APNIC and the economics of fees, reserves, and incentives

The unavoidable invoice

The smallest Internet operators often learn the economics of regional number administration through a bill that does not look political. A rural broadband provider, a university network, a small data centre, an island ISP, a bank running its own autonomous system, or a cloud operator with address holdings may treat the APNIC invoice as another cost of doing business. It is not a spectrum licence. It is not transit. It is not a tax passed by a parliament. It is not the price of a domain name or a router port. Yet it is attached to one of the more consequential ledgers in the Internet economy: the recognised registry record for Internet number resources, the account standing needed to change that record, the administrative basis for transfers, the ability to maintain reverse DNS, and the authority behind routing-security statements.

That is why APNIC's fees deserve more scrutiny than an ordinary membership schedule. A regional Internet registry sits at the junction of commons administration and private network capital. It does not own IPv4 blocks, IPv6 prefixes or autonomous system numbers in the way a landlord owns a building. But it operates the ledger through which recognised holding, transfer, certification and operational metadata are made legible to other networks. For most members there is no practical substitute. If the registry is weak, the region carries systemic risk. If the registry is too comfortable, members carry a different risk: a necessary institution may begin to price its own indispensability.

The central economic tension is simple. Reserves are insurance, and reserves are temptation. They are insurance because APNIC administers infrastructure that cannot be allowed to fail when revenue falls, a bank counterparty stumbles, a system is compromised, a natural disaster disrupts operations, a legal dispute arises, or a political conflict threatens the integrity of the registry. A registry with no buffer would be a bad bargain for members. It would save money until the year in which saving money became expensive.

But reserves are temptation because accumulated member money changes behaviour. A large cushion can dull the pressure to choose between core registry operations, regional programmes, travel, research, fellowships, legal defence, institutional diplomacy and new systems. It can turn a temporary surplus into a permanent budget base. It can make fee increases feel technically necessary when they are partly choices about institutional scope. It can finance prudent resilience, but it can also finance inertia. The same reserve that protects the ledger from crisis can protect the institution from its own members.

This is not an argument for austerity. The Asia-Pacific region is too large, too varied and too exposed to operational shocks for a thinly capitalised registry. APNIC needs secure systems, skilled staff, resilient publication infrastructure, legal competence, election machinery, member support, policy coordination, and the boring redundancy that keeps critical records available when ordinary assumptions fail. The case for a reserve is strong. The case for a reserve without a clearly understood target, composition and release rule is not.

Fees have a parallel ambiguity. A fee can be a legitimate price for maintaining a common facility. It can also become a quasi-tax when it funds broad activities only loosely tied to the payer's use of registry services. In the IPv4 scarcity era it can become still more sensitive: a charge attached to holdings or transfers may operate as a toll on scarce operational capital. APNIC must verify transfers, collect dues and protect the integrity of records. It must not allow billing, account standing or transfer charges to become a private capital-control system that affects liquidity and operational continuity beyond what registry stewardship requires.

The legitimacy bargain is therefore narrower than APNIC's public importance and broader than accounting compliance. Members accept compulsory collective funding because they need a disciplined, durable and neutral registry. In return the registry must show what is core, what is discretionary, who pays, who is protected, what risks are insured, how reserves are measured, when fees rise, and when accumulated money should return to members through lower charges or better-defined resilience. The institution's fiscal question is not whether it is useful. It is how to price usefulness without converting stewardship into leverage.

What the fee schedule really prices

The visible architecture of APNIC's revenue begins with its member fee schedule. It sets a sign-up fee, an annual fee for accounts without chargeable resources, formula-based annual fees for members holding chargeable IPv4 or IPv6 resources, charges for additional autonomous system numbers after the first two, temporary assignment charges, transfer fees, and reactivation fees after termination for non-payment. The 2026 schedule lists a base fee of AUD 1,295, rising over the following years to AUD 1,487 in 2029. Annual fees for address holdings are calculated by formula, with IPv4 and IPv6 assessed separately and the higher result applying. Transfers are charged at 20% of the annual fee applicable to the resources being transferred, subject to stated exceptions.

The schedule is neither a flat club subscription nor a simple marginal-cost tariff. It is a resource-sensitive system. Larger holders pay more, and APNIC's membership tiers also give larger members more voting weight. The formal tiers run from Associate through Very Small, Small, Medium, Large, Very Large and Extra Large; voting entitlements rise from one vote for Associate members to 64 votes for Extra Large members. The design has an understandable logic. Larger resource holders have more at stake, make greater use of the registry's recognised records, and contribute more to common costs. But the design also embeds a political economy. Those best able to absorb fees receive greater formal weight. Those most sensitive to fixed charges often have less institutional capacity to study accounts, participate in meetings, draft proposals or organise a counterweight.

A service fee is legitimate when members can see the service being funded. APNIC maintains registration data, member accounts, request processing, resource transfers, reverse DNS delegations, RPKI services, Whois and RDAP functions, policy processes, technical systems, helpdesk capacity, security controls, audit and legal support. These are real costs. Accuracy and continuity are more important than transaction volume, which means a narrowly efficient registry would still not be cheap. A small mistake in the ledger can have consequences far beyond the time required to make the mistake.

The harder question is where service pricing ends. APNIC also operates in a region where training, technical assistance, measurement, community participation, development support and global coordination have public-good characteristics. These activities may improve routing hygiene, security awareness, policy participation and operational capacity. They may be worth funding. Yet they are not identical to keeping the authoritative number-resource ledger secure and accurate. When they are funded from the same fee base, members should be able to see the difference.

This distinction matters because fees are sticky. A member can complain, vote, attend meetings, work through a national registry where one exists, or transfer resources in limited cases. It cannot buy the same APNIC-administered number resources from a rival regional registry. The registry monopoly is not a commercial abuse; it is a functional necessity. There must be one recognised ledger. But the absence of market exit means fee discipline must come from governance, disclosure and credible appeal.

The IPv4 scarcity era adds another layer. When fresh IPv4 allocations were the dominant story, a resource-based fee could be understood mainly as a way to fund distribution and registration. Today IPv4 holdings have market value independent of the registry. Blocks are bought, sold, valued in acquisitions, leased, moved between regions and treated as balance-sheet assets. The registry's entry is not the whole asset, but it is a crucial part of the asset's usability. A charge on holding or transferring such resources therefore has a capital-market effect even if it is legally framed as a membership or service fee.

This does not make APNIC's fee formula illegitimate. It does mean that economic candour becomes more important. If larger IPv4 holders are expected to contribute more because they hold scarce assets whose value depends on the common ledger, that is a policy choice. It may be fair. It may be progressive. It may be the least bad way to fund a regional utility. But it should not be hidden inside a formula that appears purely technical. Scarcity rents are real; charging against them should be acknowledged as a distributional decision, not merely as registry housekeeping.

The same is true of transfer charges. A fee attached to a transfer can be defended as cost recovery: staff review, compliance checks, documentation, account validation, registry updates and risk management all require work. But a percentage of the annual fee applicable to the resources being transferred will be read differently by different members. For a large transaction involving valuable IPv4 space, the charge may be small relative to the private value exchanged. For a small entrant buying a modest block, it may be another barrier on top of broker fees, legal advice, renumbering, routing changes, financing and future annual dues. The question is not whether transfers should be free. It is whether the fee reflects work, risk and agreed cross-subsidy rather than the registry's privileged position near a scarce asset market.

The ledger that must not fail

The strongest argument for APNIC's fee base is the simplest: the ledger must be funded. Internet number resources are useful because others can verify the recognised holder, the relevant contacts and the operational information associated with the resource. Peers, customers, transit providers, security systems, auditors, brokers, counterparties and network operators all rely on registry data to make decisions. APNIC is often described as an allocator and registry. In economic terms it operates a title-adjacent record for a class of operational capital.

That record has several layers. The public registration layer identifies resources and account holders. The operational layer supports reverse DNS, domain and route objects, portal functions, requests and updates. The security-attestation layer supports RPKI certificates and route origin authorisations, where APNIC's registry authority underwrites cryptographic statements about who may originate a prefix. The transaction layer supports transfers, mergers, acquisitions, historical resources and inter-regional movements. The governance layer supports the policies that define what can be delegated, transferred, reclaimed or certified.

Underfunding such a system can be a false economy. Members might see a lower invoice this year, while the region inherits risk in the form of weaker security, slow transfer processing, stale records, fragile RPKI publication, poor disaster recovery, staff attrition, inadequate support, or vulnerability to better-funded litigants. Some registry functions are difficult to value until they fail. Reverse DNS continuity is not glamorous. Neither is certificate publication, access control, backup, incident response, payment reconciliation, audit or legal review. Yet each can matter precisely when members have the least patience for institutional improvisation.

The cost base should therefore include insurance-like spending. Cyber-security readiness is not merely an IT expense. Disaster recovery is not waste because no disaster occurred last year. Legal capacity is not only for lawsuits already filed. Staff retention is not bureaucracy if the alternative is operational fragility. RPKI and reverse DNS make continuity a service, not a slogan. A registry that cannot finance emergency operations without a sudden levy on members is not prudent.

The ledger argument, however, cannot justify everything by proximity. A programme may be good for the regional Internet and still not be a core registry expense. A fellowship may broaden participation. A training programme may improve operational practice. A measurement project may generate useful evidence. A global coordination activity may help APNIC represent regional interests. The issue is not whether such work has merit. The issue is whether members can distinguish the insurance premium for the ledger from the contribution to a broader institutional mission.

Accounting categories alone do not make that distinction. Employee benefits, software, professional fees, travel, communications, meetings, occupancy, insurance and donations are normal line items. They do not tell members which costs maintain the registry, which protect RPKI, which support reverse DNS, which fund policy development, which support training, which relate to global coordination, which are tied to externally funded projects, and which reflect discretionary expansion. A budget can be accurate and still economically opaque.

The right disclosure follows the economics of the institution. Members considering a fee increase need to know whether the pressure comes from security hardening, registry modernisation, legal exposure, inflation, currency movement, staff growth, programme expansion, or reserve rebuilding. Each has a different claim on member consent. Security may justify a rapid response. Programme growth should face a more explicit debate. Inflation may justify indexation but not an unexamined expansion of scope. A legal shock may justify temporary funding and reserve use, but it should require post-crisis explanation.

This is where the Executive Council's role becomes central. APNIC's governance documents place budget authority and financial stewardship in member-controlled structures, with the Executive Council establishing the basis for budgets and expenditure ceilings, and the Treasurer carrying financial responsibilities with staff support. The Council also has discretion to reduce or waive fees for deserving organisations. These are meaningful controls. Their effectiveness depends on whether members are given information in a form that maps onto actual choices.

A disciplined council would treat the budget as a control surface, not a presentation. It would ask what the minimum viable cost of the authoritative registry is; which services are required by policy, contract or security necessity; which activities are discretionary but high-value; which are legacy; which have sunset conditions; which are funded by the APNIC Foundation or other restricted sources rather than member fees; which costs are rising faster than service demand; and what would be cut before fees rose or reserves were drawn down. The point is not to make the Council hostile to staff or programmes. It is to prevent the institution's usefulness from becoming an all-purpose fiscal argument.

Reserves as insurance

The reserve question is more subtle than the fee question because reserves are member money not spent in the year it is collected. In a commercial company retained earnings may fund growth, protect creditors or increase shareholder value. In a member-funded registry reserves are different. They are deferred member contributions held for continuity, risk and independence. They are not improper by nature. They are necessary if APNIC faces tail risks that cannot be financed quickly through annual invoices.

APNIC's latest public financial statements show why the issue is material. The 2025 financial report records revenue from contracts with customers of about AUD 29.1m, other income of about AUD 5.0m and finance income of about AUD 267,000. Employee benefits were the largest expense at about AUD 20.9m. APNIC Foundation funded project expenses were about AUD 3.7m. Software application expenses were about AUD 2.2m, professional fees about AUD 1.8m, travel about AUD 1.6m, and communication and meeting expenses about AUD 1.1m. At year-end APNIC had total assets of about AUD 64.2m, total liabilities of about AUD 22.1m and net assets of about AUD 42.1m. Cash and cash equivalents were about AUD 7.35m, while current and non-current financial assets together were about AUD 41.9m. Retained earnings were about AUD 38.5m, with a separate asset revaluation reserve of about AUD 3.57m.

Those figures do not prove excess. They prove that reserves are no longer a minor footnote. A member-funded registry with roughly AUD 42m in net assets is not living hand to mouth. Liquid and investment assets appear large relative to annual operating expenditure. That may be prudent. It also creates a cushion large enough to shape incentives. The relevant question is not whether the balance is morally too high. It is whether members can see a reserve policy that explains the target, the risk model, the liquidity profile, the conditions for use, and the conditions under which fees should fall or discretionary spending should be constrained.

A serious reserve policy for a registry-layer utility needs buckets. The first is operating continuity: enough liquid money to pay staff, hosting, facilities, insurance, essential suppliers and member services if revenue is interrupted. The second is disaster and security recovery: major incident response, forensic work, emergency infrastructure, parallel operations, system rebuilds and continuity after a natural disaster. The third is legal and governance risk: defence of registry authority, transfer disputes, election disputes, contractual matters, privacy and data claims, indemnities and proceedings in which the registry's institutional capacity matters. The fourth is strategic transition: planned investment in systems that reduce long-term risk, such as registry modernisation, RPKI resilience, automation of routine requests, stronger access controls and better assurance over records.

Each bucket needs a rationale and a ceiling. Without ceilings, every risk becomes an argument for more accumulation. One can always imagine a larger cyber incident, a longer legal fight, a worse natural disaster, a deeper revenue shock, a sharper currency movement, or a more expensive modernisation. Prudence then becomes unfalsifiable. Members are told that money is held for resilience, but no one can say when resilience is funded.

The reserve-to-expense ratio is the most important fiscal watchpoint. A few months of expenditure may be thin for an institution that cannot fail. Around a year may be prudent, depending on liquidity and risk appetite. More than that may still be defensible if APNIC has disclosed contingencies, planned capital projects or legal exposure. But a high ratio should trigger harder questions. Are restricted, designated and unrestricted reserves separated? How much is available in cash or near-cash? How much is invested in assets that could lose value or cannot be accessed quickly? Are unrealised gains treated like spendable money? Are members being asked to pay more while reserves already exceed the stated target? Is the target measured against core registry expenditure or total organisational expenditure?

The denominator is not a technical detail. If reserves are measured against total expenditure, programme expansion can justify a larger reserve, which then makes further expansion easier. If reserves are measured against core registry expenditure, members can see the insurance buffer for the ledger itself. The most informative approach would show both: one ratio for core registry continuity and one for total organisational commitments. That would reveal whether members are insuring the number-resource ledger or underwriting the broader institution.

Release rules matter as much as accumulation rules. If reserves fall below the band, the response may be temporary fee increases, spending restraint or phased rebuilding. If reserves exceed the band, the response may be lower fee growth, rebates, targeted relief, accelerated security investment, or explicit member-approved strategic projects. What matters is that the excess does not silently become ordinary revenue. In a membership body, a reserve target without a release rule is only half a policy.

When the cushion changes the institution

Reserves do not merely sit on a balance sheet. They alter institutional behaviour. A registry with no reserve is fragile and reactive. A registry with a substantial reserve can be independent, patient and resilient. It can also become insulated. It can survive dissatisfaction, delay fee reform, absorb weak programme performance, fund legal disputes, and preserve a staffing or project footprint without returning immediately to members for consent. The temptation of reserves is not usually spectacular misuse. It is ordinary budgetary sociology.

Departments defend their scope. Programmes acquire constituencies. Travel, meetings and representation become part of the expected institutional rhythm. New systems require maintenance. Temporary initiatives become permanent because ending them is harder than renewing them. Staff plans assume continuity. External partnerships create reputational commitments. A reserve reduces the urgency of choosing. It lets APNIC maintain multiple ambitions at once: technical operations, policy convening, training, research, measurement, community support, global representation, legal readiness, development work and institutional diplomacy.

Many of these ambitions may be useful. That is precisely why the incentive problem is difficult. Waste is easy to condemn. Useful activity is harder. But usefulness is not the same as necessity, and member fees are not a blank cheque for all useful regional Internet activity. The institutional question is whether APNIC can show which activities are essential to the registry, which are public-good contributions, which are externally funded, which are discretionary, and which should stop if money tightens.

The functional monopoly heightens the problem. If a vendor becomes expensive, customers compare alternatives. If APNIC's budget grows, members can object through governance channels, but they cannot buy the same registry authority elsewhere. The registry's indispensability is justified by the need for a single ledger. That same indispensability makes fiscal discipline depend on internal constitutional habits rather than competitive pressure.

Legal spending is one place where reserves can be both essential and dangerous. APNIC operates under Australian law, contracts with members, resource policies, election rules, by-laws, privacy obligations, service terms and relationships with other Internet institutions. It may face disputes over membership, transfers, resource recovery, eligibility, confidentiality, data handling, fraud, insolvency, sanctions, governmental requests or election conduct. A registry without legal capacity would invite opportunism. Wealthy members or market participants could pressure it if they believed it could not defend its records or decisions.

Yet a legal war chest can also make the institution less sensitive to the cost of conflict. It can encourage litigation rather than settlement, resistance rather than explanation, or rule designs that shift burdens to challengers. Legal advice can protect the registry for members, but it can also protect the institution from members. The solution is not to expose privileged material or weaken APNIC's defence. It is to report legal and governance spending by category: routine corporate advice, audit and assurance, resource and transfer disputes, election and governance advice, privacy and data matters, litigation or arbitration, security incident response, and consulting unrelated to legal risk. Members do not need the strategy memo. They need to know whether professional fees are rising because of ordinary assurance, exceptional litigation, governance stress or programme advice.

Security spending has a similar ambiguity. Members should not demand disclosures that create vulnerabilities. But security confidentiality cannot become a general exemption from fiscal accountability. APNIC can report enough to show whether money is going into operational hardening, third-party assurance, staff capability, redundant systems, incident response readiness, post-event remediation or long-term platform renewal. In a registry, security is core. In a budget, "security" should not be an unlimited container.

The temptation of reserves is strongest when no one can answer a simple question: what would APNIC do differently if reserves were lower, and what should it do differently because reserves are high? If the answer to both questions is "nothing", the reserve is no longer a disciplined insurance fund. It has become institutional ballast.

Cross-subsidy, solidarity and mixed missions

Every regional registry has a mixed mission. The narrow mission is to maintain a reliable number-resource registry. The broader mission includes the development of the regional Internet, training, security awareness, measurement, policy participation and representation in global forums. The narrow mission is easier to price because its failures are closer to the ledger. The broader mission is harder to contest because its benefits are real but diffuse.

Cross-subsidy is not inherently wrong. Larger operators may benefit when smaller networks keep accurate records, deploy routing security, understand transfer rules and participate in policy. A data centre in Singapore may benefit indirectly from better operational capacity in Nepal, Vanuatu, Cambodia or Papua New Guinea. A mobile operator in Japan or Australia may prefer a region in which RPKI, abuse-contact practices and number-resource hygiene are stronger across borders. Public goods are still goods.

But cross-subsidy must be named. Cost recovery asks what it takes to run the registry. Solidarity asks how much members should contribute to regional capability. Scarcity pricing asks whether holders of valuable IPv4 space should bear a larger burden because scarcity produces rents. Institutional expansion asks what APNIC wants to become. These are different questions. A single fee formula can blur them. Blurring makes consent easier to claim and harder to test.

Programme separation would improve the politics. Annual budgets should distinguish core registry operations, security and resilience, RPKI and routing-security services, reverse DNS and DNS operations, member support, request processing, transfer processing, policy development, training and technical assistance, measurement and research, meetings and community participation, global coordination, externally funded projects, legal and governance costs, and reserve contributions. Such separation would not diminish broader programmes. It would protect them from suspicion by allowing members to decide whether they are worth funding.

The APNIC Foundation complicates and clarifies the picture. Foundation-funded project expenses appear in APNIC's financial reporting, but members should be able to distinguish externally funded or restricted activity from member-fee-funded activity. Otherwise a line item can be misread in both directions: supporters may think member fees fund more development work than they do; sceptics may think essential registry fees are being diverted more than they are. The cure is not rhetorical reassurance. It is a plain map of funds, restrictions and outcomes.

Cross-subsidy should also be evaluated by burden, not just intention. A 50% discount for members in Least Developed Countries recognises a real distributional problem, but it does not cover every vulnerable operator. Hardship is not confined to UN categories. A rural network in a middle-income economy may be financially weaker than an urban operator in a poorer one. An island ISP may face high transit, satellite, power, equipment and shipping costs. A community network may hold resources for public-interest reasons and have limited administrative capacity. A university network may depend on slow public procurement and foreign-currency approvals. A small data centre may need provider-independent space and an ASN for resilience but operate on thin margins.

Equal rules can be unequal in effect. APNIC's first two ASNs are not subject to additional ASN charges, and the first /48 IPv6 provider-independent assignment is excluded from annual fee calculation. The fee schedule and by-laws also contain relief mechanisms, including Executive Council discretion to reduce or waive fees for deserving organisations. These are important signals. But the legitimacy of relief depends on whether hardship paths are predictable, public and measurable, rather than merely discretionary.

The right policy is neither romanticism about small operators nor indifference to them. Small networks should pay legitimate common costs. They should not be exposed to abrupt continuity risk because the payment system, currency channel, invoice timing or appeal process was designed for better-resourced organisations. Solidarity becomes credible when it is structured; otherwise it looks either arbitrary or sentimental.

The price of standing

"Good standing" sounds administrative. In a registry economy it is a control lever. APNIC's membership materials require members to remain current with fees and charges. Transfer conditions require payment before completion in several circumstances. Non-payment can lead to termination, a reactivation period and eventual risk that resources are subject to de-registration and return. These rules are understandable. A registry cannot maintain indefinite accounts for members who do not pay. It cannot allow fees to become voluntary while benefits remain unchanged.

The difficulty is that account standing touches systems whose consequences are not merely financial. RPKI is the clearest example. APNIC's hosted RPKI service allows members to create and manage route origin authorisations through its systems. In self-hosted arrangements, members operate more of the certification infrastructure but still depend on APNIC's role in the regional trust chain. Route Origin Validation depends on signed objects being current and available. A billing problem should not casually become a routing-security problem, even though the institutional relationship between holder and registry underlies the authority to maintain those objects.

Reverse DNS is another dependency. APNIC manages reverse delegations for IPv4 and IPv6, with delegations generated from registry data and published on APNIC name servers on a regular cycle. Reverse DNS does not usually determine whether packets forward, but it matters for diagnostics, mail reputation, logging, abuse handling and operational trust. For some networks it is part of customer assurance. For others it is part of institutional identity. Treating it as a mere account feature understates its reliance value.

Transfers add commercial timing risk. If a member is buying IPv4 for customer growth, closing a merger, restructuring a network, moving assets after an acquisition, or correcting historical records, a delay caused by fee standing can have consequences beyond the invoice. Transfer conditions may require APNIC accounts, supporting information, payment of relevant fees and deletion or movement of associated objects in outbound cases. A standing rule is therefore part of the continuity of number-resource capital.

The greatest risk comes from binary status models: paid or terminated, active or inactive, complete or blocked. Collection discipline requires consequences, but registry stability requires gradation. A better model distinguishes financial delinquency from operational validation and critical-continuity functions. Staged notices, technical-impact statements, payment plans, emergency continuation for critical dependencies, and appeal channels should be part of the system. The registry should preserve leverage to collect fees without creating avoidable harm to third parties or the routing system.

Metrics would matter. APNIC should be able to publish, in aggregate, how many accounts receive delinquency notices, how many enter payment plans, how many are terminated, how many are reactivated, how many resources face de-registration, how many RPKI or reverse-DNS changes are affected by standing issues, and how long interruptions last. Such metrics would not excuse non-payment. They would show whether collection policy is functioning as discipline or producing stability risk.

Hardship and appeal procedures belong in the fiscal constitution. The hardship path should be written for operators, not lawyers. It should explain when to contact APNIC, what evidence is needed, what temporary relief may be available, how critical services are treated, how disputes are escalated, and what happens if a member is affected by natural disaster, bank failure, currency controls, sanctions, public-sector payment delay or genuine hardship. The appeal path should cover fee disputes, termination, reactivation, transfer-related payment issues and exceptional standing consequences. Aggregate outcomes should be reported.

The deeper point is that financial controls in a registry are never purely financial. Because APNIC's services sit under routing security, reverse DNS and transfer records, billing rules become part of Internet stability policy. A private club can cancel access when dues are unpaid. A registry should think more like a clearing house or land-title office: payment matters, but notice, continuity, appeal and reliance by third parties matter too.

National registries and pass-through opacity

National Internet Registries add another layer to APNIC's fee economy. APNIC recognises seven NIRs in the region: APJII, CNNIC, IRINN, JPNIC, KISA, TWNIC and VNNIC. These institutions serve local communities, operate under local law, use local languages, maintain their own memberships and are not run by APNIC. In most cases organisations in those economies may choose between APNIC and the local NIR, but cannot obtain resources from both. NIRs have their own fee schedules in local currency and pay fees to APNIC. APNIC's schedule applies an additional 2.9 multiplier to NIR members' annual fee calculation.

The model has strong institutional arguments. Local-language service matters. Domestic payment channels matter. National communities may participate more effectively through a local institution than through a regional body operating primarily in English. NIRs can bridge regional policy and domestic operational realities. They can also reflect local law, industry structure and administrative practice.

But NIRs complicate price transparency. A downstream member in an NIR economy may not know how much of its local fee reflects APNIC charges, domestic registry operations, national taxes, reserves, development programmes, local cross-subsidy or currency hedging. APNIC members outside NIR economies may not know whether the NIR multiplier reflects cost, risk, history, administrative burden or a distributional bargain. APNIC may know what it charges the NIR but not how that charge is passed through. The downstream network may depend on APNIC policy and registry authority without having the same direct member relationship as a network elsewhere.

This is not a marginal issue. NIR economies include large and institutionally complex Internet markets. The fee burden may be mediated by industry associations, state-linked bodies, domestic membership classes or local regulatory realities. If APNIC's legitimacy rests on member control, the legitimacy of NIR pass-through rests on transparency between APNIC, the NIR and downstream resource holders.

A good regime would not require APNIC to control domestic fee schedules. It would require visible mapping. What does the NIR pay APNIC? What formula applies? What local categories exist? How are APNIC-related charges represented to downstream members? Are RPKI, reverse DNS, transfer and account-standing dependencies clear to those downstream networks? Are hardship paths available locally? Are APNIC fee changes passed through, absorbed, smoothed from reserves or offset by domestic policy choices? Are local members told when a fee reflects regional registry cost and when it reflects domestic institutional cost?

The permanent moratorium on new NIRs makes the point sharper. If the existing model is effectively closed to new entrants, the legacy arrangements carry more weight. They should not become a black box simply because they are familiar. APNIC should be able to show how its fees affect NIRs and how NIR arrangements protect downstream continuity. NIRs should be able to show members how regional charges, local costs, reserves and programmes interact. The regional system should not allow a network to face fee-driven continuity risk without knowing whether the source is APNIC policy, NIR policy, local payment practice or administrative failure.

Scarcity, transfers and the risk of a hidden toll

IPv4 exhaustion transformed the economics of regional registries. APNIC still has a small-policy distribution function, but the era of routine growth allocations is over. New and existing members can obtain limited IPv4 under policy, with the maximum available from APNIC constrained to a small block. Networks requiring more space must look to transfers, acquisitions, leasing-like arrangements, address sharing, carrier-grade NAT, IPv6 deployment or operational compromise.

Scarcity makes registry records part of a capital market. IPv4 blocks are bought, sold, valued, financed, leased and moved in business transactions. The registry does not set the market price, but it controls the recognised registration changes that make transfers credible. It also controls the records needed for diligence, the conditions under which resources remain in good standing, and the mechanisms by which RPKI and reverse DNS follow the holder.

This gives APNIC fee power a post-exhaustion character. A transfer fee may have been designed as an administrative charge. In a scarcity market it can be perceived as a toll on the movement of scarce operational capital. For large transactions the fee may be modest relative to private value. For small networks it may matter greatly. For outbound inter-regional transfers, the source APNIC member may face the charge. For an initial transfer to a new or addressless account, annual membership fees must be paid before completion. The fee system can influence who can move address space, when they move it and how much friction they face.

APNIC must avoid both undercharging and rent capture. Undercharging transfers would make ordinary members subsidise private transactions that require review, documentation, legal awareness and registry updates. Overcharging would turn the registry into a toll collector. The strongest justification is cost, risk and common-infrastructure recovery. The weakest is an unspoken claim on scarcity value.

Leasing sharpens the problem because operational use and registry title can separate. A lessee may rely on route objects, ROAs, abuse contacts and reverse DNS while the lessor remains the recorded holder or retains decisive control over registry actions. APNIC's formal fee relationship may be with the holder, not the operational user. If fees are unpaid or account standing is impaired, the user may suffer even after paying the lessor. If the registry does not recognise a lease as a transfer of holding, its fee and standing rules still affect the underlying capital. Private contracts then sit on public registry dependence.

The answer is not to deny the address market. It is to discipline APNIC's role near it. Transfer charges should be predictable, proportionate and explained. Exceptions and waivers should be published. Processing metrics should show whether smaller transfers face delays or friction disproportionate to their risk. Fee revenue from transfers should be compared with the cost of transfer processing and related risk management. If transfer fees contribute to general reserves or programmes, members should know that. If they are purely cost recovery, APNIC should be able to show it.

There is also a capital-control risk. A registry becomes a private capital-control lever when payment status, transfer fees or discretionary processing materially affect the movement, usability or value of IPv4 assets beyond what is necessary for accurate records and policy compliance. This is the sharpest scarcity-era danger. APNIC must enforce policy, verify transfers and maintain clean records. It should not influence address liquidity through opaque fee discretion, avoidable delay, unclear standing rules or poorly disclosed exceptions.

Several tests help distinguish stewardship from hidden taxation. Is the fee proportionate to work, risk or agreed cross-subsidy? Is the formula stable? Are exceptions governed by criteria? Are transfer fees funding transfer operations or general reserves? Does non-payment trigger staged consequences and appeal? Can hardship preserve critical functions while payment is resolved? Are NIR downstream users exposed to APNIC-originating charges without clarity? Do large and small transfers receive treatment that reflects both administrative cost and market fairness?

In the scarcity era, APNIC does not need the lowest fees to be legitimate. It needs members to believe that fees are constrained by purpose. The registry must prove that it is not monetising the fact that the ledger is unavoidable.

Budget control is not the same as member control

APNIC is member-governed, but member control is not automatic simply because members have formal rights. The cost of oversight is uneven. Large organisations may have staff who can follow meeting materials, fee proposals, financial statements and governance discussions. Small networks may not. Some operators face travel constraints, language barriers, time-zone burdens or limited familiarity with APNIC's processes. Others participate through an NIR and are one step removed from regional budget debate.

Voting weights add another complication. The tiered voting structure gives larger resource holders more votes. That may be fair because larger holders pay more and face greater exposure to registry risk. It also means marginal fee sensitivity is not evenly represented. A large operator may care about the absolute bill but treat it as small compared with network scale. A small network may face a fee that materially affects its budget yet hold little voting influence. Formal democracy can coexist with practical under-representation of those for whom billing friction is most serious.

Member meetings and annual reports are necessary but insufficient. Accounts tell members what was spent in standard categories. They do not necessarily show the fiscal consequences of policy choices, the burden by member class, the relationship between reserves and fees, the separation of core and discretionary spending, the effect on NIR-mediated networks, or the operational consequences of standing rules. Without this information, members may approve or tolerate budgets without understanding the trade-offs embedded in them.

The policy process also has fiscal blind spots. Technical participants may debate resource rules without seeing cost implications. Finance discussions may occur in member meetings without enough technical detail about RPKI, reverse DNS, transfers or support burden. A transfer rule that increases due diligence may raise staff costs. A routing-security change may require infrastructure investment. A final-allocation rule affects the population of small members. A historical-resource policy may change who enters the fee base. A national registry arrangement can shift costs between APNIC and domestic institutions. Conversely, a fee schedule can affect resource fragmentation, transfer timing, account consolidation, IPv6 adoption and direct membership.

Fiscal analysis should therefore accompany material changes. Major policy proposals should include an approximate operational-cost note. Fee changes should include a policy-impact note. Transfer-fee changes should describe likely effects on small transfers, large transfers, NIRs, inter-regional movement and historical resources. RPKI-related investments should state whether they are core security expenditure or optional service expansion. Reserve-policy changes should identify the risks covered and the member classes carrying the cost.

This need not make APNIC technocratic or slow. A short set of questions would improve discipline. What changes? Who pays? Who benefits? Which systems are affected? What is the burden on small members? How does the change affect NIR-mediated economies? Does it interact with IPv4 scarcity? Does it create or reduce continuity risk? Are hardship or appeal mechanisms needed? What metrics will be published after implementation?

Implementation metrics are the missing half of member control. A fee reform should not be judged only by the case made at adoption. Did reserves move toward the target band? Did service improve? Did transfer processing times fall? Did small-member hardship requests rise? Did RPKI availability improve? Did support tickets change? Did NIR pass-through become clearer? Did legal disputes increase? Did the burden match the forecast? Without after-the-fact reporting, policy is a promise and budget is a hope.

A fiscal constitution for a registry-layer utility

APNIC needs a fiscal constitution: not necessarily a single grand document, but a visible set of rules connecting fees, reserves, risk, services and member control. Members should not have to reconstruct the system from fee schedules, by-laws, financial statements, meeting slides, policy archives and scattered explanations. A registry-layer utility should make its fiscal logic legible.

The first element is a reserve target band. It should be expressed in months of expenditure, with separate measures for core registry operations and total organisational commitments. It should identify restricted, designated and unrestricted reserves. It should distinguish cash, short-term deposits, investment assets, property and assets not readily available for operations. It should state which risks justify the band: revenue interruption, cyber incident, disaster recovery, legal defence, governance crisis, critical supplier failure and planned capital replacement. It should say what happens above and below the band.

The second element is programme separation. Annual budgets and reports should show core registry operations apart from community programmes, research, training, grants, global coordination and externally funded activity. This would not devalue broader work. It would allow members to support it knowingly. Programme separation also prevents core-registry risk from being used as a rhetorical shield for all spending.

The third element is burden analysis for fee changes. Every material change to base fees, bit factors, transfer charges, ASN charges, reactivation fees, discounts or NIR multipliers should show class-by-class effects: Associate, Very Small, Small, Medium, Large, Very Large, Extra Large, NIR members, LDC-discounted members, IPv4-heavy members, IPv6-heavy members, ASN-heavy members, transfer recipients, outbound transfer sources and historical-resource holders where relevant. The analysis should show nominal and percentage changes, expected revenue, reserve effect and rationale. It should include currency and payment-friction considerations for vulnerable economies.

The fourth element is hardship and appeal. Relief should not depend on members guessing who to ask or how sympathetic the answer will be. The hardship path should explain evidence, timing, temporary relief, critical-service continuity, payment plans and escalation. The appeal path should cover fee disputes, termination, reactivation, transfer-related charges and exceptional standing consequences. Aggregate statistics should be published so members can see whether the mechanism is real.

The fifth element is NIR pass-through transparency. APNIC should publish its NIR fee treatment clearly, including the multiplier and exceptions. NIRs should show how APNIC-related costs appear in local fees. Downstream networks should be able to understand whether they are paying for regional registry cost, local registry cost, domestic programmes, taxes, reserves or currency protection. Where APNIC policy creates operational dependency for downstream NIR members, the continuity implications should be plain.

The sixth element is legal and governance-spend reporting. Professional fees should be broken down sufficiently to show legal, audit, assurance, consulting, election support, dispute resolution and security-advisory work. Active matters can remain confidential. The category and magnitude should not disappear. Large deviations should be explained, and insurance recoveries or reserve drawdowns should be visible in summary.

The seventh element is implementation metrics. Fees and reserves should be judged by outcomes, not intentions. The metrics should include reserve-to-expense ratios, core-registry expenditure share, programme expenditure share, legal spend by category, transfer processing times, transfer-fee revenue and cost, hardship requests and outcomes, account terminations and reactivations, de-registration events, RPKI availability, reverse-DNS update performance, support response times, NIR pass-through reporting and delivery of budgeted capital projects. The same measures should be reported consistently across years.

Such a constitution would not remove politics. It would improve them. Members could argue about the right reserve band rather than whether reserves are a black box. They could argue about solidarity spending rather than whether it is hidden in registry cost. They could argue about transfer fees rather than whether APNIC is quietly taxing scarcity. They could argue about hardship rules rather than plead for discretion in private. A member community does not need unanimity to govern well. It needs the right disagreements.

The legitimacy bargain

The economics of APNIC fees and reserves should be watched through incentives rather than slogans. Low fees are not automatically virtuous; they can starve the ledger. Large reserves are not automatically wasteful; they can buy independence and crisis capacity. Programme spending is not automatically mission creep; it can strengthen the regional Internet. Transfer charges are not automatically rent capture; private transactions create public registry work. The hard task is to keep each instrument tied to its stated purpose.

Several indicators would show whether APNIC is maintaining that discipline. The reserve-to-expense ratio should be intelligible and tied to a target band. Unrestricted liquid reserves should not rise indefinitely without a release rule. Legal spending should be classified well enough to distinguish routine assurance from governance stress. Core registry operations should be visible apart from broader programmes. Cross-subsidy should be named, especially where large holders, small operators, LDC discounts or NIR arrangements are involved. Fee changes should show distributional effects before members are asked to accept them. Hardship data should show whether billing rules are creating avoidable account-risk. NIR pass-through should be clear enough for downstream networks to understand what they are paying for. Transfer and leasing dependence should be monitored so the registry does not become the hidden price-setter of address liquidity. RPKI and reverse-DNS continuity should be treated as part of financial-control design, not merely service uptime.

These are not accusations. They are instruments for separating prudence from drift. A legitimate registry should welcome them because they protect the strongest case for funding: the case that APNIC is an insurance-backed ledger for a region that cannot afford registry failure.

The old language of membership dues is too small for the scarcity era. The invoice is attached to a ledger that helps make number resources usable. The reserve is an insurance fund for a regional utility. The transfer fee is a charge on the movement of scarce operational capital. The standing rule is a continuity control. The budget is a map of institutional ambition. The Executive Council is not only a governing body; it is steward of the line between cost recovery and private gatekeeping.

APNIC should not be starved. A brittle registry would be more dangerous than a well-funded one. The region needs secure systems, competent staff, reliable RPKI, reverse DNS continuity, credible transfer processing, legal resilience, disaster recovery and policy support. But APNIC should not be allowed to rely on indispensability as a substitute for explanation. The more valuable IPv4 becomes, the more careful the registry must be with fees attached to holdings and transfers. The more operational systems depend on account standing, the more careful it must be with termination, reactivation, hardship and appeal. The larger reserves become, the clearer their target and release rules must be. The broader APNIC's programmes become, the more clearly core registry cost must be separated from regional public goods.

The legitimacy bargain is simple to state and difficult to maintain. Members fund the registry because the registry preserves the common ledger on which private networks depend. In return, the registry must behave as if every dollar is constrained by that purpose. It may insure against risk, but it should not accumulate capital without a rule. It may cross-subsidise regional capacity, but it should name the subsidy. It may charge for transfers, but it should not tax scarcity by stealth. It may enforce payment, but it should not turn billing into avoidable routing or DNS risk. It may defend itself legally, but it should not hide governance problems inside professional fees.

APNIC's strongest fiscal position would not be the largest reserve or the lowest fee. It would be a budget that members can read as a constitutional document: what must be funded, what risks are insured, who carries the burden, who receives relief, what is discretionary, what is temporary, what is measured, and when accumulated money returns through lower charges or better-defined resilience. That is how a registry remains a ledger rather than a gatekeeper. In the scarcity era, legitimacy is earned by the discipline with which the institution prices its own indispensability.