Summary

  • Regional Internet Registries need liquid reserves because annual fee collection, cyber incidents, legal shocks, banking interruptions and service succession do not arrive on a convenient schedule. The relevant criticism is not that reserves exist, but that a reserve without a named purpose, draw trigger, upper boundary and release rule can finance institutional persistence independently of current member consent.
  • Published figures show the scale without creating a valid league table. RIPE NCC reported audited consolidated equity of EUR 33.663 million at the end of 2025; APNIC's 2024 audited accounts recorded AUD 37.703 million in managed investment funds plus AUD 6.15 million in cash; ARIN's 2025 budget opened with a USD 36.092 million investment-reserve balance. The currencies, accounting classifications, dates and liquidity differ, so adding or ranking these figures would be false precision.
  • The anti-exit mechanism works through time. A large unrestricted buffer lets a board absorb member departures, fee resistance, investment losses, delayed reform and operating deficits without immediately changing scope. That can be prudent during a temporary shock, but it also weakens the feedback that would otherwise force a captive service provider to narrow costs, improve rights or make departure credible.
  • Member-level incidence is uncertain because a legal member, a service account, an LIR account, a voting member and a network ultimately paying through another organisation are different denominators. For RIPE NCC alone, dividing EUR 33.663 million by a sensitivity range of 19,700 to 21,000 accounts produces roughly EUR 1,603 to EUR 1,709 per account, but that is a scale illustration rather than an ownership claim or a distribution entitlement.
  • A defensible reserve rule should separate core continuity from ordinary institutional spending; state stress scenarios and liquid coverage; require automatic review below and above a band; fund tested service handover; and return, rebate or explicitly re-authorise persistent excess. Reserves should make registry functions portable and administrators replaceable, not make exit financially irrelevant to the incumbent.

The balance sheet changes the bargaining table

A reserve is normally described as a shield. It protects payroll, suppliers and essential services when expected income arrives late or an unexpected expense arrives early. For an Internet number registry, the case is stronger than for an ordinary association. Registration records, RDAP and Whois publication, reverse DNS, routing-security repositories, account controls and transfer records cannot simply close for a week while a board arranges emergency finance. Networks and customers rely on stable administrative evidence even when the registry corporation encounters a banking, legal or governance shock.

The shield has another side. Cash and investments also change who must respond when members entity. An institution living close to its current revenue has a short feedback loop: departures, non-renewals, lower demand or resistance to a fee increase quickly force management to reduce costs or persuade payers. An institution with a large accumulated buffer can continue its existing plan while those signals weaken. It can wait for critics to tire, use reserves to bridge a deficit, preserve programmes that current income no longer supports, or finance the legal and administrative cost of defending the structure that members are challenging.

That is the anti-exit subsidy. It is not a claim that money is hidden, stolen or imprudently invested. It is an institutional effect. Past members paid more than the organisation spent, investment returns added to the balance, or assets appreciated. The resulting capital allows the present operator to withstand a degree of present-day withdrawal. Exit ceases to discipline the institution at the speed suggested by the annual fee relationship.

The effect matters most where exit is already weak. A network can change an accountant, conference venue or training supplier. It cannot yet move its complete number-resource registration relationship between competing providers while preserving globally recognised state, reverse-DNS administration, routing-security continuity and transfer history. Regional assignment makes the incumbent unusually difficult to replace.

When financial reserves are added to technical and contractual lock-in, the organisation can become insulated on both sides: members cannot readily leave, and the institution can temporarily absorb the revenue consequences if some do.

A reserve may still be justified. The question is what it buys. Money that buys ninety days to restore banking access is a continuity buffer. Money that buys a tested transfer of registry services is transition capital. Money that indefinitely preserves the same spending scope, powers and corporate shell despite sustained member rejection is an anti-exit subsidy. Governance begins by refusing to call all three uses resilience.

Three published balances, three different facts

The public financial record supplies useful scale, provided that unlike figures are not forced into one ranking.

RIPE NCC's 2025 Financial Report presents audited consolidated equity of EUR 33.663 million at 31 December 2025. Its later reserve policy uses essentially the same amount against a EUR 41.125 million 2026 expense budget and describes it as 9.8 months at the full planned rate. The audited statement establishes the year-end accounting position. The 9.8-month figure is a policy calculation using the following year's budget, not an audit opinion about how long every service would survive.

APNIC's 2024 audited financial report recorded AUD 37.703 million in managed investment funds at fair value and AUD 6.15 million in cash and cash equivalents. It also recorded land and a building within a larger property balance. APNIC's reserve policy treats capital across operating liquidity, property and invested funds, with a contingency objective tied to 18 months of budgeted cash operating expenditure. The audited numbers establish asset values and classifications at year end. They do not make every dollar immediately available or convert property and managed funds into identical crisis cash.

ARIN's 2025 budget showed an opening investment-reserve balance of USD 36.092 million and estimated a year-end balance of USD 35.044 million after earnings and withdrawals. The same page distinguishes a board-approved cash operating and capital budget from United States accounting adjustments. These are transparent planning numbers, not audited 2025 outcomes. The opening balance may be grounded in the prior close, while the earnings and withdrawal path remains a forecast until accounts are completed and audited.

LACNIC's published budgets add a different perspective. Its 2026 budget expects USD 11.766 million of operating revenue and USD 12.156 million of operating expense, with a small planned loss after financial results. Its financial-statement archive makes audited reports available separately. A budget deficit, an audited reserve, an investment portfolio and a cash balance answer different questions. None should be silently substituted for another.

The figures can be placed beside each other only with labels:

Institution and date Published amount Evidentiary status What the figure cannot prove
RIPE NCC, 31 December 2025 EUR 33.663m consolidated equity Audited year-end report Immediate liquidity, core-only burn or member entitlement
APNIC, 31 December 2024 AUD 37.703m managed funds; AUD 6.15m cash Audited year-end report Total same-day reserve cash or a guaranteed return
ARIN, opening 2025 USD 36.092m investment reserves Published budget baseline Audited 2025 close or actual draw and earnings
LACNIC, 2026 USD 87,000 planned loss after financial results Published budget Audited result or reserve adequacy

There is no defensible total across those rows without exchange dates, audited closing values for the same year, classification adjustments and liquidity treatment. Even then, a total would say little about governance. The useful comparison is not which registry has the largest translated pile. It is whether each institution has rules connecting accumulated capital to an identifiable continuity need and a member-authorised upper limit.

A buffer becomes a subsidy through time

The word subsidy often suggests a direct payment. Here it describes a transfer of economic protection. Members fund a buffer that reduces the incumbent organisation's exposure to the consequences of member withdrawal. The organisation receives time; the member gives up immediacy.

Suppose a registry spends 100 units a year and holds 80 units of liquid and near-liquid reserve. If recurring income falls by 10 units because a fee change is rejected, accounts close or members move to lower-cost arrangements, the institution can preserve the full spending plan for years if it is willing to draw the balance gradually. A 10-unit annual signal that would otherwise require immediate reprioritisation becomes a manageable variance. If investment income offsets part of the deficit, the delay can be longer.

This capacity can be socially valuable. A sudden 10-unit fall may reflect a recession, sanctions-related banking blockage or temporary billing failure rather than a verdict on the institution. Immediate staff cuts could damage security and increase long-term cost. Reserves allow a measured response rather than panic.

The same capacity becomes anti-accountability when no rule distinguishes temporary shock from persistent rejection. If revenue remains lower for three years, does the board reduce non-core activity? Does it ask members to re-authorise the scope? Does it publish a plan to restore the target? Does it test whether members should receive lower fees because the reserve remains above need? Or does it describe every draw as protection of stability while changing nothing material?

Time is the scarce governance asset. Management knows that members face participation costs. Network operators have customers, outages, procurement, security incidents and regulatory obligations. A contentious fee or scope debate that stretches across several annual meetings loses entities. Staff and board members remain paid, briefed and continuously present while volunteer critics rotate out. A reserve that finances delay therefore changes not only the balance sheet but the politics of endurance.

This is why a reserve can subsidise the status quo without financing any improper act. It pays salaries during the delay. It funds advisers who explain the existing model. It covers the cost of consultations whose outcome remains discretionary. It protects every activity from the need to prove its priority immediately. The larger and less conditional the balance, the longer management can treat declining consent as a communications problem rather than a binding financial signal.

Continuity is the strongest case, and the narrowest one

The strongest defence of a large reserve is service continuity. It should be accepted before it is bounded.

An RIR collects much of its income on an annual cycle. A shock can occur before invoices are paid, while funds are trapped in a banking channel, or after much of the year's income has been spent. Cyber recovery can require external expertise and replacement infrastructure before insurance responds. A court order, disputed authority or sudden loss of a key supplier can increase legal and technical expenditure. Specialist staff may need retention payments during a transition. A successor operator may need a controlled period of dual running. These are real cash demands.

RIPE NCC's Goals of the Clearing House Reserve is valuable because it names both protection and danger. It sets a 60 to 100 percent band against current budgeted operational expense, with an 80 percent benchmark. It recognises that a reserve can become too high and associate excess with complacency, spending growth, poor investment decisions and attraction to litigation. That acknowledgment should be generalised across the registry system.

Yet continuity cannot mean preservation of every current activity. A service-continuity reserve should start with outputs that running networks cannot safely lose: authoritative records, secure account changes, public registration data, reverse-DNS delegation, RPKI publication and certificate continuity, transfer-state integrity, incident response, evidence preservation and independent dispute handling. It should include the people, systems, vendors and legal capability required to sustain those outputs.

The reserve may also support broader activities in an ordinary downturn, but the claims should not be merged. Meetings, training, grants, measurement platforms, outreach and policy support can be valuable. Their value needs its own argument. If every programme is placed inside the same emergency perimeter as registration integrity, the institution can make its full historical scope financially untouchable. A narrow technical dependency becomes a shield for a wide organisational footprint.

The proper continuity question is therefore not how many months the corporation can remain unchanged. It is how long defined services can operate, how quickly non-essential obligations can be reduced, and how much money is reserved for orderly succession if the present operator cannot recover. The answer may justify a substantial balance. It will not justify an unlimited one.

Institutional survival and service survival can diverge

An incumbent registry naturally writes its continuity plan from inside its own corporate boundary. Payroll, office, contracts, board activity, legal advice and technology appear as components of one operating system. In the short run that is sensible. Services are delivered by a real organisation, not an abstract database.

In the longer run, the boundary can mislead. A company can survive financially while failing to correct a governance defect. It can pay every invoice while members lack an effective remedy. It can maintain websites and repositories while its scope continues expanding beyond the narrow coordination function that justifies compulsory fees. Conversely, essential services can be transferred or replicated while the old corporate shell restructures, shrinks or closes.

The reserve should protect the first transition, not prevent the second possibility from ever being considered. Critical infrastructure is safer when the function is separable from the operator. Replicated data, documented credentials, tested recovery, independent custody and a lawful handover path reduce the amount of capital needed to defend a single institution. A registry that can move its essential functions in three controlled months needs a different reserve from one that assumes the same office must be financed indefinitely because no successor plan exists.

This creates a revealing incentive. Management may prefer a large reserve to investing in replaceability because cash strengthens the incumbent while portability weakens it. A handover plan, external data custody and member exit rights reduce institutional indispensability. A reserve without those controls can be presented as resilience while preserving the very single point of organisational dependence that makes the reserve necessary.

The governance test should reverse that incentive. Reserve adequacy should receive credit for reducing transition time. Every successful restoration exercise, data export, credential inventory and successor rehearsal should lower the uncertainty margin. Every untested dependency, undocumented contract or person-specific control should increase it. The balance then rewards measurable resilience rather than corporate permanence.

Member counts do not produce an ownership ledger

Large balances invite a tempting calculation: divide the reserve by the number of members and call the result each member's money. The calculation can illuminate scale, but it is not an ownership statement.

The denominator is unstable across RIRs. RIPE NCC distinguishes legal members and LIR accounts, and one member may hold more than one account. APNIC includes direct members, different membership tiers and networks served through National Internet Registry arrangements. ARIN distinguishes Service Members and General Members; voting eligibility is not the same as receiving registry services. LACNIC has membership categories connected to resources and fees. A network, a legal organisation, a billing account and a vote can therefore represent different units.

The timing differs as well. A year-end account count is not an average member-month count. Organisations join, close, merge or hold multiple service relationships. Some fees may be delayed by sanctions or banking restrictions. Some resource holders receive services through another institution. A precise reserve-per-member figure can create more confidence than the denominator deserves.

A bounded RIPE NCC sensitivity illustrates the right use. The audited consolidated equity was EUR 33.663 million. If the relevant scale denominator were 19,700 accounts, the balance would equal about EUR 1,709 per account. At 20,000, it would equal about EUR 1,683. At 21,000, it would equal about EUR 1,603. The range changes by roughly EUR 106 per account because the denominator changes by 1,300.

Those figures do not mean each account owns that amount or should receive a cheque. Creditors, tax treatment, contractual obligations, risk pooling and association law matter. The calculation says something narrower: the reserve is of the same order as a substantial fraction of one annual fee for each account, and member-count choices visibly affect the incidence story.

The table should always name its denominator:

Sensitivity denominator RIPE NCC equity per account Interpretation
19,700 about EUR 1,709 Lower account denominator, higher scale allocation
20,000 about EUR 1,683 Planning-scale midpoint, not a verified average
21,000 about EUR 1,603 Higher account denominator, lower scale allocation

For APNIC and ARIN, the same discipline applies more strongly because membership classes and reserve classifications complicate the ratio. The correct report would publish reserve per billed account, per voting-eligible organisation and per service relationship, clearly separated. Where a national registry stands between APNIC and an operator, the report should identify the limitation rather than invent a direct-member equivalent.

Uncertain incidence is itself a governance finding. An institution claiming that reserves belong collectively to members should be able to explain which population financed the accumulation, which population bears replenishment risk and which body can authorise a return or alternative use.

The exit channel is already narrow

In a competitive market, accumulated cash can protect a supplier during a downturn, but customers can still switch. The buffer does not erase demand discipline indefinitely. RIRs operate under a different geometry.

The registry relationship is attached to regional administration, recognised records and surrounding services. A member dissatisfied with cost or governance cannot simply instruct another RIR to assume the same registration while leaving routes, certificates, reverse DNS and transfer history intact. Inter-RIR transfers exist for defined resource movements, not as general service-provider portability. Institutional exit may require closing an account, restructuring resources, transferring assets, changing legal entities or accepting risks unrelated to the quality of registry governance.

That weak exit makes voice more important. Members need elections, consultations, resolutions, financial reports, charging votes, petitions and review rights because ordinary switching is constrained. But voice is costly and often indirect. Boards commonly retain budget adoption and executive authority. Turnout can be low. The economic burden of proving that a programme should shrink falls on dispersed members, while staff possess the data and time needed to defend it.

Reserves widen this asymmetry. If a contested activity can be financed from accumulated capital, management does not need to win immediate willingness to pay. If departures reduce revenue, the institution can draw reserves while arguing that the departures make a stronger buffer necessary. If calls for portability are framed as threats to stability, the reserve then finances resistance to the exit mechanism that would discipline reserve size.

This circularity is the central risk. Lock-in justifies reserves because revenue and continuity are concentrated. Reserves protect the locked-in institution from member pressure. The protected institution has little incentive to create portability. The absence of portability is then cited as evidence that the institution must remain financially invulnerable.

Breaking the circle requires a rule that treats exit capacity as a reserve asset. Money dedicated to data succession, credential handover, dual running and member-directed portability is genuine resilience. Money dedicated only to preserving the incumbent's ability to refuse those changes is not.

Investment income can soften current consent

APNIC and ARIN illustrate another layer. A reserve that earns investment returns can finance part of the institution without current fees. This can reduce fee pressure and protect long-term purchasing power. It can also create an income stream less responsive to present member preferences.

APNIC's audited history shows both sides. Managed funds reached AUD 37.703 million at the end of 2024 after a fair-value loss of AUD 4.03 million in 2022 and later gains. ARIN's 2025 budget estimated earnings on long-term and operating reserve funds while planning withdrawals. Returns and losses are ordinary consequences of investment policy, not evidence of misconduct.

The governance question is how returns enter the social contract. If market gains reduce future fees symmetrically, members benefit visibly from accepting risk. If losses trigger replenishment through higher charges while gains expand programmes or remain invested without an upper limit, the institution receives a one-way option. Current members bear downside through future fees or reduced services, while management gains discretion over upside.

Investment income also blurs the signal sent by departures. A registry losing recurring fee revenue may still report a positive annual result because markets rose. That result does not prove that the spending model retains member support. Conversely, a market loss can make a sound operating model look weak. Reports should separate operating surplus, investment result and reserve draw so members can see whether ordinary activity is self-financing.

The proper metric is structural operating balance before investment returns and planned reserve use. A registry may intentionally run a temporary deficit, but the board should state the duration, cause and exit rule. Persistent dependence on market gains to fund recurring activity turns a continuity portfolio into an endowment. That may be a valid institutional choice only after explicit member authorisation of the purpose and ceiling.

A reserve can finance delay without financing reform

Boards usually say that reserves create room to make orderly decisions. The quality of that claim depends on what happens during the room purchased.

Consider a three-year stress period. In the first year, income falls and the board draws five percent of reserves while commissioning a review. In the second, the same programmes continue because contracts and staff cannot yet be changed. In the third, management proposes a fee increase to restore the reserve. The institution has used member capital to postpone adjustment and then asks members to replenish the capital that funded the postponement.

That sequence may be unavoidable if essential services genuinely required the spending. It becomes anti-accountable if the review never identifies core and optional costs, if no contract calendar is published, and if members never receive a choice between scope reduction and replenishment. The reserve has protected management's preferred baseline from comparison.

A reform reserve should work differently. The first draw would trigger publication of the shock, the amount and the services protected. The second would activate cost and contract review. A persistent draw would require a member decision on the future scope. Capital used for transition would carry milestones: data replication, service separation, supplier exit, staffing changes, fee redesign or portability testing. Missed milestones would narrow board discretion rather than produce a larger request.

This approach respects the practical difficulty of changing institutions. Employment law, leases, security obligations and vendor contracts cannot be unwound overnight. It also prevents those commitments from becoming a permanent answer. The reserve buys an orderly path from old obligations to a member-authorised structure; it does not erase the destination.

Too little reserve creates its own captivity

The anti-exit critique can be misunderstood as a demand to empty the treasury. That would make members more vulnerable, not more powerful.

An underfunded registry facing a shock may need emergency support from a government, large member, supplier, peer registry or global coordination body. The rescue provider can acquire influence over policy, staffing, data custody or timing. Small members may have even less voice. A cash-starved organisation can threaten service interruption to obtain a rapid fee increase. It can sign expensive short-term contracts, defer security work or accept conditions that a well-funded board would reject.

Adequate reserves therefore protect independence. They allow the registry to resist a donor with political conditions, a creditor demanding control or a major member seeking preferential treatment. They let the institution preserve the last verified state during litigation rather than monetise disputed assets. They fund retention of technical staff while authority is clarified.

The target is not minimum cash. It is minimum captivity. Excess reserves can make the institution captive to its own perpetuation; limited public evidence reserves can make it captive to a rescuer. A sound policy finds a band within which neither management nor an emergency financier gains unreviewable leverage.

That band must be based on scenarios. A bank-access delay may require weeks of liquidity. A cyber compromise may require an immediate surge. A prolonged income loss may require months of core operations and a restructuring budget. A service transfer may require parallel systems and legal work. Each scenario needs an amount, probability range, activation condition and recovery plan. Adding the largest cost of every scenario would overstate need because some risks overlap; assuming none coincide would understate it.

The board should publish a base case, a severe but plausible combined case and a reverse stress: what event exhausts liquid coverage before a successor can operate? The reverse stress is especially useful because it exposes whether the institution relies on hope, cash or tested replaceability.

The reserve needs a release valve

A minimum without a maximum protects accumulation. A maximum without an automatic response merely starts another discussion. An enforceable reserve policy needs both boundaries and actions.

Below the minimum, the board should present a replenishment plan that distinguishes cost reduction, fee change, recovery of receivables, insurance and asset reallocation. Essential services should be protected first. The plan should not assume that every current activity survives until members pay more.

Within the band, management should report liquidity, risk uses, investment performance and changes to stress assumptions. Ordinary fluctuation should not trigger annual political theatre. The whole purpose of a band is to allow disciplined discretion.

Above the maximum, the excess should follow a default rule. Options include a fee rebate, lower future charges, a one-time defined capital project, a separately approved transition fund or a member vote to retain the amount for a named risk. Doing nothing should not be the automatic choice. Otherwise the maximum becomes a descriptive line that reserves can cross whenever management develops a new rationale.

RIPE NCC's policy is notable for requiring a reduction or redistribution plan above its upper point and a replenishment plan below its lower point. The stronger version would add deadlines, member decision rights and treatment of repeated near-boundary years. It would also define whether the comparison uses audited year-end equity, liquid reserve assets or another measure.

For APNIC and ARIN, an upper rule is especially important where invested capital can grow through returns. A percentage of annual expense can provide a moving band, but it creates a ratchet: spending growth raises the permitted reserve, and the larger reserve can make spending growth easier. The policy should use both an expense-based measure and an independently defined core-continuity need. If full organisational expense rises while core service cost remains stable, the maximum should not automatically validate the larger institution.

Purpose labels must change decisions

It is useful to describe parts of capital as operating liquidity, contingency reserve, long-term reserve, transition fund or invested surplus. Labels alone do not constrain use.

APNIC's investment policy improves clarity by separating a Contingency Reserve Fund from an Operational Reserve Fund and by giving them different horizons and return objectives. RIPE NCC defines a reserve band and enumerates scenarios. ARIN distinguishes operating and long-term investment funds in member presentations. These structures are more informative than one cash total.

The governance test is what happens when money moves. Who may approve a draw? Which purposes qualify? Is a draw reported gross or net of investment gains? Does use for ordinary deficit automatically require a corrective plan? Can a board relabel a recurring programme as continuity? Does capital reserved for handover remain available to defend the incumbent in litigation about handover?

A strong policy would require a decision record for every material draw: amount, authority, scenario, service protected, alternatives considered, expected duration and replenishment or release consequence. Confidential legal or security detail can be protected while the financial purpose remains public. An annual audit can verify classification and authorisation without deciding whether the policy choice was wise.

This article does not attempt the next accounting question of which assets are legally restricted, donor-restricted or immediately available. That requires close reading of each financial statement and legal instrument. The present point is narrower: even economically available capital should be governed by purpose. Unrestricted in accounting terms should not mean unbounded in institutional terms.

The member decision must be capable of saying no

Consultation can improve a reserve policy, but accountability requires a decision that changes the outcome.

Members should not vote on individual trades or daily cash management. Those tasks require professional execution. They should authorise the constitutional envelope: purpose, minimum, maximum, liquidity floor, maximum stress loss, valid draw categories, excess-capital rule and transition obligation. Boards and treasury committees can then operate within it.

The mandate must be withdrawable. If changing the reserve policy requires a supermajority of the entire membership, while keeping the policy requires only board inaction, inertia favours accumulation. If members can approve a fee increase but cannot direct use of excess capital, they fund the reserve without controlling its destination. If low turnout is treated as consent, the institution can claim member ownership of a decision few members observed.

A periodic affirmative vote solves part of the problem. Every three years, the policy could lapse into a conservative default unless members re-authorise the band and purposes. Material expansion beyond core continuity could require a separate resolution. The board would publish scenario evidence at least ninety days before the vote and respond to alternatives using the same cost model.

The electorate also matters. The population paying fees may not match the population holding votes. One organisation may hold multiple service accounts but one vote, or voting eligibility may depend on status and designation. National registry arrangements can separate the operator bearing cost from the direct institutional member. The reserve report should show incidence across those groups rather than presenting one undifferentiated community.

No vote can manufacture exit. It can, however, stop reserves from substituting for it. A revocable mandate turns accumulated capital back into an instrument held on defined terms rather than a permanent source of board autonomy.

A practical anti-subsidy test

Members and boards can test a reserve against eight questions.

First, is the protected service named? “Stability” is too broad. The rule should identify records, publication, security, account control, dispute preservation, staffing and succession outputs.

Second, is the amount connected to a reproducible scenario? A percentage of the full budget is a starting point. The calculation should also show core burn, liquidity timing, one-time shock cost and transition duration.

Third, is the denominator honest? Audited balances, budget forecasts and fair-value investments must remain distinct. Member, account and voter counts must be labelled, with sensitivity ranges where a single denominator is unavailable.

Fourth, does a draw trigger change? Using reserves for an ordinary deficit should activate cost review, milestones and member decision. It should not simply postpone the same request.

Fifth, is there an enforceable upper boundary? Persistent excess should reduce fees, fund a named one-time purpose or return for explicit authorisation.

Sixth, does the reserve fund replaceability? Data custody, tested restoration, credentials, dual running, dispute notation and lawful service handover should be budgeted as continuity assets.

Seventh, can members revoke the mandate? A policy that can be expanded by the board but narrowed only through an impractical threshold is structurally one-way.

Eighth, does exit affect management before reserves are exhausted? Persistent member loss or fee resistance should trigger review well before financial distress. The board should not wait until the buffer is gone to accept that the signal was real.

A reserve that passes these tests can be large without being anti-accountable. A reserve that fails can be modest and still subsidise lock-in. The problem is not a universal number; it is the relationship between money, purpose and consequence.

What the auditors establish, and what members must decide

Audited financial statements are indispensable. They establish whether reported assets, liabilities, income, expense and fair-value measurements are presented under the applicable accounting basis. They can expose investment losses, operating deficits, commitments and changes in equity. They make it harder for management to invent a balance.

An audit does not establish the proper size of a reserve. It does not decide whether a conference programme belongs inside compulsory funding, whether 18 months is better than nine, whether portability should replace some cash insurance, or whether excess capital should reduce fees. Those are governance judgments.

Budgets answer another question: what the board expects and authorises for the coming year. A budgeted reserve withdrawal is not an audited withdrawal. Estimated earnings are not realised gains. An approved expense ceiling is not proof that every amount was spent. Articles about registry finance should keep those statuses visible because false precision often begins by combining a confirmed opening balance with forecast returns and an uncertain member denominator.

The public record used here supports several firm conclusions. RIPE NCC, APNIC and ARIN hold capital large enough to influence more than short-term cash management. Their published documents recognise continuity, investment and governance considerations. The balances have different liquidity and accounting meanings. Member populations do not supply one comparable denominator. Reserve policies vary in how explicitly they address upper limits, draw purposes and member control.

The record does not support a claim that any named reserve is currently excessive in every plausible scenario. Protected contract terms, cyber-recovery estimates, insurance, banking access, staff obligations and service-handover costs are incomplete in public. It also does not support a claim that every accumulated unit is necessary. That uncertainty is the reason for a scenario-based mandate, not a reason to defer to an unexplained total.

Resilience should make the incumbent less necessary

The best reserve is partly self-liquidating. It finances changes that reduce the amount of institutional insurance required next time.

A cyber reserve should fund controls and restoration that lower future recovery cost. A banking reserve should support diversified, lawful payment channels. A legal reserve should preserve independent adjudication and the last verified operational state rather than reward escalation. A transition reserve should build data and service portability. A revenue reserve should fund a measured cost adjustment rather than maintain a permanent structural deficit.

Under this standard, success is not the preservation of a large balance. It is the reduction of unmanaged dependency. The registry becomes easier to audit, easier to restore, easier to replace and less able to convert technical reliance into unrelated authority. Members gain a credible exit even if they rarely use it. The threat of departure disciplines service and scope before a crisis.

That approach aligns treasury prudence with a narrow coordination mandate. The registry still holds enough liquid capital to protect uniqueness, accuracy, security and running networks. It does not treat accumulated fees as an endowment for indefinite expansion. It does not confuse the continuity of the ledger with the continuity of every institutional choice made around it.

Cash reserves become an anti-exit subsidy when they remove consequence from member dissatisfaction while leaving the exit barrier intact. They become continuity capital when they preserve services, shorten recovery, fund portability and return excess authority to the people bearing the cost.

The distinction can be written into policy. Name the service. Publish the scenario. Label the accounting status. Show the member-count sensitivity. Set the lower and upper actions. Require a decision for persistent excess. Rehearse succession. Make the administrator replaceable before asking members to finance its permanence.

That is not hostility to reserves. It is the discipline that makes a reserve defensible.

Evidence and limits

The audited baseline for RIPE NCC is its 2025 Financial Report; the policy conversion and reserve band come from Goals of the RIPE NCC Clearing House Reserve. The first establishes the year-end accounting position, while the second supplies a board-and-member governance model using the 2026 budget.

APNIC's asset values, cash, fair-value investments and 2022-2024 market movements come from its 2024, 2023 and 2022 audited reports. Its governance authority and budget roles are described in the APNIC By-laws and the Executive Council description.

ARIN's planning figures and accounting adjustments come from the 2025 Budget. Its audited reports are listed in the annual-report archive, and its 2024 investment presentation separately shows asset allocation and reserve reporting. LACNIC's 2026 Budget is used only as a planned income-and-expense comparator; its financial-statement archive is the source class for audited results.

No cross-currency total, ranking of institutional prudence, legal ownership claim or exact reserve-per-member entitlement is asserted. The sensitivity range illustrates denominator risk. A complete adequacy conclusion would require current liquidity ladders, contractual commitments, insurance terms, average member-month counts, restricted-fund treatment, stress costs and tested handover durations that are not fully disclosed in the cited public material.