Summary

  • An IPv4 price can reflect more than the number of addresses. If a transfer path exposes the seller to longer completion, uncertain application approval, repeated document requests, bilateral coordination, or a significant chance that the transaction cannot be registered, buyers may demand compensation and sellers may accept a lower net price. The discount is the price of institutional risk, not evidence that the region's addresses are technically inferior.
  • The appropriate unit of analysis is a directed registry path: from source RIR to destination RIR, separating intra-regional and inter-regional cases. A label such as 'APNIC price' or 'African price' groups different recipient rules, resource histories, contract statuses, and compatible paths into a misleading average.
  • No public dataset currently combines the entire population of negotiated and abandoned transactions with the contract price, net proceeds, block quality, submission dates, registry-controlled delay, party-controlled delay, needs justification outcomes, and failed paths. Public transfer logs identify recognised changes but generally omit price, negotiations, refusals, and withdrawals. Private broker reports only observe their own books.
  • A credible estimate would match cases by contract month, block size, prefix fragmentation, reputation status, historical or contractual status, buyer type, seller urgency, payment terms, and path eligibility. It would then compare directed registry paths and place administrative time into pre-declared delay bands, while preserving failed or withdrawn cases rather than studying only completed transfers.
  • Delay is endogenous. Weak ownership evidence, contested space, poor reputation, complicated corporate history, or an unprepared buyer can both slow a case and reduce its price. A study must separate these defects from registry-controlled time, use pre-filing attributes for matching, and test policy changes or path openings where possible. Otherwise, it will assign a price to difficult transactions and call the result a regional penalty.
  • The Number Resource Society can support an access-control-free evidence layer: common case definitions, confidential contribution, portable timestamps, independent aggregation, and public uncertainty ranges. Its role should be to reveal whether institutional frictions are capitalised into price, not to certify a correct price or add another transaction authorisation.

A regional discount would be a waiting cost, not an address defect

Two clean IPv4 blocks of the same size can perform the same numbering function. Their packets do not carry the calling code of the registry that updates the record. A router does not reduce its preference because a transfer office took three weeks instead of one. If a block sells for less while keeping comparable technical and commercial attributes constant, the difference requires an explanation outside the protocol.

The administrative path is one possible explanation. A seller who only receives funds after registration is completed has capital tied to an uncertain date. A buyer planning a migration cannot be certain when they can present the new registration to upstream providers, customers, security teams, and geolocation providers. An escrow can protect the funds, but it does not bring the registry decision. A broker can coordinate documents, but cannot make an incompatible path executable. Lawyers can allocate the risk of failure, but the contractual allocation does not remove the underlying risk.

The effect on price can appear in several places. A buyer may offer less upfront because they expect a delay. A seller may grant a later discount after repeated requests make closing uncertain. The parties may keep the nominal price but shift fees, escrow cost, or remediation work to the seller. A broker may quote a wider margin for a difficult path. A financing provider may advance less or charge more because completion is less predictable. A pressed seller may abandon the path and accept a lower offer from a smaller set of eligible buyers.

Calling all these effects a 'regional discount' is convenient but dangerous. The resource is not worth less because of the people or networks in a region. The discount, if it exists, attaches to a specific path under specific rules at a given time. It is a cost imposed by institutional conditions. This distinction matters because a path can be changed by policy, mutual recognition, service performance, and portability. Geography is not fate; administrative design is a variable.

The price being discounted must be defined before it is measured

Market commentary often says 'price per IP' as if every observation were comparable. It is not. One figure may be the buyer's gross contract price. Another may be the seller's net proceeds after broker commission. A third may include escrow, legal review, registry fees, reputation cleaning, or managed migration. Some contracts fix the price for an entire block; others spread value across addresses, services, and guarantees. A temporary right to use addresses is not the same product as a recognised permanent transfer.

The primary endpoint of a study on regional friction should be stated in advance. Net proceeds per address for the seller is a good choice when the question is whether difficult administration discounts the holder's asset. All-in acquisition cost per address for the buyer is a different and equally useful endpoint when the question is whether slow administration taxes deployment. The two must not be conflated. A transaction can produce lower net proceeds for the seller and higher cost for the buyer at the same time if the friction absorbs the spread.

Contract date also matters. The relevant market reference is the price available when the parties fixed the commercial terms, not the date on which a registry later published the completion. If a case closes months after the agreement and the market has moved in between, comparing its price at the closing date with a fast case may attribute a general market shift to regional delay. The dataset therefore needs at least offer date, agreement date, registry submission date, approval date, and final record modification date.

Currency conversion requires the same discipline. Use the exchange rate specified in the contract when available, and otherwise a market rate stated at the agreement date. Do not convert every historical transaction at today's rate. Do not treat a financing charge as part of the IPv4 unit price in one region but as a separate service elsewhere.

The word 'discount' should then mean a conditional difference relative to a contemporary benchmark after observable product and contract attributes have been matched. It should not mean that a broker's quarterly average is lower than another broker's asked-price list. Asked prices, accepted offers, and realised net proceeds are different classes of evidence. A credible study can publish all three, but it must not silently substitute one for another.

The path, not the region, is the economic treatment

Every transfer has a direction. The starts under one RIR. The recipient may stay in that service region or be under another. Source rules may determine whether the block is eligible to leave. Recipient rules may determine whether the buyer is qualified and for what quantity. In an inter-RIR case, both institutions may need to approve and synchronise the change. The same source block can therefore face different conditions depending on the destination.

This makes the directed pair the minimum useful treatment: ARIN to ARIN, ARIN to APNIC, RIPE NCC to ARIN, LACNIC to RIPE NCC, and so on. Even this is not sufficient by itself. A path may differ by resource class, historical status, allocation history, recipient account, National Internet Registry involvement, application pre-approval, and policy version. The analysis needs cells by path and period with these attributes preserved.

The current RIPE NCC transfer guide gives a simple reason for this design. It states that inter-RIR transfers require approval from both registries and take longer than transfers within its own service region. Its current policy also applies a recipient utilisation plan to incoming transfers from regions that require a compatible needs-based policy, even though ordinary intra-RIPE IPv4 transfers do not use a general forecast for the recipient. The condition follows the path.

APNIC's current policy similarly assigns source conditions to the source RIR and recipient conditions to the destination RIR. An APNIC recipient must provide a detailed 24-month utilisation plan, while the source is evaluated under the rules in force where the source account is. ARIN requires recipients specified in its region to demonstrate provisioning up to 24 months. A transaction directed into ARIN therefore carries a different recipient filter than an otherwise similar transfer directed into a region without general needs review.

A single 'RIPE price' could mix fast intra-regional cases, longer bilateral cases, corporate reorganisations, historical resources, and ordinary policy transfers. A single 'LACNIC price' could combine outbound, inbound, and intra-regional transactions even though the counterparty and recipient obligations differ. Regional averages can be useful descriptive summaries, but they cannot identify the institutional cost.

Needs evidence enters the negotiation before an application is filed

The most visible delay begins when a registry opens a case. Some of the most important economic effects occur earlier. A buyer who expects detailed needs scrutiny may seek pre-approval before contacting sellers. They may reduce the requested quantity to fit a conventional utilisation plan. They may avoid a large block that would require a more difficult forecast. They may choose a source on a path that their advisers know. A broker may refuse to represent a buyer who cannot assemble acceptable evidence.

These choices affect price without appearing as registry processing time. Pre-approved buyers can negotiate from a stronger position because they offer a cleaner path to closing. Sellers may accept a lower nominal offer from them rather than from a higher bidder whose eligibility is unresolved. Conversely, an approved buyer facing few compatible sources may pay a premium. The sign of the observed effect is therefore not guaranteed. Slow administration can discount the seller's block, increase the buyer's all-in cost, or redistribute value toward parties holding scarce approvals.

ARIN and APNIC offer pre-approval mechanisms that place needs assessment upstream of a specific source. This can reduce uncertainty, and a rigorous study should identify whether the buyer held a valid approval at the time of the agreement. But pre-approval is not a random feature. Repeat buyers with good legal counsel are more likely to obtain it, and those buyers may also negotiate better prices. Comparing pre-approved and unapproved cases without matching would confound sophistication with administrative certainty.

The invisible sample begins with inquiries. How many buyers were told that a quantity, corporate arrangement, or destination was unlikely to qualify? How many reduced demand? How many chose leasing because permanent transfer seemed too uncertain? How many sellers accepted an intra-regional offer because an inter-regional offer carried approval risk? These are economic outcomes even though no transfer line was created.

A proper panel must therefore start before filing. Brokers, escrow providers, and direct parties should record a pseudonymous deal file when serious terms are exchanged, not only when a registry accepts an application. The panel then tracks that file through agreement, submission, information requests, approval, withdrawal, expiry, or failure. Otherwise, the cleanest visible transactions will define the market and the most consequential frictions will disappear.

Incompatible paths reduce the buyer set and may depress the best available offer

A transfer path may be closed, conditionally open, or operationally uncertain. Where no compatible inter-RIR path exists, a seller cannot simply choose any global buyer. The address remains globally routable in the technical sense, but the recognised change of holder is not globally portable. An administrative boundary becomes a market access boundary.

The immediate effect is a reduced number of eligible counterparties. In an auction, fewer serious bidders can reduce the highest bid even if each remaining bidder is well-funded. In a negotiated sale, the seller's fallback option weakens. A broker may need more time to find a recipient in the permitted geography or may suggest a structure that leaves the record with the source. The commercial value of a clean, permanent change is then replaced by the value of a narrower or more conditional arrangement.

The discount does not need to appear as an explicit line. A seller may wait longer, pay a higher commission for a difficult placement, accept deferred payments, provide longer warranties, or retain liabilities that a fully portable transfer would end. A buyer may insist on a termination right tied to registry approval. The expected cost of that right is capitalised in the negotiation.

Compatibility can also produce premiums. If a newly opened path connects a group of sellers to a high-demand destination, eligible blocks may become more valuable. If only a subset of blocks can cross due to status or holding restrictions, those blocks may acquire scarcity value. This is another reason not to presume the direction or size of a regional coefficient.

AFRINIC's position illustrates the importance of dates and executable service. AFRINIC ratified a broader digital resource transfer policy in February 2026, including an inter-RIR framework subject to reciprocity. Ratification, counterparty recognition, and operational availability are distinct states. A study must code when a directed path could actually be completed, not merely when a proposal was discussed or approved. A paper opening without tested service should not be treated as equivalent to a mature bilateral path.

Uncertainty can be costly even when the final answer is yes

Approval rates alone cannot measure institutional cost. A service may approve almost all filed requests and still impose substantial uncertainty if decisions come late, evidence requests vary, applications are repeatedly restarted, or advisers cannot predict which documents will satisfy the next reviewer. Transaction cost is created by variance as much as by refusal.

Consider two paths with the same mean completion time. On the first, almost every complete case ends in a narrow window. On the second, half close quickly while a significant tail remains unresolved for months. A seller with a financing deadline and a buyer with a migration date will value the first path more. The second path forces them to price the possibility of becoming a tail case.

This risk can be represented without claiming to know a universal discount. The parties estimate the probability of completion, the distribution of completion dates, the cost of carrying funds, the cost of replacement addresses, the chance of missing a customer launch, and the legal cost of termination. The negotiated adjustment reflects their own exposure and bargaining power. A large cloud operator may tolerate delay but pay heavily for uncertainty around a fixed capacity event. A small ISP may have a lower daily cost but no margin in its balance sheet for extended escrow.

Institutional uncertainty also affects diligence. If examiners can reopen historical authority or corporate succession, sellers spend more before the agreement to assemble evidence. If a recipient needs detailed needs evidence, buyers expose customer forecasts and network plans. If two registries coordinate by email without a published step clock, neither party can say which institution controls the next action. Each uncertainty produces precautionary work.

The empirical question is therefore not simply whether one RIR is 'slower'. It is whether a directed path has a different conditional distribution of registry-controlled time, information requests, quantity changes, withdrawals, and failures, and whether those differences enter the commercial terms. A region may have a reasonable median and an expensive tail. A path may be fast after a complete submission but expensive before the completeness threshold is reached.

The public registry does not contain a global transaction panel

The five RIRs publish transfer information in forms that support valuable research. Common transfer files can identify resources, parties, dates, and source or destination registries for recognised events. Some RIR pages distinguish policy transfers from mergers or other changes. These records help reconstruct custody movement and compare volumes.

They do not disclose a full market. The published date is typically the recognised transfer date, not necessarily the contract date. The line normally does not report price, broker fees, escrow cost, failed offer, initially requested quantity, approved quantity, registry questions, party response time, or the transaction that was never filed. A commercial sale split across multiple prefixes may appear as multiple lines, while a corporate reorganisation may appear near market transfers. Public logs alone cannot reveal the effect of delay on price.

Private reports fill part of the gap. Brokers may publish price observations by block size and period. Advisers may describe typical completion ranges. Their data is useful, especially when method and coverage are clear. But each firm sees a selected order book. A broker specialising in fast, clean cases will report a different distribution from a broker handling contested historical space or difficult inter-RIR paths. Direct transactions and competitor deals remain invisible.

Academic research on reported and inferred transfers shows both the value and the limits of registry data. Published transfers can be compared with routing and registration changes, but inference from network observations is noisy and a route change is not evidence of a sale. This research discipline should be transposed into price analysis: evidence that a block moved or began to be routed does not reveal the commercial contract.

There is therefore no defensible public figure for 'the discount' across all regions. The absence is not a reason to assert that friction has no price. It is a reason to specify the missing observations and build a study whose limits are visible.

The case, not the prefix line, should be the unit of observation

The dataset should assign a pseudonym identifier to a commercial case. A case begins when a defined buyer and seller negotiate a specified set of resources under a common price agreement. It may include multiple prefixes and multiple registry tickets. If a /16 is split into four /18s for the same buyer under a single contract, the economic unit remains one case. If four buyers acquire those blocks under separate conditions, there are four cases.

This definition prevents block fragmentation from inflating the sample. Public transfer files are resource-centric, so a complex transaction can produce many observations. Treating each line as an independent price observation would over-weight large split transactions and under-weight the shared delay.

Each case should record commercial milestones: serious quote, accepted offer, signed agreement, escrow funding, first registry submission, completeness confirmation, each information request, each party response, source approval, destination approval, registry record modification, fund release, and technical switch. Timestamps allow total elapsed time to be decomposed.

The case also needs outcome states broader than finalised and rejected. Useful categories include: no agreement, agreement expired before filing, withdrawn by buyer, withdrawn by seller, quantity reduced, source replaced, destination changed, converted to lease, abandoned sole-record structure, registry refusal, unresolved dispute, and finalised. A reason taxonomy should distinguish a stated decision from a party's interpretation.

Commercial confidentiality can be preserved. Names and exact prefixes do not need to enter the research table. A trusted auditor can verify that the contributor actually participated and that duplicate reports from buyer, seller, broker, and escrow refer to a single case. Block size can be banded for publication. Price can be normalised and published only in cells that meet a confidentiality threshold. The goal is not to expose contracts; it is to make the institutional incidence measurable.

Matching must make compared blocks commercially plausible substitutes

A matched design asks a practical counterfactual: what would this transaction probably have cost on another executable registry path if the block, timing, and parties were otherwise comparable? No observation can answer that perfectly because the same block cannot be sold simultaneously under two custodians. Matching reduces the difference to cases that a buyer could reasonably have considered as substitutes.

The first matching dimensions are agreement period and block size. IPv4 prices move over time and often vary by prefix size. A /24 acquired in one quarter should not be lightly compared with a /16 acquired years earlier. Matching can use exact calendar windows and prefix bands, with sensitivity tests for narrower windows.

The second group describes the resource state before negotiation: historical or contractual status, eligibility date, fragmentation, registration history, dispute status, routing history, documented control under letter of authority, RPKI state, and overall reputation state. Reputation should be measured from a stated set of feeds before the agreement, not from subsequent abuse that may be caused by the recipient.

The third group describes transaction conditions: permanent or temporary, direct or intermediated, cash or financed, escrow structure, warranties, indemnities, fee allocation, bundled remediation, and whether the buyer already held approval. Seller urgency should be approximated by stated deadline bands, insolvency status, public auction, planned shutdown, or normal sale. Urgency is hard to observe but too important to ignore.

The fourth group describes the parties without exposing identity: new or repeat entity, network type, approximate existing holdings, buyer jurisdiction, seller jurisdiction, related-party status, and whether an NIR participates. A repeat institutional buyer and a small new firm do not face the same information cost.

Exact matching on every field could leave too few cases. The study can use coarse exact matching on the most decisive attributes and a propensity or distance method inside those cells. Each matched comparison should publish balance diagnostics. If slow-path and fast-path cases remain very different after matching, the correct conclusion is weak identification, not a confident regional coefficient.

Delay bands should reflect transaction exposure rather than flattering averages

Administrative time should be partitioned by control. Party response days start when an institution requests information and stop when a complete response arrives. Registry-controlled days start when a complete submission or response is acknowledged and stop at the next institutional action. Bilateral transfer days cover the interval between source approval, destination action, and synchronised completion. Escrow time and technical switch lie outside the registry clock unless an institutional step holds them.

The study should pre-declare delay bands. A feasible design is completion within seven calendar days, eight to fourteen, fifteen to thirty, thirty-one to sixty, sixty-one to ninety, and over ninety, with unresolved cases kept as censored observations. The bands are research categories, not claims about current duration in any region. Their purpose is to show whether price changes monotonically with exposure and whether the long tail carries a distinct adjustment.

Calendar days are preferable for economic exposure because funding and opportunity cost do not stop on weekends. Service performance reports may also show business days to compare institutional commitments. Public holidays and declared closure periods should be coded, not selectively used.

The start event must be consistent. Measuring from initial filing favours institutions that discourage early filing; measuring from staff-declared completeness may exclude the very document cycle under study. The panel should publish both total institutional engagement and post-completeness decision time. It should also show the number of completeness cycles.

Delay bands need path and policy version. A path opened in one year or a changed application rule in the middle of the sample cannot be pooled as if constant. Intra-regional and inter-regional time should also not be merged. The result is a matrix: directed path, policy period, resource class, delay band, and outcome.

Price outcomes should be reported for expected delay known at agreement time and for realised delay after filing. The first captures what the parties priced ex ante. The second captures renegotiation and unforeseen cost. Mixing them would leave a later problem explaining an earlier price that could not have anticipated it.

Registry-controlled delay is not the same as a difficult seller

The central identification problem is endogeneity. Cases do not become slow at random. A seller with incomplete corporate records may take longer to prove authority. A block with contested chain, stale registration, or retransfer restriction may require more work. A buyer seeking an unusually large quantity may face deeper needs scrutiny. These same attributes can reduce price even if the registry handles the case perfectly.

The dataset must therefore preserve the state known before filing. Was the registered holder identical to the contracting seller? Were signatories already documented? Was the space outside a holding period? Was the recipient pre-approved? Was the source free of identified dispute? Did the parties submit the required forms together? These variables separate preparation from institutional response.

Information requests should be coded by subject. Identity verification, authority, resource eligibility, requested quantity, contract status, sanctions, dispute, payment, membership, and technical update are different. Repeated requests for missing mandatory documents should not be counted the same way as a new interpretation introduced after a complete submission.

Examiner discretion remains hard to observe. A case may be delayed because the policy is ambiguous, because staff capacity is low, or because the institution sees a genuine risk that contributors cannot disclose. The study should not presuppose bad faith. It should compare similar cases, publish reason codes, and allow institutions to correct factual timestamp errors without vetoing results.

Seller selection creates another bias. Holders of portable, well-documented space may choose the fastest destination, while more difficult blocks stay in the source region. Buyer selection works the opposite way: sophisticated buyers may accept difficult paths because they can manage them. A path coefficient may therefore reflect who chooses the path.

Policy discontinuities can help. When a directed path opens, a holding restriction changes, or a needs requirement is removed, the study can compare eligible and unaffected cases around the change. Such events are not perfect experiments; market conditions and entity composition may change at the same time. But they provide a more robust test than a simple cross-sectional average.

Failed and abandoned transactions are part of the price evidence

Completed transactions condition the sample on success. This is a serious error when the proposed mechanism includes incompatibility and uncertain approval. A path that finalises only the easiest cases may look fast and well-priced because all the difficult cases disappear before publication.

Suppose sellers on two paths receive the same price when deals close. On the first path, almost all signed cases close. On the second, a substantial share expires, converts to lease, or accepts a different buyer after months of delay. The comparison of finalised prices would report no discount. The expected value for a seller is nevertheless lower on the second path because the probability and time of completion differ.

The analysis should estimate at least three outcomes: the conditional price among finalised cases, the probability of completion in each time band, and the expected discounted net proceeds from the initial agreement. The third combines price, time, fees, and failure without pretending that an abandoned transaction has an observed sale price. It can be presented as a range based on stated fallback assumptions.

Withdrawals need reasons. A buyer who disappears after a market decline is different from a buyer unable to satisfy a needs test. A seller who cannot prove authority is different from a seller who refuses to accept a document request. Contributors may disagree. The dataset can keep both versions and mark the reason as contested.

Attrition before filing is harder. Brokers may report serious cases they took on, but the threshold for 'serious' must be common. A signed term sheet or verified proof of funds can define inclusion. Informal inquiries should not become failed transactions simply because a visitor asked for a quote.

Survival methods can keep unresolved cases in the analysis without imputing a fictional completion date. Competing-risk models can distinguish approval, withdrawal, refusal, and path change. The statistical method matters less than institutional honesty: a market study cannot erase the transactions most affected by the barrier it purports to measure.

Block quality must be separated from administrative provenance

IPv4 blocks carry histories that affect value. Some have clean, stable registration and routing histories. Others have been heavily used for mail, hosting, or services that attract abuse complaints. Geolocation databases may place them incorrectly. Route objects and RPKI state may require changes. A buyer may need to warm reputation, contact external databases, or renumber around fragmentation.

These conditions can be correlated with region. Historical allocation patterns, industries, enforcement practices, and broker specialisation differ. If a study observes lower prices for one path but does not measure block state, it may call a reputation discount an administrative discount.

Quality measurement must be pre-agreement and reproducible. Use broad bands derived from multiple specified sources, because a single reputation list may be incomplete or commercially biased. Record whether a prefix was routed, dormant, recently originated from multiple networks, or covered by a route origin authorisation. Preserve uncertainty when evidence is contradictory.

Registration status also matters. Historical space may have different contractual consequences across paths. A transfer may result in a change of service agreement or status. A buyer may value access to RPKI and support, while a seller may value preservation of a historical position. These conditions belong in matching.

Block size is not enough. A contiguous /17 and an equivalent address count assembled from many /24s impose different routing, operational, and reputation costs. A partial transfer carved from a larger allocation may require more work than an intact block. The dataset should record number of prefixes, contiguity, and whether the contract allows subdivision.

Research must avoid a more subtle stigma. If it publishes a regional discount without showing that quality was balanced, market actors may apply the label to every holder in that region. This can create the very discount the study claims to discover. Path-specific results, confidence intervals, and balance evidence are safeguards against turning incomplete analysis into a pricing convention.

A matched estimate should be a range with multiple robustness tests

The base model can regress the logarithm of net price per address on directed path and expected delay band, with fixed effects for agreement period and controls for matched attributes. The logarithm makes proportional differences easier to compare across price levels. But a regression coefficient should not be presented alone.

First, publish raw matched pairs or cell summaries under confidentiality thresholds. Readers should see how many cases support each path, the distribution of block sizes, the balance of historical status, and the share with pre-approval. A result based on a handful of intermediated cases is not global simply because the formula is sophisticated.

Second, vary the matching. Use exact month rather than quarter, narrower block bands, finalised cases only, all serious cases, direct deals only, and intermediated deals only. Exclude distressed sales, then show them separately. If the estimated sign changes under reasonable choices, the evidence is unstable.

Third, compare expected and realised delay. A price discount associated with path at agreement time is evidence that entities anticipated friction. A later price concession associated with unforeseen registry-controlled delay is a different mechanism. Both can exist, but causality should follow chronology.

Fourth, test placebo outcomes. Technical performance after transfer should not deteriorate simply because a service was slow, once block quality and operator deployment are controlled. If every characteristic appears 'regionally discounted', omitted variables are likely doing the work.

Fifth, examine policy changes and path openings. Did the bid spread narrow after removal of a needs condition, making a bilateral path executable, or increasing predictability of step times? Compare with paths unaffected by the change. Report anticipation and transition periods rather than choosing the most favourable threshold.

Finally, use partial identification when data is missing. If broker coverage is known only within a range or abandoned transactions lack final fallback prices, publish upper and lower bounds. A wide but honest interval is more useful than a precise made-up discount.

A discount can be decomposed into costs that institutions can actually reduce

The economic effect is easier to manage when split into mechanisms. One component is the time value of money: escrowed or delayed proceeds have a financing cost. Another is opportunity cost: the seller cannot firmly offer the same block elsewhere while a conditional deal is alive. A third is replacement cost: the buyer may need temporary addresses or book another source.

There is also failure risk. If registry approval is a condition of settlement, the expected price is multiplied by the probability of completion and reduced by the cost of termination. A party can insure part of this risk through escrow conditions, but no contract can create an unavailable transfer path.

Information cost is distinct. Needs evidence, authority proof, and historical documentation require staff and advisers. Part of the work is necessary to prevent errors. Repeated or unpredictable work is a policy cost. Confidentiality also has value: a buyer may disclose customer plans or capacity forecasts they would not otherwise reveal.

Market access cost arises when incompatibility reduces the number of counterparties. It can be approximated by the difference between the best eligible global offer and the best executable offer, both independently verified. This comparison should include the chance that the highest offer actually qualifies.

Uncertainty carries an option premium. Parties value the ability to plan migration, financing, and customer commitments. A predictable fifteen-day process may be cheaper than a process averaging ten days but ranging from two to ninety. Publishing only the mean service hides the option cost.

These components point to reforms. Faster clocks reduce financing cost. Pre-stated evidence reduces information cost. Compatible paths widen the bidder set. Independent review reduces arbitrary failure risk. Portable records reduce reconstruction. The purpose of measurement is not to stigmatise a region. It is to identify which institutional choices reduce value for the holder and increase cost for the operator.

The distributional burden falls most heavily on parties with the least margin for uncertainty

Large repeat buyers can maintain pre-approvals, legal templates, broker relationships, and alternative inventory. They can file multiple cases and absorb delay. Large sellers can wait or choose among buyers. Smaller operators and one-time holders have less redundancy.

A small network buying for a customer launch may need a specific block size by a specific date. If the transfer slips, it may lose the customer or lease temporary space at additional cost. A novice buyer may not know what forecast format is convincing. A seller raising funds for a network investment may accept a lower clean offer rather than risk a higher conditional offer.

Regional capital conditions amplify the effect. Where financing is expensive or exchange rates volatile, an uncertain closing date can change affordability. This does not justify assigning a fixed penalty to the region. It reinforces the need to measure buyer and seller exposure rather than treating the nominal price as the full outcome.

The burden can also shift to customers. Operators who pay more for acquisition and delay may pass the cost into hosting, access, or service prices. Operators unable to acquire a contiguous block may assemble fragments, increasing routing and management work. A transfer rule that seems limited to registry members can thus affect service expansion.

There is a policy risk in the opposite direction. Allegations of a regional discount can be used to demand broad control over transfers in the name of protecting holders. If the proposed protection closes more paths or increases discretion, it may deepen the discount. Evidence should support narrower administration: predictable verification, executable portability, and transparent review.

Registries can publish decisive evidence without publishing private contracts

Every RIR already controls timestamps and decision events that no broker can fully reconstruct. It can publish quarterly distributions of elapsed time between acknowledged submission and first action, completeness, decision, and record update. It can separate institution-controlled and party-controlled intervals, show percentiles rather than only averages, and preserve the age of unresolved cases.

It can also publish motivated outcome aggregates: approved as filed, approved at reduced quantity, withdrawn, expired, refused, converted to another case type, and still pending. Needs-review cases can be grouped by requested and approved size bands. Inter-RIR cases can be shown by directed pair, with both institutions reconciling the count.

No price needs to appear in the registry's own publication. The public performance table can be joined, under independent confidentiality, with commercial observations contributed by parties. Stable case hashes can enable audit without exposing organisation names or exact prefixes. The institution should not receive a veto on the commercial output simply because it provides timestamps.

Service definitions need versioning. If a clock starts only after staff declares a case complete, also publish pre-completeness cycles. If a bilateral case waits for the counterparty, identify the handoff. If a legal proceeding suspends a case, use a declared category. Undefined pauses are where accountability disappears.

Appeals and corrections should be visible in aggregate. A decision reversed on review is evidence of uncertainty. A data correction should leave a revision trail. Confidentiality can protect documents while revealing how often a rule changes quantity or timing.

The first result may show that no path-specific price effect can yet be estimated because commercial coverage is too sparse. That is a useful finding nonetheless. It tells members and market actors what evidence is missing and prevents unsupported regional allegations from hardening into accepted fact.

Market actors can build the missing panel without giving up confidentiality

Brokers, escrow providers, lenders, buyers, and sellers each hold a fragment. A cooperative panel can combine these fragments if contribution rules are strict. Participation should be open to direct parties, not limited to established brokers. Each contributor should attest to their role and allow duplicate reconciliation by an independent auditor.

The panel should publish its coverage. It can compare contributed finalised cases with public transfer logs by path, period, and size band, while acknowledging that one commercial case may match multiple lines. It should state the share of the contributor's activity represented and whether direct deals are missing. A price chart without a coverage denominator invites overinterpretation.

Exact prices can stay confidential. Public outputs can show medians and interquartile ranges only when enough independent parties populate a cell, accompanied by matched estimates and uncertainty intervals. Cells dominated by one contributor should be suppressed or flagged. Suppression is better than false anonymisation.

Contributors need portable copies of their own event history. If a broker leaves the panel or the collector fails, parties should retain verified timestamps and classifications. Research definitions should be open so that another institution can reproduce the aggregation.

Funding must be disclosed. A seller platform gains from results showing path discounts; a registry may prefer results that minimise them; a broker may want its own execution history to appear superior. Conflicts do not invalidate data, but hidden influence would invalidate trust.

The panel should invite institutions to respond before publication to factual errors, then disseminate the response alongside the result. It should not negotiate the conclusion. A registry can show that a delay was party-controlled. A contributor can contest that classification. Both versions can stand if evidence is not resolved.

The Number Resource Society can make evidence portable without becoming a price authority

The Number Resource Society is well-placed to define a lightweight evidence service because the problem is fragmented records, not a lack of organisations claiming power. Its useful role would start with common definitions: commercial case, directed path, complete submission, registry-controlled day, party-controlled day, needs reduction, incompatible path, withdrawal, and finalisation.

The Number Resource Society could publish an open event format and accept signed contributions under confidentiality. It could reconcile reports, preserve duplication links, publish coverage, and commission independent statistical review. A buyer or seller could export their own case history and use it in an appeal, audit, or future transaction without depending on a broker's private ticket system.

The public output should show what is known and what is not. If a path has too few matched observations, the Number Resource Society should not publish a ranking. If price coverage is selective, it should state the selection. If an apparent discount disappears after matching for block reputation, that correction should be highlighted.

The Number Resource Society must not certify that a regional price is fair, require parties to disclose their contracts as a condition of transfer, sell accelerated recognition, or make its own approval necessary. It should not turn a research panel into another checkpoint. Its authority comes from reproducible definitions, consented evidence, and replaceability.

The positive institutional direction is portability. A holder should be able to carry a verified identity, resource history, dispute status, eligibility dates, and transaction timestamps between custodians. Better portability reduces both delay and the uncertainty that can be priced as a discount. Evidence about the discount then becomes a performance test for the service, not a justification for permanent control.

The Number Resource Society can also organise the matched study across regions without claiming to speak for every operator. Participation rates, contributor composition, and funding should be published. The analysis should be re-runnable by independent researchers on protected data. An institution that defends operator rights should be especially careful not to manufacture a figure that serves its argument.

Reform should target the mechanism before anyone announces a regional figure

First reform is transparency of step times. Publish comparable percentiles, backlog age bands, information request counts, and bilateral handoffs. A region cannot refute delay allegations with a completion average that excludes unresolved cases.

Second is pre-filing eligibility clarity. Holders should know whether a resource is transferable, what status will result, and what documents prove authority. Buyers should know objective recipient conditions before negotiating. Pre-approval can help where needs review remains, but it should have a defined clock, a reasoned quantity decision, and a path for appeal.

Third is path compatibility based on registration integrity. A source registry should authenticate the holder and resource. A destination registry should authenticate the recipient. Differences unrelated to truthful, unique, technically coordinated registration should not silently close the path. Where a path remains closed, both institutions should publish the exact condition and evidence for maintaining it.

Fourth is limited discretion. New evidence requests should cite a policy clause and a fact. A refusal or reduction should identify the correction path. Review should be independent of the initial decision and fast enough to preserve the deal where possible.

Fifth is evidence portability and continuity. Corporate authority, chain of custody, status, RPKI, reverse DNS, and transfer history should not have to be reconstructed from one institution's private correspondence each time. A standard export reduces repeat cost while preserving verification.

These reforms are justified even before a price discount is estimated. Delay, opacity, and incompatibility are service defects when they cannot be tied to a necessary safeguard. Price research adds a measure of incidence: it shows who pays and how much of the cost reaches asset value or deployment.

The honest conclusion is a research plan, not a manufactured discount

It is reasonable to expect that institutional frictions enter IPv4 negotiations. Sellers care about the probability and date of payment. Buyers care about the probability and date of usable, recognised control. Brokers, escrow providers, and lenders price uncertainty. Incompatible paths reduce options. Needs review and unpredictable document cycles shift bargaining power.

This chain is an economic hypothesis with observable mechanisms. It is not yet a full empirical estimate. RIR public logs omit prices and failed negotiations. Private reports have selective coverage. Completed cases hide attrition. Regional labels mix directed paths and changing policies. Block quality, seller urgency, and entity sophistication can confound any simple comparison.

The correct study begins with serious commercial cases, preserves failures, matches plausible substitutes, separates registry time from party time, codes policy versions, and reports delay bands. It estimates seller proceeds and buyer all-in cost separately. It tests path openings and rule changes. It publishes balance, coverage, and uncertainty rather than one dramatic global figure.

If a robust discount survives, the conclusion should be narrow: under specified conditions and during a specified period, a directed institutional path reduced expected value relative to matched alternatives. That conclusion would support targeted reform, not prejudice against holders in a region. If no discount survives after matching, that result is equally valuable; it would challenge market folklore and redirect attention to block state or deal selection.

The most important figure today is therefore missing by design, not waiting to be guessed. No institution has published the complete panel needed to know it. The governance failure is not that a particular discount has already been proved. It is that organisations with power over timing and compatibility do not publish enough comparable evidence for operators to test whether their rules are capitalised in price.

Digital resources are global operational inputs. Their administrative path should be measurable, verifiable, and portable. A region should never be reduced to a price stigma. A slow or closed path should be treated as a repairable institutional condition, and the evidence should be strong enough to show whether the repair changes the deal.

Sources