RESEARCH DOSSIER
Insurance Distributor Churn and Market Mobility Dynamics
Abstract. Insurance distribution is a dynamic allocation of intermediating capacity rather than a stationary population of sellers. Individual producers enter, leave, return, change insurers or agencies, and cross territorial boundaries; intermediary firms transfer ownership or cease trading; customer portfolios move on a partially independent timetable. The measurement of distributor churn requires a longitudinal distinction between these events. This monograph examines the economic organization, regulatory observability, statistical identification, and market consequences of distributor mobility. Comparative institutional evidence from the European Economic Area, Hong Kong, India, the United States, the United Kingdom, and Australia is examined alongside research on intermediary economics, channel coexistence, and event-history methods. Three contrasting numerical cases demonstrate why net registration change, cohort licence retention, and appointment turnover measure different quantities. A methodological synthesis specifies risk sets, competing events, relational records, exposure denominators, and identification conditions for market-outcome research. The findings establish that competition, continuity of coverage, and geographic access depend on which relationships are interrupted or transferred, rather than on an undifferentiated rate of departure. The research is an institutional and methodological synthesis, including transparent calculations from published aggregates and illustrative examples; it does not estimate causal effects or claim a new named theory.
Keywords: Insurance intermediation; distributor churn; insurance agents; insurance brokers; producer licensing; carrier appointments; customer books; agency mobility; market entry; intermediary exit; competing risks; multistate models; market concentration; insurance distribution; coverage continuity; longitudinal data.
Introduction
A distributor is economically active when the institutional relationships needed to intermediate an insurance transaction are available to that distributor. These relationships include professional authorization, contractual access to risk carriers, information about customers and their exposures, operational resources, and a system through which advice, placement, and servicing can be delivered. Each relationship has its own duration. The loss of one relationship may reduce distribution capacity, redistribute it, or leave it substantially intact. The empirical object of this monograph is the evolution of those relationships and the consequences of their movement through the insurance market.
Producer churn denotes recorded turnover within a specified distributor population or contractual relationship during a defined observation period. Market mobility denotes movement between observed professional roles, intermediating organizations, insurer relationships, ownership arrangements, geographic markets, or licensing jurisdictions. The terms are complementary: turnover describes the frequency of transitions within a defined boundary; mobility describes their direction and institutional content. Precision requires declaring that boundary before selecting a denominator.
The research concerns distributors and their commercial relationships. Policyholder switching is an outcome or jointly occurring event, not a substitute for distributor mobility. Brokerage mergers matter here insofar as they change the opportunities, affiliations, or service responsibilities of intermediaries. A separate examination of merger valuation or consumer switching would address a different unit of analysis. Distinguishing these objects preserves the coherence of the parent scientific discipline, Insurance Distribution and Intermediation.
The investigation uses a comparative reading of regulator publications, registration guidance, published aggregate measurements, and selected academic research. The evidence is differentiated by access level: original reports and official guidance support their specific institutional and numerical statements; repository abstracts support only the findings visible in those abstracts. Public register descriptions establish which categories of administrative information potentially exist, not that a complete longitudinal panel has been collected. The case calculations below are reproducible from published numbers; hypothetical computations are labeled. No proprietary customer-level data, experimental evidence, independent survey, or external peer review is represented as having been obtained. All jurisdictional propositions refer to the identified instrument or published guidance, whose implementation and edition must be verified independently for legal or commercial use.
The substantive argument develops through seven chapters. The first establishes longitudinal identity; the second analyzes entry and exit; the third examines insurer relationships; the fourth studies customer-book continuity; the fifth considers market structure and incentives; the sixth evaluates comparative regulatory evidence; the seventh integrates research design and market outcomes. Their common premise is that observed transitions have meaning only relative to a stable population, a clearly identified relationship, and an explicit observation process.
Producer Population and Longitudinal Identity
Persons, legal entities, and productive relationships
Insurance distribution contains several populations. An individual professional may advise or solicit customers, exercise delegated authority, place risks, or work under a licensed firm. A corporate intermediary may hold appointments, employ licensed persons, and administer contracts independently of particular employees. A branch may serve a local market without being a separate legal entity. An insurer may itself distribute insurance. These distinctions align with the supervisory attention to intermediary activities and licensing in the IAIS (2024) Insurance Core Principle 18. The relevant legal treatment depends on jurisdiction and activity; the common research task is to identify the entity performing the function.
A person identifier, an entity identifier, and a relationship identifier therefore belong in separate observational tables. A licence number is a property of a regulatory authorization, not necessarily a permanent person identifier. Two concurrent authorizations may belong to one person; a regulatory migration may produce a replacement number; a legal entity may operate under several brands; a firm may retain its identity after changing owners. A data set that treats each new number or brand as a new person risks creating artificial professional entry and exit.
A longitudinal register should preserve the natural-person or legal-entity key as far as reliable evidence permits, together with authorization identifier, type, jurisdiction, validity interval, source, and extraction date. A separate appointment table records the intermediary, insurer, jurisdiction, applicable line or contract scope where known, and beginning and ending dates. An employment or affiliation table records intermediary-to-firm relationships, and a historical ownership table records the changing control of firms. These are conventional relational data-management distinctions applied to the insurance distribution problem; they do not presuppose a new economic theory.
The distinction becomes especially important when data are acquired through snapshots. Suppose a registered entity appears in a December extract and disappears in January. The data establish an observed disappearance within that interval. They do not identify the effective date of termination, the reason, the next occupation, or the status of customers. If the source subsequently restores the entity, the interruption may represent temporary inactivity, administrative delay, reinstatement, or genuine exit followed by return. Event-history research requires dated events or explicit interval censoring, rather than imputing exact dates without evidence.
Constructing the population at risk
A churn rate is interpretable when numerator and denominator refer to the same entity, eligibility rule, time, and geography. Consider an opening population of active individual producers. An exit incidence over the subsequent year can be computed as the number of opening-cohort producers who experience a first observed qualifying exit during the year divided by the number initially at risk, provided follow-up is complete. If entry occurs throughout the year, counting all exits and dividing by the opening population produces a flow-to-stock ratio, not the probability that a member of the opening cohort exits.
When exact person-time is available, a rate may be calculated as exits divided by the sum of observed time at risk. A person who enters halfway through a year and remains active contributes half a producer-year. A person who leaves after three months contributes one-quarter of a producer-year. The resulting rate has units of events per producer-year. It may exceed one where repeated events are permitted. For first-exit analysis, each person stops contributing time to the risk set at the first exit, censoring, or end of observation.
The definition of active must be distinguished from licensed, appointed, employed, and commercially productive. A person can retain a licence while recording no new business. A broker may maintain a portfolio and service renewals without new sales. An employee may change responsibilities while staying within an organization. A firm may retain its legal registration while ceasing to offer a product line. These conditions require distinct flags and, when available, independently recorded activity such as placing a policy or servicing an existing account. Regulatory authorization is often the most readily observed layer, but it is an imperfect proxy for output.
A defensible linkage process should begin with deterministic matches on stable identifiers, then investigate incompatible or missing keys using independently governed matching procedures. Name, address, and employer similarity can help locate candidates. They also risk conflating relatives, common names, shared offices, and successor firms. A record linkage study should report the frequency of uncertain matches, the sensitivity of results to treatment of those matches, and whether confidential identifiers can be used lawfully. Public reproduction should be based on appropriately aggregated or synthetic data when person-level disclosure is inappropriate.
Transitions and classification
The following conceptual distinctions organize the empirical chapters: first observed entry into a register; verified first entry into a profession; temporary absence; reinstatement or return; termination of a particular appointment; change of firm affiliation; relocation; corporate ownership transfer; business closure; and exit from insurance distribution across the study's observable boundary. Each event can be represented with an effective date or bounded interval, a source date, a predecessor state, and a successor state.
The categories should be coded at their own level rather than forced into a single mutually exclusive label across all layers. An individual can change agency while retaining an insurer appointment. A corporate acquisition may coincide with a brand change and appointment renegotiation while employees remain in place. Multiple transitions can occur on one date. A researcher modeling a single competing-risk process must decide which mutually exclusive event types are relevant to that particular process, while preserving overlapping events in the original relational data.
A change in the information system should also be recorded as an event in the observation process. The Hong Kong Insurance Authority's register transition guidance distinguishes arrangements associated with the introduction of its statutory licensing regime. Treating all newly issued identifiers during such a transition as first professional entry would convert administrative migration into a labor-market phenomenon. Administratively induced changes can be identified from source documentation, crosswalks, overlapping dates, and cohort restrictions.
The principal contribution of longitudinal identity is interpretive. A stationarity assumption about a distributor population is unwarranted when the people, appointments, and firms that constitute it are constantly changing. At the same time, a large number of administrative changes does not imply a large loss of productive capacity. Reliable measurement begins by assigning each observed event to its appropriate entity and relationship.
Entry, Retirement and Agency Exit
Stock-flow accounting
A population can grow even when attrition is substantial. If N0 and N1 denote the opening and closing stocks of a given population, E counts additions, X counts removals, and A denotes net corrections or boundary changes, then
N1 = N0 + E - X + A.
This identity is meaningful only when the stocks and flows use a consistent unit and marking convention. Repeated entry and removal of the same person may legitimately contribute to event counts, while a deduplicated cohort analysis counts persons under a different definition. Corrections, inclusion of a newly reporting jurisdiction, or migration of a register must not be silently absorbed into professional inflows.
The identity helps establish a data-quality check: if published movements fail to reconcile, the discrepancy should be investigated before comparing rates. Agreement of the identity establishes accounting consistency, not accuracy of every classification. For example, a cancellation followed by reinstatement can balance the stock exactly while generating two observations that have no equivalence to permanent occupational exit.
The difference between gross and net movement is economically consequential. A rapidly expanding distribution market can recruit many new agents while experiencing weak retention. A stagnant market can retain experienced agents but fail to recruit replacements. An identical net stock change can be generated by many combinations of entry and exit. Consequently, net growth cannot identify workforce stability, the frequency of disrupted customer relationships, or the cost of recruitment and training.
Entry pathways and occupational selection
Entry into insurance distribution can follow employment with a captive insurer, independent agency work, corporate brokerage recruitment, or transition from another financial occupation. The relevant commitment includes licensing expenditure, supervision, training, customer acquisition, technology access, and the opportunity cost of time. The degree to which these costs are borne by the producer, a sponsoring insurer, or the employer influences incentives to enter and remain.
Early exits may occur where anticipated commissions, sales opportunities, hours of work, or professional requirements differ from expectations. Experienced distributors may leave because alternative employment, geography, family circumstances, business succession, or organizational strategy changes. These mechanisms are plausible economic explanations rather than separately established causal findings. Observed early departure is compatible with both voluntary selection and organizational termination; administrative records usually cannot discriminate between them without supplementary evidence.
Entry cohorts should therefore be analyzed by initial relationship type, product line, cohort year, and the observable recruitment environment. Baseline experience is particularly important. A first observed appointment in one insurer does not establish a first entry into the profession. A new licence after a lapse can represent an experienced returner. Recruit-level retention analysis should distinguish first licensing, transfer from another jurisdiction, reinstatement, migration of an existing register, and first observed appearance after the study begins.
Cohort evidence from Hong Kong
The Insurance Authority (Hong Kong, 2025) reports in its intermediary turnover analysis approximately 70,600 first-time entrants from the start of the statutory licensing regime on September 23, 2019, through December 31, 2024. Approximately 44,000 remained licensed at the end of 2024, corresponding to about 62.3% of the reported entrants. The source reports rounded counts. Its separate 2019–2021 entrant group comprises about 29,600 people, of whom approximately 10,500 remained licensed, with about 19,100 no longer licensed at year-end 2024. The corresponding endpoint retention is 35.5% and the endpoint absence share 64.5%, calculated from the rounded figures. Approximately 15,600 left before the end of their first three-year licensing period, around 52.7% of that entrant group.
These proportions describe the source's definitions and observation dates. They are not annual probabilities, and the 2019–2021 entrants have unequal durations of follow-up at the common endpoint. The partial 2019 recruitment period also differs from a full calendar year. The full-period 62% active share includes younger cohorts with less exposure to potential exit; comparing it directly to the older cohort's 35% as evidence of an improving retention trend would confound cohort composition and follow-up.
The same publication differentiates first entrants from returning or experienced individuals, licence-type changes, and the transition of deemed licensees. These exclusions are not incidental. They determine the cohort's meaning and explain why total changes in licences can exceed the number of people entering or leaving the industry. The regulator also discusses the risk of orphaned life policies and remuneration arrangements. Those concerns motivate research on servicing continuity; the published cohort counts do not establish the causal effect of a compensation practice on individual departure.
A researcher with person-level event histories could estimate first observed loss of authorization at specified durations, reporting censoring and reinstatement separately. The public figures support a narrower but important result: distributor retention depends on the recruitment boundary, the distinction between new and experienced participants, and the interval during which a cohort has been exposed to exit.
Retirement, succession, and business closure
Retirement raises a special identification problem. A producer's age is informative about potential career stage but does not directly measure retirement. Some older producers reduce new business while continuing to service existing customers; others sell a firm, remain advisers, or retain a licence after transferring day-to-day responsibilities. Younger individuals leave for other reasons. A researcher should use a directly recorded retirement reason, a survey with defined response categories, or a carefully qualified proxy rather than equating exit with an age threshold.
A succession transaction can result in economic continuity with a change of control. The outgoing owner may retire, yet the agency's legal identity, appointments, employees, and customers remain. Conversely, termination of a firm with no successor may leave customers without the familiar service channel even when the former owner remains licensed elsewhere. The occupational exit and the firm survival estimands serve different scientific questions.
Agency closure also needs an explicit rule: dissolved legal entity, expired registration, loss of all observed relevant appointments, or cessation of commercial activity. Each proxy identifies a different boundary. Product-line withdrawal should generally be recorded separately from total closure, since an agency ceasing commercial motor placements may still distribute personal lines or life insurance.
The Agency Universe Study overview (Big I, 2024) estimates approximately 39,000 independent property-casualty agencies in the United States. This industry estimate concerns organizations in a defined channel. It does not enumerate all licensed individuals, direct insurer sellers, or exits during the year. A stock estimate of agencies, even when repeated across surveys, cannot establish the probability that a particular agency survives or the number of producers who leave. Sampling changes, definitions, and mergers must be addressed before deriving longitudinal conclusions.
Economic significance of exit patterns
Exit matters through the resources that must be replaced and the service relationships that must be maintained. Recruitment and supervision can create costs before revenue is generated. Experienced intermediaries may possess specialized market knowledge, familiarity with policy conditions, and client-specific information. Their departure can change productivity or service quality even where nominal headcount is maintained. The impact is heterogeneous: replacing an adviser serving complex industrial risks may require different capabilities from replacing a distributor of a standardized personal product.
Retention alone is therefore an insufficient performance objective. A high retention rate can coexist with poor advice, weak service, restrictive market access, or inefficient attachment. Entry and exit can facilitate reallocation toward better matches. Analysis should distinguish service disruption and lost specialized capacity from benign turnover, and evaluate the distribution of outcomes across customer segments. The desirable inference follows from measured outcomes, not from an assumption that any particular level of churn is optimal.
Carrier Appointments and Network Transitions
Appointment as a distinct economic and legal relation
An insurer appointment represents a particular relationship between an intermediary and a carrier under the relevant legal and contractual arrangement. It can authorize distribution of designated products or activities, subject to limitations and the applicable law. A licence represents the regulatory permission to undertake defined activities. Employment records describe another relationship. These layers can coincide in some distribution systems but remain analytically distinct.
The NAIC (2005) Producer Licensing Model Act demonstrates the distinction through its optional appointment provision in section 14 and termination notification provision in section 15. Section 15 of the model specifies notification within 30 days of the effective termination in the stated circumstances. The model is an instrument for state adoption; its provisions are not asserted as the uniform law of every US jurisdiction. An appointment termination recorded under a state's implementation cannot, by itself, establish that the producer has surrendered a licence, ceased all insurance work, or transferred any customers.
One individual or intermediary firm may have several appointments. The loss of one insurer can narrow the set of available products while leaving the producer active. Loss of the last observed appointment may have stronger implications, particularly for a role that requires such a relationship, but the inference depends on register coverage, other roles, and the jurisdiction's rules. A new carrier relationship may supplement an existing portfolio rather than replace it.
Appointment data should retain insurer identities across corporate reorganizations. A change of legal underwriting entity within an insurer group could create a new contractual relationship with limited change to commercial access. Where evidence permits, analysis should report separate carrier-level and ultimate-group-level transitions. The study should also distinguish authority to quote, authority to bind, access via wholesale partners, and direct appointment if these details are relevant and actually available.
Gross relationship flows in India
The IRDAI Annual Report 2024–25 provides a useful reconciliation for the life-insurance agency channel. Its published table reports an opening agency stock of 2,895,438, appointments of 1,115,026, terminations of 887,020, and closing stock of 3,123,444. The difference is 228,006, exactly equal to appointments less terminations. Using the opening stock as a common denominator, appointments are 38.51%, terminations 30.63%, and net growth 7.87%. Combined gross movement is 69.14% of the opening stock. These are flow-to-opening-stock ratios.
The public-sector row begins at 1,414,743 and records 494,530 appointments and 422,422 terminations, giving 1,486,851 at closing. The corresponding private-sector counts are 1,480,695, 620,496, 464,598, and 1,636,593. Both rows reconcile. The report's note concerning exclusion of Sahara India Life following business transfer to SBI Life is material to coverage of the published series. The source constitutes an institutional annual account, not a deduplicated panel of all Indian insurance professionals.
These relationships illustrate how large appointment flows can underlie moderate net expansion. However, terminations can include people who were appointed during the year, agents moving among insurers, and repeated relationships. The 30.63% flow-to-stock ratio must not be called the probability that a member of the opening cohort leaves the insurance profession. Nor can the combined gross flow be interpreted as the percentage of unique agents who changed insurer. Identifying either requires linkable individual histories.
A simple hypothetical transfer clarifies the problem. If an intermediary leaves insurer A and joins insurer B within the same measured market, one termination and one appointment occur. The person remains in insurance distribution. If the individual also transfers business to B, that customer movement is an additional event. A market-level agency count might remain unchanged even though network composition changes and incumbent customers need to decide whether their contracts should move.
Contractual mobility incentives
Appointment duration reflects the expected value of the relationship to both parties. From a producer's perspective, carrier access can generate commission revenue, suitable coverage for customers, reliable claims support, and operational convenience. From an insurer's perspective, an appointment offers access to customer demand and distribution expertise while introducing monitoring and servicing obligations. The relationship is viable when both parties expect its continuation to be advantageous, subject to contractual and regulatory requirements.
The value of a relationship may change after premium adjustments, underwriting restrictions, changes in capacity, altered commission schedules, different technology requirements, or a shift in the producer's customer mix. An insurer can alter the range of products available to distributors without terminating every appointment. Thus the economically relevant exposure is often product-specific and time-varying rather than a simple binary affiliation.
The abstract of Cummins and Doherty's (2006) Economics of Insurance Intermediaries describes commercial property-casualty intermediaries as market makers and identifies compensation components tied to business characteristics. This supports attention to the placement and incentive functions of intermediaries. It does not establish a universal relationship between commission design and distributor exit. Comparative treatment of compensation should consider timing, contingency, continuing service obligations, and the possibility that remuneration changes coincide with other shifts in insurer strategy.
Consider a producer evaluating two possible carrier portfolios. An established carrier offers familiar systems and predictable processing; an alternative offers better coverage for a subset of customers but imposes onboarding costs. The producer's expected net benefit depends on product suitability, demand, revenue, the cost of transition, and service obligations. A difference in remuneration may matter without being decisive. Where an insurer's underwriting capacity contracts, the availability of alternative carriers may dominate compensation entirely.
Network representation
A bipartite network represents intermediaries on one side and insurance carriers on the other. An edge denotes a specified observed relationship, such as a direct appointment during a month. The degree of an intermediary is the number of distinct connected carriers; the degree of a carrier is the number of observed intermediaries connected to it. Weighted variants can describe placed premiums or policy volume when those measures are available and comparable. Appointment counts and premium weights should not be treated as interchangeable.
The analysis should preserve the temporal network. A static graph formed by pooling many years gives an intermediary apparent simultaneous access to carriers that may never have been available together. A monthly graph allows estimation of the number of relationships added and removed, degree distributions, persistence, and the number of alternative carriers connected to a particular intermediary. It should also distinguish direct links from mediated access through wholesale brokers or delegated arrangements.
For example, an intermediary with three product-relevant carrier relationships loses one. If the remaining two offer comparable capacity and the affected customers can remain covered, the reduction in appointment degree may have limited immediate welfare significance. If the terminated carrier is the sole market for a specialized risk, even unchanged aggregate degree across unrelated products does not preserve the original customer opportunity. Interpretation requires product and geography, not simply network degree.
A carrier's termination of many appointments on the same date may indicate a coordinated book withdrawal, contractual redesign, merger integration, or administrative cleanup. Observed clustering can motivate investigation of common shocks. It also violates the assumption that appointment failures are independent events. Inference should account for dependence among producers exposed to the same insurer, agency group, or regulatory event.
Customer Books, Portability and Switching Frictions
What is transferred when a distributor moves?
A customer book comprises a set of economically valuable relationships and associated records, obligations, and expected future business. Its commercial value can depend on expected renewals, service revenue, customer loyalty, contractual permissions, information quality, and the costs of maintaining it. Its legal treatment depends on applicable contracts and laws. A book should not be represented as a universally transferable property right over customers, who retain their own rights and choices under the relevant arrangements.
An existing insurance policy is a contract with an insurer. A change in producer or servicing intermediary need not alter the insurer, the policy's risk-bearing obligations, the coverage period, or the premium. Conversely, a renewal could go to a different insurer while the broker remains the same. At least four observable situations should be separated: same intermediary and carrier; different intermediary with same carrier; same intermediary with different carrier; and movement of both. A fifth type, lapse or nonrenewal without replacement, is particularly important when studying protection.
The FCA's Wholesale Insurance Broker Market Study (2019) reports relationship considerations and infrequent switching in its examined wholesale context. The accompanying demand-side analysis describes qualitative evidence of clients following individuals who move between brokerage firms. Together these materials show why the distributor's personal mobility and the customer's organizational mobility can be related. Neither report establishes a representative producer-level churn rate or a universal percentage of customers that follows departing employees.
To study co-movement, one must observe both the effective departure of a producer and the customer's subsequent service or placement assignment. If a policy remains with the same insurer and a replacement servicing broker is assigned, the customer relationship changes while insurance coverage can remain intact. If a producer's employer changes but the customer relationship persists, a firm-level analysis will record redistribution while a person-level analysis will record continuity.
Transaction costs and information
The customer's transition cost includes searching for an appropriate distributor, conveying information about exposures, understanding alternative coverage, assessing service quality, and arranging documentation. Commercial customers may need specialized knowledge about historical claims, risk modifications, contractual obligations, or coverage exclusions. Personal customers may be sensitive to time, trust, accessibility, and ease of renewal. The importance of each cost depends on the product and the customer's circumstances.
Claims-history information is especially relevant in some commercial lines. The Irish CCPC's (2020) public liability insurance study identifies information access as a potential obstacle in the insurer-switching context. This evidence concerns insurer selection and information frictions, not a direct measurement of broker departures. Its transferable analytical relevance lies in the dependence of financial matching on usable records.
Information portability and service portability are separate. The ICO's data portability guidance addresses a defined personal-data right, including conditions on the information covered and its transfer. The guidance is under review and is not a determination of rights in every insurance transaction. An individual's rights in personal data do not establish that an employee, agency, or successor broker owns or is entitled to transfer an entire insurance book. Contracts, confidential commercial information, consent or another lawful basis, data-protection duties, and record-retention requirements may all bear on a specific transfer.
A scientific data design should record authorized transfers and distinguish them from incomplete observation. It should not assume that a missing customer identifier implies that a customer departed; a customer's record may have been pseudonymized differently, retained in a separate system, or become unavailable to the researcher. Matching customer data across firms requires appropriate permission and governance, particularly where health, financial, or other sensitive information is involved.
Measuring renewal continuity
A renewal analysis begins with policies due for renewal within an observation window. The denominator might be all eligible policies, or a defined subset whose servicing producer departed sufficiently before renewal. Outcomes could be renewed with the same insurer; renewed with another insurer without a gap; lapse or nonrenewal followed by new coverage after a gap; and unresolved status. Policyholders with multiple policies should not be counted as distinct persons unless that is the explicitly chosen unit.
Suppose a departing producer services 200 policies approaching renewal, and 160 renew on schedule with the original insurer, 20 move to another insurer with no gap, 12 have a confirmed interruption, and 8 cannot be linked to an outcome. This hypothetical example yields 180 known continuous renewals, 12 known interruptions, and 8 unknown outcomes. The known uninterrupted fraction of all 200 policies is 90%; the known interruption fraction is 6%. Among the 192 whose outcomes are known, the interruption fraction is 6.25%. If every unresolved case involved interruption, the overall interruption fraction would be 10%. Reporting both bounds makes the missing-data dependency explicit.
Even timely renewal does not establish equivalent protection. A replacement contract may have different limits, exclusions, deductibles, conditions, or insured entities. The empirical analysis of welfare consequences needs coverage detail appropriate to the product. Differences in premium alone cannot determine value, particularly when coverage changes.
Customer-level effects also depend on event timing. A producer's departure one week before renewal may create a different service burden from departure several months before renewal. A planned succession with communication, reassignment, and access to records may preserve continuity. Abrupt departure in the absence of replacement capacity may increase uncertainty. Comparing these cases requires data on handover arrangements and time between distributor change and coverage expiration.
Book economics and incentive alignment
A book valued as a stream of future net revenue is sensitive to customer persistence and cost of service. In a simplified hypothetical calculation, 100 customers each generate annual net service revenue of 100 monetary units, annual retention is 90%, and the annual discount rate is 10%. With first revenue received in one year and a constant geometric decline thereafter, the present value is 10,000 divided by 1.10 times the reciprocal of one minus 0.90/1.10, or 50,000 monetary units. This is a stylized present-value illustration; it omits growth, taxes, contractual restrictions, servicing heterogeneity, and uncertainty.
Its relevance is that retention changes expected economic value. A producer who expects customers to follow to a new brokerage may evaluate a move differently from one whose customer relationships are primarily attached to the incumbent firm. The firm may accordingly invest in team-based service, contractual arrangements, customer communication, and succession planning. The customer's right to choose a provider and the applicable legal framework remain independent constraints.
Compensation structures can also assign acquisition and servicing revenue on different schedules. A departure occurring after acquisition remuneration has been paid but before many service obligations have arisen can create asymmetric organizational costs. Conversely, continuation of an uneconomic relationship can trap productive capacity. The Hong Kong turnover analysis raises servicing concerns in connection with remuneration and orphaned policies. Its institutional observations support posing the research question; establishing its magnitude or causal direction requires transaction-level evidence.
The appropriate outcome is the quality and continuity of service and protection, considered alongside freedom of choice. A market in which customers can readily move with a preferred distributor may support competition; a market in which departure leads to interrupted or unsuitable coverage may impose costs. Evaluating these effects requires observing the policy and service relationships after the distributor event.
Market Concentration and Mobility Incentives
Market boundaries and concentration
Distributor mobility occurs within a market defined by product, geography, customer segment, and available alternatives. A national count of producers can conceal concentration of specialized placements within a small number of brokers or underwriting markets. A count of legal entities can overstate independent competitive decision makers where multiple subsidiaries are controlled by one group. Conversely, aggregate ownership data can understate local operational diversity where teams serve distinct segments.
The Herfindahl-Hirschman Index is the sum of squared market shares. The US Department of Justice's HHI explanation defines the familiar percentage-share convention. In a hypothetical four-firm market with shares 40%, 30%, 20%, and 10%, the HHI is 1,600 + 900 + 400 + 100 = 3,000. Combination of the two smaller firms, with other shares unchanged, produces shares 40%, 30%, and 30%, giving 3,400. The difference is 400, equal to twice the product of the combining percentage shares. These calculations illustrate arithmetic, not a legal conclusion under any merger regime.
The relevant shares might measure brokerage revenue in a defined market, placements, or another appropriate indicator of economic activity. Premium volumes reflect the cost and quantity of risk transfer and may be affected by rate changes. Headcounts describe labor resources rather than directly measuring market revenue or customer choice. The measure must conform to the scientific question and to comparable definitions across firms and periods.
A network may exhibit high intermediary concentration without equally high carrier concentration, or the reverse. Intermediary exit can therefore affect customer access even where the number of risk-bearing insurers is stable. If a specialist brokerage closes, customers may lose an effective placement route to otherwise active insurers. Conversely, several competing intermediaries may depend on the same narrow set of underwriting sources.
Mobility and channel diversity
Insurance distribution encompasses organizational forms whose economic functions differ. The abstract of Berger, Cummins, and Weiss (1997), multiple distribution systems, reports evidence supporting a product-quality explanation for coexistence of property-liability distribution channels. The result cautions against treating a channel's cost or productivity measure as a complete description of its economic role. It does not establish that one organizational form has a particular distributor retention rate.
Mobility between captive and independent channels, corporate firms and network structures, or geographically differentiated markets may reflect a change in the production of intermediation services. A move can widen access to carriers, expand specialization, reduce administrative costs, or alter the client's service pathway. It can also narrow the feasible market if relationships, data, or specialist resources are lost. Comparison requires information on the attributes of the origin and destination, not just transition counts.
The expected value of moving depends on the producer's outside opportunities and the cost of transition. A distributor specializing in a market with several competing employers and accessible contractual relationships may have more alternatives than one tied to a single local organization. Higher mobility can discipline compensation and service arrangements where alternatives are genuinely available. But observed low mobility can also reflect satisfaction or accumulated relationship-specific capital. Nonmovement alone is not evidence of exclusion.
Consolidation, transfers, and competitive effects
A merger can change group concentration immediately without changing the number of licensed persons. It may also generate later staff departures, office closures, network changes, or redistribution of customer relationships. These are separate outcomes. An acquisition announcement can affect behavior before the transaction closes; a failed transaction can affect expectations without changing ownership.
The US Department of Justice's (2021) Aon and Willis Towers Watson case records a challenge to a proposed combination and its subsequent dismissal following abandonment of the transaction. The case illustrates why announced mergers, challenged mergers, completed mergers, and realized changes in market concentration must be coded separately. The proposed transaction cannot be used as an observed completed-merger treatment or as evidence of realized postmerger distributor departures.
For a completed transaction, a researcher should establish the closing date, affected legal entities, ownership crosswalk, product and geographic overlap, and intended operational changes. An observed decrease in entity count could be a reporting consolidation. A change in person-level affiliations could be administrative reclassification. A decrease in available services needs independent evidence, such as the closure of offices, altered placement opportunity, or measurable customer outcomes.
Mobility analysis can also expose concentration in recruitment and visibility. If most departing producers move to a single successor firm, the destination distribution differs from a market with many receiving firms. Origin-destination counts and group-level concentration can identify these patterns. The inference should account for total size: a large firm can receive many new entrants simply because it hires more workers, not because it has a uniquely strong tendency to capture competitors' producers.
Local access and distribution resilience
A market with numerous appointment relationships may still lack accessible service in a particular locality. Geographic analysis should distinguish location of a registered office, location of a producer, service territory, and residence or business location of customers. Digital distribution can reduce some geographic constraints while leaving others, such as local language, specialized advice, or service standards, in place. Registration address alone does not establish operational service availability.
A local access indicator can measure the number of active, product-relevant intermediaries within a defined service area. A capacity indicator might count carriers reachable through them. Both require a product scope and a date. Better evidence could include eligible inquiries, quotations, bind rates, wait times, renewal gaps, and reasons for unsuccessful placement. These outcome measures distinguish nominal presence from usable access.
Consider two local markets that each lose one of five intermediaries. In the first, the other four can absorb clients and reach the same carriers. In the second, the departing firm is the only distributor of a particular specialized cover. The equal headcount change of 20% has different implications. A network-aware assessment would identify the second market's loss of a specific placement path and investigate whether alternatives actually became available.
Resilience also involves correlated shocks. A change in an insurer's distribution strategy may remove common access across many firms. A technology outage can interrupt distributors relying on the same platform. A regulatory change can affect a particular type of licence or role. Diversity of firms, carriers, and operational arrangements matters insofar as it provides viable substitutes when a relationship fails.
These analyses show why a concentration statistic should be paired with mobility and outcome evidence. Concentration can describe structure; movement can describe reallocation; coverage and service outcomes can describe consequences. None alone establishes the welfare effect of distributor churn.
International Registration and Licensing Evidence
Comparative framework
International comparison begins with a mapping of regulatory categories to research units. The IAIS (2024) intermediary supervisory standard supplies an internationally relevant supervisory framework while allowing jurisdictional implementation. Studies of actual mobility must observe the categories and authorization arrangements established by each jurisdiction. An international standard does not create a single global registry or a uniform producer licence.
A comparative register table should specify the competent authority, legal population, individuals versus entities, coverage of ancillary or direct channels, dates, status categories, appointment information, treatment of cross-border activity, and availability of historical records. Any comparison should then be restricted to a harmonized subset where meaningful or presented as a comparison of distinct institutional quantities.
European Economic Area registration series
EIOPA's (2026) third IDD application report provides registration statistics for 2020–2024. The aggregate values reported in Annex I are 884,337 in 2020; 860,634 in 2021; 834,360 in 2022; 824,402 in 2023; and 818,128 in 2024. The endpoint change is minus 66,209, or minus 7.49% using the published endpoint totals, consistent with the report's rounded 7.5% decrease.
The table is based on reporting by national competent authorities, with differing coverage across years; some jurisdictions begin reporting after 2020 or have gaps. The annex identifies these differences. Consequently, the aggregate change can combine genuine changes in reported national stocks with variation in reporting coverage. It does not estimate a balanced-panel gross exit count. A researcher seeking like-for-like stock change should construct a common jurisdiction sample from the underlying series, verify changes in categories, and report both the restricted and published aggregate results.
Registration is a stock of authorizations or registered intermediaries under the report's definitions. It does not enumerate all sales staff or establish the number of people who retired, joined another channel, changed insurer, or retained customer relationships. Separate individual and corporate counts should not be combined without checking national units. The earlier report uses a different reporting set, which further limits comparisons based solely on headline aggregates.
EIOPA's intermediary register portal provides access to national registers maintained by competent authorities. It is an infrastructure for checking registration rather than an acquired panel of every European distributor. The 2026 report also describes growth in cross-border passport counts of about 12% from 2020 to 2024, while identifying limits to evidence of actual cross-border activity. Permission to operate across a border is analytically different from observed placements, offices, or customers in the host state.
Hong Kong statutory licences
Hong Kong's register and published cohort analysis provide two complementary perspectives. The register information documents arrangements associated with the transition to statutory licensing. The 2025 turnover article examines first-time entrants and distinguishes them from other licensing movements. This supplies an example of a regulator explicitly separating new recruitment from administrative or career reclassification.
The cohort figures provide a basis for retention analysis within the defined covered population. They are unsuitable for direct comparison with the EIOPA net registration change because the numerator, denominator, unit, and time exposure differ. A harmonized cross-jurisdiction project would need individual first-authorization dates, comparable active-status definitions, a fixed follow-up duration, and rules for transfer, reactivation, and censoring.
United States producer and appointment records
In the United States, producer licensing and appointments operate through state regulatory systems. The NAIC model act is useful for understanding the distinction between licence and appointment without substituting a model provision for actual state law. A producer working across states can generate several licensing records, and interstate licensing should not be interpreted as an equivalent number of unique workers.
The NIPR reports and alerts service describes licensing, appointment, and regulatory action information available through its products. Such records can be potentially valuable for linking person and carrier histories, but this research did not purchase or obtain a subscriber-level longitudinal extract. The public product description establishes potential information categories, not the completeness, availability, cost, or lawful access arrangements for a particular research panel.
A study using these records should establish whether the temporal information represents effective dates, reporting dates, or both; how terminations and reinstatements are recorded; and how absent appointments are handled in jurisdictions or roles that do not use the same appointment practices. Matching individuals across states and carriers requires a stable authorized key. State-level changes in adoption or data provision should be accounted for as observation and institutional changes.
United Kingdom and Australia
The FCA's Financial Services Register describes the relationship of appointed representatives to authorized principals and indicates that an appointed representative may have more than one principal. A transition whereby one principal relationship ends may therefore differ from exit of the firm or representative from all relevant regulated activities. The scope of responsibility and permissions must be identified for each observed relationship.
Australia's ASIC AFS representative guidance distinguishes authorized representative arrangements from certain employee and director roles, and describes circumstances involving multiple AFS licensees. Counting only authorized representative appointments may therefore omit employees operating under a different relevant arrangement. These public institutional descriptions demonstrate the need to map the regulatory relationship before calculating churn. They are not a comprehensive statement of current UK or Australian law for every category of insurance activity.
Both systems illustrate the relationship-centered nature of mobility. A firm may acquire or lose authority through a principal or licensee without necessarily dissolving; an individual may change employer without acquiring the same type of register entry as an independent representative. A panel spanning both countries would need category-specific indicators rather than a single pooled count of “agents.”
Comparability of the three quantitative cases
The evidence supports three distinct statistical propositions. The EIOPA published aggregate declined by about 7.5% between two endpoints. The Hong Kong regulator reports substantial observed absence from the older first-time entrant cohort at a common endpoint. The India life-insurance agency table reconciles high volumes of appointments and terminations with positive net growth. All are meaningful within their own definitions.
There is no defensible way to rank the three markets by the figures 7.5%, 64.5%, and 30.63%. The first is a multi-year net stock change with variable reporting coverage; the second is an endpoint cohort absence proportion with unequal follow-up; the third is a one-year relationship flow normalized by an opening stock. Their signs, magnitudes, and durations measure different events. A league table would produce numerical precision without scientific comparability.
International measurement should publish a concordance instead: the institutional unit, event type, risk set, denominator, observation period, data completeness, and interpretation for each study. Cross-market rankings become meaningful only once these elements are comparable. Until then, comparative institutional analysis is more informative than pooled numerical aggregation.
Duration Models, Networks and Market Outcomes
Duration, risk sets, and censoring
Event-history analysis provides a way to describe time until a specified transition while respecting unequal follow-up. A first-exit survival function S(t) is the probability that an initially eligible individual remains free of the defined first-exit event through duration t. In a simple setting with exact event times and independent right censoring, the product-limit estimator multiplies the terms one minus the number of exits divided by the number at risk at each event time. Different entry cohorts can then be compared at common durations, subject to coherent definitions and censoring assumptions.
Left truncation occurs when a subject enters observation after already spending time in the state of interest. A register first acquired in 2024 may show people who have been licensed for many years. Treating them as 2024 entrants changes the estimated duration distribution. Where actual entry dates are known, delayed entry can define when individuals begin contributing to the observed risk set. If entry dates are unavailable, the analysis should use a clearly defined observation-time estimand, not claim career-duration survival.
Right censoring occurs when a person's event status after the last observation is unknown. A subject active at study end is right-censored for a first-exit analysis, not classified as permanently retained. A lost linkage or cessation of a data supplier may be informative censoring. Sensitivity analysis should consider whether individuals with incomplete records differ systematically from fully observed individuals.
Interval censoring is important when only periodic snapshots exist. A licence active in January and absent in March implies that the observed disappearance occurred between these dates, assuming the snapshots are complete and comparable. Assigning February 1 to every such event may be computationally convenient but creates artificial temporal patterns. Methods appropriate for interval-censored events or explicitly stated discrete observation intervals are preferable.
Competing events and multiple states
Suppose persons initially work within a defined carrier agency network. Their first event can be a move to another network, a transition to another insurance role, retirement, or another observed exit. The first events compete because occurrence of one precludes simultaneous classification into another first-event type under that model. Putter, Fiocco, and Geskus (2007) provide a methodological competing risks and multistate tutorial; the application to insurance distributor relations is the present study's synthesis.
The cumulative incidence of cause k by time t can be expressed as the integral from zero to t of the probability of remaining free of any first event immediately before each time, multiplied by the cause-k hazard increment. Estimating one minus a single-cause survival curve while treating competing events as ordinary censoring can overstate the actual probability of the event of interest. The distinction matters when retirement and carrier transfer are common alternatives.
A hypothetical discrete illustration uses 100 people initially in a state. During the first period, 10 transfer, 5 leave the profession, and 85 remain. During the second period, 8 of the remaining 85 transfer, 7 leave, and 70 remain. The two-period cumulative incidence of transfer is 18%, and of professional exit 12%; 70% have experienced neither first event. The second-period conditional transfer probability is 8/85, or 9.41%. The cumulative transfer probability is not the unweighted sum of that conditional percentage and the first-period 10%.
A multistate approach extends the analysis to repeated or sequential transitions. States could represent active affiliation with insurer group A, with group B, with an independent network, temporary absence, or confirmed professional exit. Where multiple appointments coexist, states or separate marginal processes must reflect that multiplicity. Reentry may make recorded occupational exit a nonabsorbing state. Confirmed death may be absorbing; temporary nonlicensing is not.
A transition matrix P for a fixed period has entries p(i,j) giving the conditional probability of ending in state j given beginning in state i, for the population and observation convention. Each row sums to one. Under a homogeneous first-order Markov assumption, P squared gives two-period transition probabilities. That assumption should be tested or justified: time since entry, previous carrier affiliation, duration of absence, or prior appointments may predict future movement even conditional on current state. Duration dependence and historical variables can be incorporated without claiming that a simple Markov model captures all relevant mechanisms.
Discrete-month models may be particularly practical when administrative dates are recorded only at monthly resolution. Cause-specific models can estimate associations between covariates and each type of event. A researcher should separately report predicted transition probabilities and uncertainty when probabilities, rather than relative hazards, are the substantive outcome.
Recurrent relationships and clustering
Appointment events may recur. A producer can add a carrier, terminate the relationship, and later renew it. A first-event design discards subsequent mobility by construction. A relationship-level recurrent-event design can capture it, with the risk set consisting of eligible active appointment spells. An individual-level analysis of any appointment change answers a different question. Both designs should state how simultaneous events and multiple exposures are treated.
Events within the same person, intermediary firm, insurer group, and market may be correlated. A compensation change implemented across all agents of one insurer is a common exposure. A merger can affect many people through a shared event. Standard errors that presume independent person-period observations can misstate precision. Depending on the research design, clustered or multilevel methods, or other appropriately justified dependence adjustments, may be needed.
Network observations also involve dependence. If two intermediaries share an insurer whose product line closes, the induced changes are not independent. Descriptive analysis can report edge survival, changes in degree, and availability of alternative paths. Causal analyses should account for shared exposure and possible spillovers. For example, a treatment that causes one producer to exit may shift customers and workload to untreated producers, altering their outcomes.
Identification of economic causes
A descriptive association between commission policy and exit can reflect selection. An insurer may alter commission schedules in response to weak performance or strategic changes. Agents experiencing declining sales may seek other carriers before the new policy begins. A comparison of those exposed and unexposed to the policy may then combine the effect of the policy with existing differences.
A research design should first define the target effect. One plausible estimand is the difference in twelve-month first-appointment termination probability caused by a specified remuneration change for initially appointed agents in the affected carrier, relative to the counterfactual under no change. Another is the effect of that change on customer renewal gaps. These effects differ in outcome, risk set, and potentially in timing. A causal claim requires identification assumptions and appropriate data, not merely a regression coefficient.
A difference-in-differences design could be used if comparable carriers or markets change a policy at different times and untreated or not-yet-treated groups provide a justified counterfactual. Callaway and Sant'Anna's (2021) multiple-period difference-in-differences addresses staggered treatment timing under specified parallel-trends conditions. An application should establish treatment dates, investigate pre-event trends, identify anticipatory or spillover effects, and report heterogeneous effects rather than treating every change as identical. The abstract verifies the methodological relevance, while this monograph does not implement the estimator.
Natural experiments require a plausibly exogenous source of variation with a well-defined affected population. A regulatory rule change may provide variation, but implementation can coincide with other reforms, economic shocks, or reporting changes. Even where a reform affects appointments, it does not necessarily affect professional exit or customer outcomes through the same channel. The pathway from treatment through relationships to outcomes should remain explicit.
A matched cohort study can improve comparability on measured variables, such as tenure, product line, geography, and previous activity. It cannot automatically eliminate unmeasured differences in ability, customer relationships, or employer strategy. Sensitivity analysis should consider alternative risk-set definitions, selection into observation, treatment timing, missing outcomes, and uncertainty in linkage. Results from one channel or jurisdiction should retain their external-validity limits.
Outcomes and the empirical welfare question
The consequences of distributor mobility can be organized into producer, firm, insurer, customer, and market-level outcomes. Producers experience income, employment continuity, costs of acquiring new authorization, and changes in opportunity. Firms experience recruitment and service costs, customer retention, changes in capability, and ownership transitions. Insurers experience redistribution of distribution access and placement flows. Customers experience service continuity, quality, costs, and availability of suitable protection. Markets experience changes in effective alternatives, geographic access, and concentration.
These outcomes have different denominators. Producer income is measured per person or job spell; firm revenue per organization; placement premium per contract or period; coverage continuity per eligible policy; claims outcomes per claim or exposure; availability per inquiry or relevant customer population. An integrated panel should maintain these distinctions instead of combining heterogeneous outcomes into an unvalidated index.
An illustrative outcome study might follow policies due for renewal in the six months surrounding a distributor departure. The exposed group contains policies whose designated service relationship is interrupted before renewal; a comparison group contains similar policies with an uninterrupted relationship. The outcomes are known coverage gaps, time to reassignment, quote availability, and changes in terms. The study should record departure timing and whether reassignment was planned. Its analysis must consider the possibility that difficult-to-place risks cause both higher workload and intermediary departure.
A second design might evaluate geographic access after office closure. The relevant population is customers and eligible prospective customers in a predefined service area. Observations could include product-specific alternatives, unsuccessful inquiries, remote service access, and time to obtain appropriate quotes. A stable national producer count would not invalidate evidence of local access deterioration, because it is a different aggregate boundary.
A third study might examine the market entry of new carriers through existing intermediary networks. The exposure is actual availability of a carrier-product relationship rather than announcement of entry. Outcomes could include broader quote sets, changed placements, or movements between insurer groups. The analysis should distinguish substitution of existing relationships from genuinely expanded access. A nominal appointment with no commercially usable capacity is weak evidence of enhanced competition.
The empirical welfare question thus concerns the joint evolution of mobility and outcomes. A producer transfer may facilitate a better match and increase competition where customers retain informed choice. An equivalent departure count may indicate service interruption where viable replacement capacity is absent. These conclusions are conditional mechanisms, not numerical causal estimates.
Reproducible empirical protocol
A future empirical study should specify the administrative authority and observation window, the population's stable identifiers, definitions of active and eligible, treatment of authorizations and appointments, handling of transfers and reentries, and outcome definitions. It should record data extraction versions and independently audit stock-flow reconciliations, duplicate records, missing values, unmatched transitions, and discrepancies between effective dates and report dates.
The analytic output should include cohort retention by common duration, incidence rates with explicit person-time, origin-destination tables with consistent units, relationship-level termination flows, and comparable measures of market and network concentration. It should report uncertainty appropriate to the sampling and dependence structure, including sensitivities to alternative definitions and observation gaps. Reproducibility requires preserving calculation code, data dictionaries, and lawful access arrangements or public synthetic examples, rather than disclosing confidential personal information.
The calculations in this monograph supply elementary audit fixtures for such a study. The EIOPA endpoints reconcile to a net decline of 66,209; the Hong Kong older cohort reconciles to 29,600 less 10,500 equaling 19,100; the India stock-flow identity reconciles nationally and by the reported sector rows; the hypothetical network concentration and competing-risk calculations have explicit inputs and results. These fixtures establish the internal arithmetic of the reported examples; they do not substitute for acquiring, validating, or analyzing the underlying individual histories.
Conclusion
Insurance distributor churn is a family of transitions across several linked institutional populations. The scientific object includes licensed persons, intermediary firms, insurer appointments, professional affiliations, ownership groups, and customer service relationships. Each population has a distinct stock, set of events, and period at risk. A valid estimate of turnover must identify which entity or relationship is changing, distinguish first events from recurrent events, and establish the denominator and observation process.
The published numerical evidence demonstrates the consequences of this requirement. EIOPA's registration series describes net changes in reported stocks with varying jurisdictional coverage; Hong Kong's first-entrant analysis describes cohort licence status at a common endpoint; India's agency table describes gross relationship flows and net stock growth. These results coexist without contradiction. None can be substituted for the others or transformed into a global distributor-churn rate.
Carrier mobility is an economic reallocation of access to insurance markets. The termination of an appointment may remove one placement path without ending a professional career. The transfer of a producer or firm may preserve customer contracts while changing service responsibility. Corporate concentration can change independently of the number of individuals distributing insurance. Register, relationship, firm, and policy data must therefore be linked before a distributor event can be interpreted as a change in customer protection.
An original analytical synthesis of the reviewed evidence is the distinction between three economically consequential dimensions of movement: turnover in authorization, redistribution of placement opportunity, and continuity of customer protection. These are analytical distinctions within established insurance economics and event-history methods, not a proposed named theory or validated composite model. They clarify the causal and measurement questions raised by otherwise ambiguous headline churn statistics.
Competition may benefit when mobility enables intermediaries and customers to change relationships and access better alternatives. Continuity may deteriorate where departure interrupts specialist capacity, timely servicing, or suitable coverage. Evaluating either mechanism requires measuring actual outcomes alongside observed events, and credible counterfactual designs where causal language is used. The current evidence supports these conditional interpretations; no common causal magnitude, universal optimal turnover rate, or new empirical discovery is claimed.
The research establishes a scientifically defensible foundation for future longitudinal analysis of the distribution workforce, appointments, intermediary networks, and customer books. Its limitations are explicit: the selected public sources use heterogeneous definitions; some scholarly evidence was inspected at abstract level; no proprietary linked panel was acquired; the calculations from rounded published values are approximate; and the hypothetical examples illustrate methods rather than observed outcomes. The conclusion follows within that evidentiary scope: market mobility must be measured as the evolution of specified distribution relationships and evaluated through their consequences for effective access, competition, and continuity of insurance protection.
References
International Association of Insurance Supervisors (IAIS). 2024. Insurance Core Principle 18: Intermediaries. ICP online tool. Source.
European Insurance and Occupational Pensions Authority (EIOPA). 2026. Third Report on the Application of the Insurance Distribution Directive. EIOPA-BoS-26-077, March 30. Source.
Insurance Authority (Hong Kong). 2025. Ever wondered what the turnover for licensed insurance intermediaries is? Conduct in Focus, issue 10 (Spring), 17–18. Source.
Insurance Regulatory and Development Authority of India (IRDAI). 2025. Annual Report 2024–25. Hyderabad: IRDAI. Copy hosted by the Life Insurance Council. Source.
National Association of Insurance Commissioners (NAIC). 2005. Producer Licensing Model Act. Model 218. Source.
National Insurance Producer Registry (NIPR). n.d. Reports and Alerts. Source.
European Insurance and Occupational Pensions Authority (EIOPA). n.d. Register of Insurance Intermediaries. Source.
Financial Conduct Authority (FCA). n.d. Financial Services Register. Source.
Australian Securities and Investments Commission (ASIC). n.d. AFS Representatives. Source.
Insurance Authority (Hong Kong). n.d. Registers of Insurance Intermediaries. Source.
Financial Conduct Authority (FCA). 2019. Wholesale Insurance Broker Market Study: Final Report. MS17/2.2. February. Source.
Financial Conduct Authority (FCA). 2019. Wholesale Insurance Broker Market Study: Annex 5, Demand-Side Characteristics. MS17/2.2. Source.
Competition and Consumer Protection Commission (CCPC), Ireland. 2020. Public Liability Insurance Market Study. Public study summary, December 23. Source.
Information Commissioner's Office (ICO). n.d. Your Right to Data Portability. Source.
Berger, Allen N., J. David Cummins, and Mary A. Weiss. 1997. The Coexistence of Multiple Distribution Systems for Financial Services: The Case of Property-Liability Insurance. Journal of Business 70 (4): 515–546. Source.
Cummins, J. David, and Neil A. Doherty. 2006. The Economics of Insurance Intermediaries. Journal of Risk and Insurance 73 (3): 359–396. DOI: 10.1111/j.1539-6975.2006.00180.x. Source.
Putter, Hein, Marta Fiocco, and Ronald B. Geskus. 2007. Tutorial in Biostatistics: Competing Risks and Multi-State Models. Statistics in Medicine 26 (11): 2389–2430. DOI: 10.1002/sim.2712. Source.
Callaway, Brantly, and Pedro H. C. Sant’Anna. 2021. Difference-in-Differences with Multiple Time Periods. Journal of Econometrics 225 (2): 200–230. DOI: 10.1016/j.jeconom.2020.12.001. Source.
United States Department of Justice. 2021. U.S. v. Aon plc and Willis Towers Watson plc. Antitrust Division case file. Source.
United States Department of Justice. n.d. Herfindahl-Hirschman Index. Source.
Independent Insurance Agents & Brokers of America (Big I). 2024. Agency Universe Study. Public study overview. Source.