%, regardless of where either sits on a continuous diversification index. Continuous measures capture the quantity of diversification, but Know Your Ground is designed to capture the qualitative strategic environment that funding composition produces.

### 3.3 Measuring strategic capacity

Beyond funding composition, two historical indicators capture organisational capacity to navigate resource constraints over time.

The Reserves-to-Income Ratio was calculated as total reserves divided by annual gross income, expressed in months. Organisations above the sample mean (17.2 months) were classified as "High Reserves," indicating financial cushioning and risk tolerance. Those below were classified as "Low Reserves," suggesting vulnerability and operational urgency. These are proxies for strategic stability — the capacity to endure disruption and plan beyond the immediate funding cycle.

Unrestricted Funds was calculated as unrestricted funds divided by total funds. Organisations above the sample mean (79.4%) were classified as "High Flexibility," indicating strategic autonomy. Those below were classified as "Low Flexibility," indicating programmatic constraints from donor restrictions. These are proxies for strategic discretion — the capacity to act on opportunities rather than simply fulfil funder requirements.

Together these measures create four strategic capacity profiles that interact with funding archetypes to define the actual strategic space available to organisations at any given position in the funding structure.

References

The analysis compares positions rather than organisations, identifying systematic patterns in how funding structure shapes strategic capacity. The analytical logic follows directly from the strategic selectivity framing: if funding environments are selectively structured, we should observe that strategic capacity varies not randomly across organisations but in patterns that correspond to structural position. The test is whether organisations in similar funding positions face similar capacity constraints regardless of their size, sector, or management quality — and whether those constraints are severe enough, and widespread enough, to constitute a structural feature of the funding landscape rather than a collection of individual organisational situations.

You can find a plain English summary of the approach here. This page provides the full technical methodology (including variable construction, statistical analysis, and academic references).