RESEARCH DOSSIER
Behavioral Persistence and Insurance Renewal
Abstract. Insurance renewal is a repeated decision under uncertainty in which the previous contract becomes both an informational reference point and an institutional starting position. Renewal rates therefore combine deliberate continuity, stable preferences, information acquisition, search and switching costs, attention constraints, insurer pricing, changes in risk and income, experiences with claims, and contractual or regulatory rules. This monograph investigates behavioral persistence: the propensity for a policyholder's earlier insurance choice to affect a later continuation decision after accounting, as far as the evidence permits, for the contemporaneous economic and contractual attributes of available alternatives. The central analytical challenge is identification. A renewed policy can represent a well-matched product, an expensive but rationally retained relationship, an uninformed default, a response to learned insurer quality, or an inability to switch. Observing continuation by itself does not distinguish these explanations. The research integrates insurance-specific field experiments, administrative studies of health insurance and prescription drug plans, observational evidence on automobile insurance and life insurance lapse, flood-risk learning, behavioral decision research, and selected European and United Kingdom supervisory instruments. It separates persistence within an existing contract from default effects at first enrollment, new-business acquisition, and the legal treatment of nonrenewal. Particular attention is given to evidence that cuts against a universal inertia explanation: learning and improvement in some Medicare cohorts, heterogeneous responses to premium disclosures, substantive switching after third-party claims, and life-insurance lapses associated with economic and demographic shocks. The synthesis establishes that persistence has no invariant welfare sign. A continuity-preserving default can prevent protection gaps and reduce transaction costs; the same architecture may allow poorly matched coverage, weak price discipline, and exploitative renewal pricing. Conversely, an intervention that improves isolated choices can change the risk composition of insurance pools and reduce aggregate welfare under adverse selection. Effective evaluation therefore measures individual expected costs and coverage, the feasible alternative set, insurer and market responses, and the distribution of outcomes across risk and vulnerability groups. The dossier concludes with identification standards, transparent policy evaluation criteria, and a research agenda. It introduces no new named theory, doctrine, or conceptual system.
Keywords: Insurance renewal; behavioral persistence; consumer inertia; status quo bias; automatic renewal; policy retention; switching costs; insurance lapse; health insurance; Medicare Part D; life insurance; automobile insurance; flood insurance; claims experience; information frictions; adverse selection; consumer welfare; insurance regulation.
Introduction
Insurance is purchased for a period, yet its economic relevance usually extends across many periods. A household that insured a car last year faces another decision when the policy expires. An employee choosing health benefits may retain a plan despite altered premiums, networks or anticipated medical expenditure. A person with life insurance may continue premium payments, surrender a contract, reduce coverage, or lapse for reasons that include liquidity, health, family obligations, and contract-specific value. These decisions are sequential: the existing contract shapes what the decision maker knows, notices, expects, and can do next.
Behavioral persistence is used here as a descriptive and explanatory object: a systematic relationship between prior insurance affiliation and later continuation or switching, conditional on relevant current alternatives. The term does not presuppose irrationality. Renewal is an observable transaction or contractual process; retention is continuation of a relationship or product; lapse is a cessation under the relevant contractual definition; switching is movement to an alternative insurer or plan; and persistence is the dynamic pattern that may connect them. Their denominators, opportunity sets, and legal meanings differ. Treating all non-switchers as inert or every lapse as voluntary abandonment introduces classification error before causal analysis begins.
The classic experimental account of status quo bias by Samuelson and Zeckhauser (1988) established that an existing option can attract disproportionate choice. Insurance adds distinctive economic features to this behavioral setting. Future losses are uncertain; some contract characteristics become visible only at claim time; changing carriers can alter underwriting, exclusions, continuity, network access, or service expectations; and adverse selection makes the composition of the remaining pool endogenous to switching. Insurance renewal consequently involves both individual decision-making and market equilibrium.
Modern evidence demonstrates the relevance of this interaction. Handel's study of health insurance inertia (2013) uses a major employer plan change to identify persistent choices and estimates welfare effects in a setting with risk sorting. Marzilli Ericson's Medicare Part D pricing analysis (2014) links persistent enrollment to the prices firms charge established customers. Polyakova's analysis of switching costs and selection (2016) shows why lower private switching costs and insurance-pool outcomes require joint evaluation. These are rigorous, setting-specific findings, not a coefficient applicable to every country or line of insurance.
The relevant observations also include responses when consumers are actively informed. Kling and colleagues' randomized personalized plan comparison (2012) produced higher switching in the treated Medicare sample. In a different market, the Financial Conduct Authority's insurance renewal field trials (Adams et al., 2015) found a response to previous-premium disclosure among participating home insurance customers but little effect for the two motor insurers involved. The divergence is analytically important. An intervention that changes behavior in one environment reveals a mechanism there; its size and even direction cannot be transferred automatically to another environment.
This research has three principal questions. First, how much of observed contract continuation can be attributed to historical choice once risk, prices, benefits, alternative offers, regulation, and household circumstances change? Second, which combinations of habit, attention, transaction cost, informational learning, loyalty, and rational valuation explain the pattern? Third, when does the pattern protect welfare, when does it transfer surplus to firms or sustain poor matching, and what can empirical designs credibly establish across insurance lines?
The evidence strategy privileges published primary studies and official regulatory sources. Administrative datasets identify actual selections and realized expenditures; field experiments isolate specific information interventions; structural studies estimate latent frictions subject to maintained assumptions; and legal instruments define concrete duties in specific jurisdictions. Evidence across methods is synthesized by mechanism, population, available alternative, and observation horizon. No pooled effect size is estimated because the identified treatments and dependent variables differ substantially. The contribution of this dossier is a disciplined comparative synthesis and a transparent separation of supported results from interpretive deductions.
Insurance Renewal Processes
The renewal decision as a sequence
A renewal decision begins before the formal expiration date. A firm or intermediary may determine an offer, assess changes to the insured risk, communicate adjustments, and implement a continuation or opt-in process. A policyholder receives information through channels of varying reliability and compares the incumbent contract with alternatives known or discoverable at the time. A payment instruction, automatic premium collection, plan carryover rule, or failure to respond may result in continuation. The same observed outcome may therefore arise from an affirmative choice or from a process design in which remaining covered requires less action than changing cover.
The time structure matters. A customer may have selected an insurer rationally at entry and thereafter faced periodic comparison costs. A customer may also have accepted an initially attractive price expecting a comparable long-term offer, only to confront renewal increases. A previous claim can alter anticipated service quality and premiums at the same time. A financially constrained household can value protection highly yet be unable to meet the next payment. Evidence on renewal needs to distinguish these pathways and the dates at which information became available.
The UK Financial Conduct Authority defines renewal broadly for its transparency requirements, including carrying forward a policy at expiry where the same intermediary remains involved although the underwriting insurer changes. Its renewal transparency guidance illustrates why renewal, change of insurer, and broker-mediated continuity are not equivalent operational categories. Researchers should use the applicable contract and market definition, disclose it, and avoid labeling an intermediary-level renewal as carrier-level retention when the underlying risk carrier has changed.
A useful accounting distinction is between opportunities, decisions, and realized coverage. The renewal opportunity population consists of contracts eligible to continue; an observed offer requires evidence that renewal terms were actually communicated or made available; a switch requires an identified alternative; and a gap in coverage requires contract-effective dates rather than an inferred lapse label. If an insurer refuses renewal, a customer cannot be counted as having chosen to leave. If cover ends because the product is withdrawn, noncontinuation cannot automatically be attributed to consumer attention.
Persistence and the choice set
The preceding contract is a benchmark only relative to a feasible choice set. Feasibility depends on underwriting eligibility, local insurer entry, broker access, product availability, legal obligations, affordability and sometimes the existence of a comparable network or benefit structure. A nominally cheaper alternative with a newly excluded condition can be economically inferior. For a customer with uncertain future claims, the value of a stable network or an established service relationship may be real even when not directly observable in administrative price comparisons.
This concern is particularly visible in insurance markets with substantial consumer heterogeneity. Handel and Kolstad (2015) combine claims, enrollment and survey evidence to distinguish information frictions and hassle costs from risk preferences. The empirical significance of these variables warns against inferring risk aversion from observed generosity of cover without accounting for how consumers understand plan attributes. It equally warns against inferring passive inertia from the absence of switching when unmeasured service, knowledge, or availability differs.
The process can be studied at insurer, contract, coverage, or household level. An individual may keep the insurer but alter the deductible; retain coverage while purchasing through a different intermediary; shift to a different policy series; or renew one of several household contracts. Aggregate renewal rates flatten these transitions. A careful study reports the event of interest, its denominator, and the degree to which it separates passive continuation from conscious and beneficial retention.
Repeated contracting and insurer responses
Insurers observe retention behavior and may learn which customers appear less price-sensitive. A firm can invest in service quality to preserve an economically valuable relationship, or can exploit consumer search costs through renewal pricing. These responses can coexist in a market, making the relationship between tenure and price depend on regulation and competitive conditions.
Marzilli Ericson (2014) documents a distinctive form of differential pricing in Medicare Part D. His published enrollment and price study reports that older plans carried premiums approximately 10 percent higher than comparable new plans in the observed 2006–2010 period. The paper also identifies persistent effects of initial defaults. The estimate applies to that market, data period, and comparison specification; it should not be recast as a general percentage increase in all insurance renewals.
The fundamental economic implication is that renewal cannot be evaluated solely at the instant of acceptance. The possibility of later retention affects the acquisition price, contractual presentation, firm incentives, and competition over time. A policy may appear cheap at entry and costly in subsequent periods, while another may offer comparatively stable prices and superior service. Evaluations restricted to first-year premiums conceal this intertemporal dimension.
Habit and Status Quo
Habitual retention and the informational reference point
Habit can reduce the cognitive and administrative demands of repeated transactions. If a customer found a satisfactory product, remembering that decision may be efficient. Over time, however, an established routine can become disconnected from changed conditions. The question for research is whether the relationship between a prior choice and later behavior remains after the researcher has measured the value of the incumbent and the realistic costs of changing it.
Samuelson and Zeckhauser's status quo experiments and evidence (1988) demonstrate that prior allocation can affect preferences expressed in choice. Their analysis also notes consequential decisions in employee benefit programs. Within insurance, reference dependence may appear when losing an existing deductible structure, provider network, or perceived claim reliability feels more salient than the equivalent gain from a new insurer. Yet such an interpretation requires evidence about preferences and perceived tradeoffs. A high retention rate is compatible with both a status quo effect and justified satisfaction.
Losses and gains at renewal can be multidimensional. A premium increase is a monetary loss relative to the prior payment. Switching may offer savings but create uncertainty about claim treatment. A plan with a lower premium may have higher contingent out-of-pocket losses. A switch may also demand effort now for benefits realized probabilistically later. The resulting asymmetry is a plausible behavioral explanation, but its magnitude must be established empirically, ideally by changing the relevant reference or default independently of objective contract value.
Automatic continuation and default mechanics
Automatic renewal changes the action required for each outcome. When continuation occurs absent a response, the task of switching involves noticing the renewal, assessing offers, meeting deadlines and completing a transaction. The default is a material part of the decision environment, with a simultaneously valuable function: it can prevent an unintended uninsured interval. A policy intervention that mandates fresh affirmative action may reduce inertia and increase involuntary loss of protection among inattentive or time-constrained households.
The default at initial enrollment and the default at annual renewal have different causal roles. Initial allocation may determine a lasting incumbent without the consumer ever choosing it, whereas renewal defaults maintain a previously accepted or assigned policy. The initial allocation literature should therefore inform the renewal analysis without being substituted for direct longitudinal evidence of renewal behavior. Marzilli Ericson's initial-default discontinuity evidence is especially useful because it traces persistence into later plan choice rather than presuming that assignment affects only the entry period.
A default can also be conditional on eligibility or product continuation. Regulatory plan transfers and insurer withdrawal may interrupt automatic carryover, producing reassignment or forced rechoice that looks like unusually high consumer engagement. Researchers should exclude, stratify, or explicitly model such events before comparing persistence across years.
Attention as an endogenous resource
Attention is scarce and competes with other demands. Insurance characteristics are complex, and the expected financial reward from comparison can be difficult to compute. The intervention studied by Kling et al. (2012) is particularly informative because the comparative information was already freely available; recipients of a personalized letter were nevertheless more likely to change plans. Their randomized field experiment reports switching by 28 percent of treated Medicare Part D beneficiaries versus 17 percent in the comparison group, together with an approximately $100 decline in predicted annual consumer cost among letter recipients. The study distinguishes supplying usable information from merely allowing access to it.
These figures identify the effect of one intervention for a particular study population. They do not prove that the full 11-percentage-point difference was caused exclusively by habit. A personalized comparison can simultaneously reduce calculation cost, direct attention to a specific alternative, alter beliefs about affordability and increase the salience of foregone savings. The most defensible conclusion is that a substantial choice friction existed in that experiment and could be reduced through direct communication.
A stronger empirical design measures whether the improvement persists beyond one renewal. A reminder can produce a one-time response without changing long-run attention; it can also initiate learning that changes future choices. Studies with only an immediate switching outcome leave this distinction open.
Persistence with deliberate preferences
Consumer inertia is frequently discussed as a departure from efficient choice. Yet an insurer's reputation, claim service, trusted intermediary, or preferred medical network can produce rational loyalty. Those benefits may be hard to price and may not be fully visible before a loss occurs. An observed willingness to pay for continuity can therefore reflect private information about service quality as much as limited attention.
The distinction demands more than a survey item asking whether the customer felt satisfied. Customers may rationalize past choices, and satisfied consumers may differ systematically in unmeasured risk and quality. Better evidence combines a history of insurer interactions, comparable alternative offers, premium differences, actual claims, and exogenous information changes. A credible assessment asks which account predicts behavior when one element changes while other relevant features are held reasonably constant.
Contract Retention and Lapse
Operational and statistical measurement
Retention measures how many eligible contracts continue across a specified period. It is commonly expressed as a ratio, but that simple ratio conceals important design choices: whether the base is written policies or exposed risks; whether midterm cancellations count; whether a broker transfer is an insurer change; and whether nonpayment, insurer-initiated termination, death, product closure or migration to a successor plan belongs in the numerator or denominator.
A renewal-event rate should be defined over policies genuinely at risk of voluntary renewal or departure. For multiyear life contracts, a continuation observation may be the timely payment of a periodic premium rather than replacement of an expiring annual contract. A renewal rate derived from one-year motor insurance is therefore not equivalent to a life policy persistency rate over its duration. The researcher should preserve each contract's time scale.
Duration analysis offers a more informative alternative to a single rate. Let the observed event be voluntary switching or lapse, and let contract tenure measure exposure to the risk of the event. A hazard describes the conditional probability of that event during a period among those still eligible immediately beforehand. Competing events—death, insurer termination, product closure, and changes in legal eligibility—should be separated because they can remove the opportunity to continue. Repeated contract episodes within a household also require dependence-aware standard errors or an appropriate multilevel design.
The descriptive fact that retention increases with tenure can result from selection: low-persistence customers leave early, leaving a progressively more loyal cohort. Without measuring individual heterogeneity, this process can imitate true duration dependence. Entry-cohort comparisons, policyholder fixed effects where feasible, and sufficiently rich risk and offer information are essential to interpreting the gradient.
Life-insurance lapsation and economic shocks
Life insurance highlights the limits of a generic inattention narrative. A policyholder's future coverage need changes with dependent family members, assets and liabilities. Some life contracts contain surrender value or guarantees that make switching economically costly; others are term contracts with different renewal and repricing provisions. A missed premium can reflect a liquidity shock even when the household wants continuing protection.
Fang and Kung's dynamic analysis of life insurance lapse (2021) uses Health and Retirement Study information and a structural discrete-choice approach to examine income, health, bequest-motive and idiosyncratic shocks. Their estimated counterfactuals suggest that a substantial part of lapsation among younger policyholders is attributable to shocks not directly correlated with the modeled income, health, or bequest motives, whereas the relative importance of these underlying motives becomes greater at older ages. This is evidence that lapse cannot be reduced to a common decision defect.
Their findings are model-dependent and sample-specific. The interpretation of residual choice shocks is especially sensitive to functional assumptions and unobserved household circumstances. It would be incorrect to infer that the study identifies the exact psychological content of every residual shock. Nevertheless, the analysis strongly supports treating household financial states and evolving insurance needs as necessary parts of persistence research.
A welfare assessment of life-policy continuation also differs from analysis of annual price shopping. The value of remaining in an underwritten policy may depend on deteriorated health, new underwriting barriers, accrued guarantees and surrender provisions. A household retaining an apparently expensive contract could be protecting a valuable option. Conversely, continued premiums on a contract whose purpose has ended may burden consumption. The relevant comparison is the full expected value of available strategies, not merely a premium at one point in time.
Health insurance and prescription drug plans
Health insurance provides unusually rich linked datasets covering premiums, plan characteristics, claims and medical expenditure, but its institutional structure varies. Employer-sponsored plans, exchanges, Medicare drug coverage and other public-private systems expose consumers to different eligibility rules and continuation mechanisms. These differences matter for generalization.
Abaluck and Gruber's Medicare plan choice analysis (2011) finds that sampled beneficiaries weighted premiums more heavily than their expected out-of-pocket costs and appeared to undervalue aspects of risk reduction. Their partial-equilibrium welfare comparison illustrates potential costs of poorly matched plans; the underlying welfare figure is conditional on their model and cannot be simply read as measured cash savings from switching. In their later study of evolving plan choices (2016), foregone savings rose with changes in plan characteristics and they estimated little learning at either the individual or cohort level in the studied setting.
There is meaningful counterevidence. Ketcham and colleagues' Medicare Part D learning study (2012) examines 2006–2007 panels and finds reduced overspending, with average improvement of $298 in their sample and gains for 81 percent of observed enrollees. Much of the improvement occurred among those who overspent most and those who switched. This does not negate the Abaluck and Gruber findings: the studies differ in samples, years, welfare constructions, and the treatment of plan changes. Together, they prohibit a categorical claim that consumers neither learn nor improve after enrollment.
Retention as an organizational performance metric
Insurers and intermediaries often treat persistency as evidence of successful distribution, sound pricing or customer satisfaction. A retention metric can be operationally useful but becomes misleading when treated as a direct welfare measure. A rising retention rate may reflect improved service, reduced competitive alternatives, tighter switching frictions, or strategic repricing. A falling rate may reflect improved consumer comparison, deteriorating trust, unaffordable increases, or a beneficial reduction in unnecessary coverage.
Evaluators should therefore report a vector of outcomes: retention, coverage adequacy, premium trajectory, realized claims service, switching effort, consumer understanding, complaints and financial vulnerability. These measures capture different aspects of welfare and should be interpreted jointly, rather than mechanically combined into a single unsupported score.
Learning after Claim Experience
Claims as information about insurer quality
Insurance is a contingent service. Policyholders know the premium and some contractual terms before buying, but their practical experience of the insurer may be limited until a covered event occurs. A claim can reveal communication quality, documentation burden, settlement speed, dispute practices and the realized interpretation of coverage. It can also reveal new facts about the policyholder's own risk and change the next premium. Learning about insurer quality and learning about risk must therefore be separated in empirical analysis.
A favorable claims experience may rationally increase retention by resolving uncertainty in the incumbent's favor. A poorly handled claim may induce a switch even if the premium is unchanged. But claim occurrence is not randomly assigned. Loss probability, behavior, coverage, location and vulnerability influence whether a claim arises; claims can affect renewal offers; and customer perceptions of responsibility can change the interpretation of the same event. A naive before-and-after comparison confounds several mechanisms.
Jõeveer and Kepp's car insurance switching and claims study (2023) analyzes renewal choices with information on policies and claims. They report a significant role for monetary gains and a difference between consumers experiencing self-induced accidents and other claim categories: at-fault claimants were less inclined to switch than customers whose claims involved other causes, conditional on the study controls. The observation complicates any simple rule that a claim necessarily induces switching. Causal interpretation remains limited by claim selection and by the observational identification used.
The more recent CEPR discussion paper by Perrone and Valiati (2025) emphasizes the role of observable alternative offers in automobile insurance. Its reported findings indicate that omitting consideration sets and the incumbent's renewal premium overstates inertia and understates price responsiveness; claims history also predicts switching. As a discussion paper accessed through its abstract, it provides a relevant methodological indication rather than an independently replicated or definitive market-wide estimate.
Risk salience and disaster experience
A catastrophe can raise both objective attention to risk and subjective beliefs about future losses. The effect may persist for a period and then attenuate. Flood insurance is a particularly informative domain because the insured event is uncommon, widely publicized, and often spatially concentrated.
Gallagher's flood insurance learning study (2014), using a nationwide U.S. panel of major regional floods and flood insurance policies, finds a spike in insurance take-up after a flood and subsequent decline toward baseline. Residents of unflooded communities within the same television media market also increased participation, at a lower rate. The pattern is consistent with belief updating with forgetting or incomplete recall of past floods. These are findings about take-up and its subsequent dynamics; they should not be described as a clean experiment measuring renewal of the same individual policy.
The conceptual connection to renewal is nevertheless important. An existing policyholder may reassess coverage after an event, and a previously uninsured household may purchase temporary cover and later discontinue it. Disaster-induced salience can generate both acquisition and later lapse, with implications for the stability of risk pools. To identify renewal-specific persistence, researchers require policy-level panels and accurate exposure, rate and mortgage-mandate information; regional take-up statistics alone cannot establish voluntary continuation behavior.
Memory, experience, and contract interpretation
The timing of claim information matters. A claim submitted immediately before renewal may be unresolved, leaving uncertainty about outcome quality. A favorable settlement received months later may influence the following renewal instead. Customer service and intermediary conduct can modify the perceived signal. A study that assigns all claims to the loss date and ignores subsequent settlement may mismeasure the mechanism and its lag.
The best empirical datasets separate loss event, claim submission, investigation, settlement, complaint and renewal offer dates. They observe whether the consumer saw a price change and whether other insurers were willing to offer comparable cover. The model should accommodate learning that is favorable, unfavorable or absent, while acknowledging that the customer's preference for familiar claims handling can be economically legitimate.
Consumer Frictions and Institutional Design
Information frictions, hassle costs, and switching costs
Insurance decision costs have several sources. Search cost is the effort needed to identify alternatives. Comparison cost is the effort required to translate different deductibles, coverage limits, networks and exclusions into expected outcomes. Switching cost includes application, underwriting, documentation, timing and potential disruption. Information friction arises when consumers misunderstand objective product features or do not receive information in usable form. These mechanisms may occur together yet respond differently to interventions.
Handel and Kolstad's survey-linked insurance choice research (2015) shows that explicitly measured consumer information and hassle costs can materially affect estimated risk preferences. This matters because a structural model that absorbs every unexplained preference for the incumbent into a fixed switching-cost parameter risks misclassification. A parameter can fit observed persistence while combining ignorance, trust, risk heterogeneity, unavailable alternatives, and procedure burden.
Comparison interventions offer partial identification. The Kling et al. personalized letters (2012) increased switching, identifying a response to delivered information. Yet a letter bundles provision of information, salience and reduced effort. A factorial design would compare equivalent content with different modes of delivery, timing or required actions to separate these components. Such a design would yield a more precise causal account than a single treated-versus-control contrast.
Choice architecture also affects how consumers weight attributes. Ericson and Starc (2016), studying a change on the Massachusetts Health Insurance Exchange, find that product standardization altered choices toward more generous options and different brands, without making consumers more price-sensitive in the observed comparison. The result supports the view that presentation and product design can affect decisions through attribute valuation and availability. It does not establish that standardization always intensifies price competition or improves welfare.
Renewal communications and comparative disclosure
The most directly relevant randomized evidence for annual general insurance renewal comes from the FCA's field trials with one UK home insurer and two motor insurers. The renewal communication experiments (Adams et al., 2015) tested including last year's premium, simplified notices, reminders and other communications. Displaying the previous premium caused 11–18 percent more home insurance consumers to switch or negotiate, especially among customers offered a larger premium increase. For the participating motor insurers, the researchers found little evidence of renewal price increases and no effect of the previous-premium treatment. Other tested communication changes produced little or no detectable impact.
The distinction between an 11–18 percent increase in the number acting and an 11–18 percentage-point change in the action rate is material. The source reports the former. It also combines switching with negotiation, so the estimate cannot be represented as a pure switching effect. A consumer negotiating a lower incumbent premium remains insured by the same company but has acted on the information. This is exactly why retention alone undercounts consumer engagement.
A notification's effectiveness depends on whether the customer can use the information. Showing a prior premium helps identify a change in the price of a similar product, but may be insufficient if coverage, excesses, discounts or risk characteristics have changed. Presenting a comparison without explaining material contract changes can create a false appearance of price equivalence. An effective disclosure makes the current decision legible while preserving the differences relevant to expected coverage and claims.
The FCA's transparency requirements for insurance renewal became relevant to consumer notices in 2017. Their stated objectives include better comparison and engagement, and treatment of consumers wishing to switch or cancel. A later supervisory review of renewals reported shortcomings in implementation. Regulatory adoption of disclosure obligations and evidence of increased consumer attention are distinct from proof that every communication design produces a lasting welfare improvement.
Price walking and firm incentives
Price walking occurs when a provider progressively increases prices for established customers in ways not fully explained by relevant cost changes, while new customers receive more favorable offers. In insurance, the ability to implement such strategies depends on search frictions, price sensitivity, market conduct duties and the competitive environment. Price walking is a pricing practice, whereas retention is a consumer outcome; a market can have strong retention without price walking, and consumers can switch actively even where price discrimination occurs.
The FCA's general insurance pricing remedies (2021) addressed price walking in UK home and motor insurance. Under the current ICOBS 6B.2 renewal-price rule, an insurer within the rule's scope must not set a renewal price higher than the equivalent new-business price at the relevant point. The core price provisions began in January 2022, subject to stated implementation transitions. The rule concerns equivalent new-business prices, not a universal obligation to keep a customer's premium unchanged from the preceding year. Risk changes, market-wide rates and product differences remain relevant within the instrument's terms.
The relevance for research is straightforward: regulatory price constraints alter the opportunity for firms to monetize inertial demand. Historical estimates of renewal price increases from before the remedy should not be interpreted as current evidence of lawful UK pricing practice. Comparisons across time also need to account for altered risk classification, insurer exit, inflation and the simultaneous operation of other conduct rules.
Contract continuity and consumer protection
Insurance continuity can have high social and private value. Renewal processes should be designed with attention to gaps in coverage, legal mandates, financially vulnerable consumers, digital exclusion and the effort required to leave an unsuitable policy. A friction-reducing mechanism is beneficial only if it improves the feasible choices and outcomes of consumers in the actual system.
The EU Insurance Distribution Directive (Directive (EU) 2016/97) establishes distribution conduct, information and product-governance duties for its applying jurisdictions. Article 25 concerns product oversight and governance, including review of whether products remain consistent with the identified target market. These requirements provide a regulatory framework relevant to the quality of products offered and retained; they do not supply a single harmonized EU behavioral renewal effect. National transposition, product category and the exact legal duty require jurisdiction-specific treatment.
EIOPA's 2025 insurance consumer trends heatmap identifies broader market concerns including claims management, digital distribution, value for money, and protection gaps. Those supervisory themes justify research attention to renewal outcomes and vulnerable groups; the heatmap is not itself a randomized estimate of renewal inertia.
A well-designed renewal regime enables reliable continuation, understandable comparison, realistic exit, and scrutiny of long-tenure outcomes. These goals can reinforce each other when information arrives early enough and alternatives are accessible. They can also conflict when a forced rechoice introduces a coverage gap or a comparison rule changes who remains in an insurance pool.
Methods, Identification, and Measurement
The central identification problem
A correlation between past and current insurance affiliation is not a causal effect of past affiliation. High-quality customers may remain because they were well matched at entry. High-risk policyholders may face less favorable alternative offers. An insurer may reward long-tenure customers with service or discounts. Product availability can shrink, and consumer health or income can change. All these variables create persistence or switching patterns even without behavioral inertia.
The empirical object must therefore be defined precisely. A researcher may seek the effect of automatic continuation relative to mandatory active choice; of a reminder relative to no reminder; of receiving a personalized price comparison; or of a claim experience on the probability of renewal. Each is a distinct treatment. Claims about a single latent phenomenon require a model linking such interventions to the relevant mechanism, and that model must be justified rather than implied by common vocabulary.
Randomization is persuasive for an intervention's direct effect when assignment, compliance, follow-up and attrition are appropriately managed. It is often less informative about the underlying mechanism when the intervention changes several things at once. The Kling et al. field experiment provides a credible information-delivery effect, while the FCA renewal experiments reveal heterogeneous behavioral responses to a specific premium disclosure. Neither permits substitution of one market's treatment coefficient into another market without an external-validity argument.
Natural experiments can exploit unanticipated plan changes, staggered communications or institutional discontinuities. Their credibility depends on identifying assumptions: affected and comparison groups must be appropriately comparable in the absence of treatment; other contemporaneous changes must be considered; and the measured outcome must represent the intended economic decision. The large employer transition used by Handel (2013) offers an example of a design that goes beyond raw retention counts.
Choice models, latent frictions, and structural interpretation
Structural insurance-choice models estimate parameters for risk preferences, plan valuations and various switching frictions. Their strength is the capacity to simulate counterfactual decisions and equilibrium responses that are difficult to observe directly. Their weakness is reliance on assumed utility functions, information sets, error distributions and insurer behavior. Two models with different behavioral foundations may fit the same observed transitions and assign different welfare significance to them.
Handel and Kolstad (2015) explicitly show why information and hassle measures matter for the estimation of risk preferences. Polyakova (2016) adds supply and risk-sorting dynamics to the structural analysis of switching. These studies are complementary but answer different questions. One cannot treat their parameters as direct psychological measurements independent of maintained assumptions.
A disciplined study should publish the choice set observed by the analyst, the choice set plausibly considered by the policyholder, and the difference between them. Missing outside offers are particularly dangerous because a model may mistake inability to obtain a feasible new policy for a preference to remain. The Perrone and Valiati discussion paper specifically emphasizes this concern in auto insurance. Researchers should use the paper as a warning about specification rather than as proof that any particular level of inertia is an artifact.
Hazards, panel structure, and selection
Duration and panel designs track the same policies through multiple opportunities. At minimum, the data should distinguish entry date, offer date, quoted price, renewal choice, effective coverage, risk change, service events and censoring. A customer who becomes ineligible and is removed from the denominator should not be recoded as a voluntary switch. Clustering by household, broker or insurer may be necessary because decisions share common information and pricing practices.
Selective attrition changes the composition of a renewal sample. A panel of people remaining after five years excludes earlier switchers and lapsers, so the observed loyalty of incumbents is conditional on survival. Likewise, experiments whose outcome is recorded only for respondents may lose the very consumers whose inattention is the scientific question. Attrition, survey response and data coverage should be reported rather than assumed harmless.
A sound outcome framework distinguishes the intention-to-treat effect of a communication from the effect among those who actually read it; the two parameters require different assumptions. It also distinguishes moving to a cheaper policy from improving expected welfare. The cheapest premium may purchase less coverage, and price comparisons without claims distributions can misclassify the result.
A reproducible research dataset
The minimum useful contract-level longitudinal file would include a pseudonymous policyholder identifier, contract and insurer identifiers, jurisdiction, product line, original enrollment pathway, renewal date and method, incumbent and alternative quotes when available, coverage terms, assessed risk, payment modality, consumer transaction, claims events and resolution, cancellation reason, and coverage-end date. For personal insurance it should include appropriately protected financial and demographic measures, subject to law, consent and research ethics.
The dataset should record whether alternatives were actually shown to the consumer, whether switching was legally and practically possible, and whether automatic continuation occurred. Missing competitor quotes must be coded as missing, not as evidence of consumer indifference. Consent, data minimization, reproducible aggregation and disclosure control are essential when claims, health and financial records are involved.
For cross-country comparison, legal meanings of renewal and lapse should be harmonized at the analysis level while preserving country-specific contract definitions. Direct pooling without those definitions creates false comparability. A defensible international study first establishes within-market identification and then tests whether differences in prices, default rules, product design and consumer protection can explain variation in responses.
Sensitivity and falsification
Every empirical interpretation should be challenged with alternative explanations. If the coefficient interpreted as inertia declines sharply after observing competitor quotes, the original model was partly measuring restricted choice information. If a renewal reminder has no effect where premiums do not rise, the treatment's mechanism may be price salience rather than general activation. If switching rises after a third-party claim but not after a self-induced claim, the difference may implicate learning about service, blame attribution, or insurer pricing; these hypotheses require separate tests.
Useful robustness checks include price-equivalent comparison sets, alternative definitions of switching, policyholder tenure interactions, insurer fixed effects, pretrend inspection, balance tests for randomized assignment, different follow-up windows, and sensitivity to unobserved risk and service quality. Where an instrument or quasi-experiment is used, the exclusion restriction deserves explicit scrutiny. Negative or null results should be preserved because they delimit the circumstances under which an intervention is effective.
Policy Outcomes and Comparative Evidence
Consumer costs and plan matching
The strongest case for active comparison is observed when consumers remain in materially inferior contracts despite accessible superior alternatives. Abaluck and Gruber's Medicare Part D choice findings show how premiums, expected spending and risk exposure may be weighted inconsistently. The Kling et al. comparison experiment demonstrates a feasible communication intervention that changed behavior and predicted costs. But plan cost is only one part of insurance quality. Treatment evaluations should report coverage levels, network access, expected risk-bearing and realized outcomes as available.
Bhargava, Loewenstein and Sydnor's study of dominated health plans (2017) provides additional evidence that complicated insurance menus can produce financially inconsistent choices even without measuring a renewal decision directly. In their U.S. employer setting, employees selected from plans differing largely in premiums and cost-sharing, allowing investigators to detect dominated options. The relevance is to decision quality and comparative complexity, not a direct causal effect of automatic renewal.
Market equilibrium and adverse selection
Interventions can change the characteristics of consumers who remain insured in particular plans. If lower-risk consumers are more able to recognize and act on a savings opportunity, a policy that reduces inertia may disproportionately move them out of comprehensive plans. Premiums and plan viability can then adjust. Individual gains calculated at unchanged prices may overstate welfare after equilibrium responses.
Handel's adverse selection and inertia results (2013) explicitly identify such a tradeoff. In the modeled employer environment, a policy reducing inertia improved individual choices at given options while increasing adverse selection enough to reduce aggregate welfare in the relevant counterfactual. The result establishes that welfare cannot be assigned a positive sign to all switching-promoting nudges. It does not imply that preserving inertia is universally beneficial.
Polyakova's Medicare Part D market analysis (2016) provides a different quantified setting: the study reports potential annual consumer-surplus gains of approximately $400–$600 per capita from active decision-making under its estimated policy environment. The two analyses do not contradict one another merely because one emphasizes a possible aggregate welfare loss and the other a gain. Their environments, models, risk sorting and counterfactual policies differ.
The legitimate synthesis is conditional. An intervention's welfare effect depends on who changes coverage, how differences in risk feed into premiums, what quality and benefit changes occur, and whether public subsidies or risk adjustment mitigate sorting. Policy analysis should report both partial-equilibrium and plausible general-equilibrium outcomes whenever the market structure permits.
Distributional consequences
Inertia and switching burdens are unevenly distributed. Some consumers possess more time, financial reserves, digital literacy or access to advice. Others face complex health risks, mobility constraints, language barriers or difficult household circumstances. The resulting distribution of costs may differ from aggregate average effects even when a market-wide reform appears favorable.
Adams et al. (2015) show why intervention assessment requires subgroup and institutional detail: in the home and motor renewal trials, displaying last year's premium had different effects by line and insurer. The findings do not justify treating age alone as a cause of inertia, although age and vulnerability can be relevant for distributional monitoring. Reporting actual differences in comprehension, offers, access and outcomes is preferable to imputing behavioral traits from demographic categories.
The FCA Consumer Panel's 2024 renewal research also illustrates continued concern over price increases and the burden of negotiating home and motor renewals. It is a consumer-experience source, not a controlled causal test. Survey results should not be confused with randomized treatment effects or transported uncritically to other jurisdictions.
The value of continuity
Continuity is itself an outcome. Failure to renew may produce uncompensated loss, contractual penalties, inability to access health care or an obligation to replace legally required coverage. In life insurance, termination may be difficult to reverse after health changes. A consumer who avoids an interruption may gain welfare even if the incumbent premium is not the market minimum.
The relevant policy choice is often between improving the existing default and abolishing it. Earlier, clearer notices, comparable contract summaries, price-change explanations, easier cancellation, independent advice and protection against exploitative renewal pricing can reduce harmful inertia while preserving continuity. Which combination performs best is an empirical question. Each component should be tested against both engagement and coverage outcomes, not evaluated solely through an increase in switching.
International comparability and limits
The available well-identified behavioral literature is concentrated in U.S. health and prescription drug markets and selected UK general insurance settings. Its design is strong in those institutional environments, but the international insurance system includes compulsory and voluntary products, life and nonlife contracts, public schemes, microinsurance and different legal regimes. Cross-market inference requires explicit common mechanisms and careful attention to transferability.
The EU Insurance Distribution Directive supplies a comparative regulatory example through product governance and conduct duties; the UK ICOBS renewal pricing provisions show a distinct targeted remedy. These instruments do not establish a global legal rule. A synthesis must disclose where a jurisdiction's law begins and ends, and avoid converting a local policy design into an international prescription.
Empirical gaps are significant for renewals in emerging insurance markets, broker-led commercial and household portfolios, inclusive insurance, digital micro-duration products and contexts where switching options are poorly observed. These gaps call for direct research, especially where automatic continuation may serve a critical protection function and where financial constraints govern nonrenewal.
Integrated Scientific Assessment
What the evidence establishes
The literature supports four stable conclusions at the level of scientific assessment. Previous enrollment and contract affiliation can materially influence later choice, as shown in the Medicare Part D pricing and default evidence and the health insurance inertia study. Comparative information delivered in usable form can raise engagement, demonstrated experimentally by personalized Medicare plan letters and, under different conditions, by the UK home insurance renewal trials. Explanations and treatment effects vary materially across markets, consumer populations and choice environments. Finally, the direction of welfare effects depends on both individual contract value and wider market responses, especially adverse selection and pricing.
These conclusions are narrower and more useful than declaring insurance consumers globally irrational or structurally loyal. The evidence identifies particular mechanisms and interventions, while credible explanations of observed persistence often require several mechanisms at once. The specific meaning of inertia is empirical: a parameter in one model should be interpreted in light of its identifying information, not automatically as a psychological trait.
What remains uncertain
Several questions remain unresolved by the reviewed studies. How often does insurer quality learned through claims dominate switching hassle in ordinary nonlife portfolios? To what extent are long-tenure renewal premiums explained by cost and risk changes rather than exploitation of low search intensity? How do consumer-level behavioral interventions affect lapse and protection gaps beyond the immediate renewal window? Which vulnerable consumers benefit from defaults, and which experience the largest hidden costs? How stable are these effects outside the U.S. and UK environments with relatively rich datasets?
The evidence is also limited in how it observes actual attention. A consumer who does not switch may have compared plans and selected the incumbent, while a consumer who switches may still choose a dominated alternative. Many administrative datasets reveal the transaction rather than the deliberation. Surveys add information but introduce recall and response bias. The research record supports disciplined inference, not omniscient reconstruction of motives.
An integrated evaluation practice
A renewal intervention should be evaluated in stages. The researcher first establishes the contract, legal and market setting, including exactly which consumers can renew or switch. The study then defines a causal treatment, measures prices and benefits for the incumbent and feasible alternatives, and observes changes in behavior over time. It distinguishes the effect on decision engagement from the effect on insurance coverage and welfare. Finally, it tests heterogeneity and, where relevant, insurer pricing and risk-pool responses.
This sequence is an application of established empirical research principles to the insurance renewal problem, not a proposed new named model. It improves interpretation by refusing to treat every retained contract as failure or every switch as success. It also disciplines regulatory claims: better disclosure is valuable when it improves informed and feasible decisions, while fairness at renewal may require scrutiny of offer quality and institutional incentives beyond disclosure alone.
Research priorities
Future work should build linked multi-year panels with accurate alternative offers and contract-level claim histories; use randomized or quasi-experimental variation in information and renewal procedures; estimate both short-term engagement and multiyear coverage outcomes; and test whether measured friction differs by insurer quality, age of policy, income and legally constrained switching options. Studies should document the competing explanations and pre-specify the behavioral and welfare estimands before the results are observed.
For actuarial and policy analysis, the practical implication is that retention forecasts should include risk and price changes without interpreting residual persistence as pure behavioral bias. For consumer research, a documented ability to compare and leave is important alongside observed satisfaction. For insurance governance, the objective is durable appropriate protection under fair renewal conditions. All three perspectives benefit from better data on what consumers could actually choose.
Conclusion
Behavioral persistence in insurance renewal is the dynamic effect of prior insurance affiliation on subsequent continuation and switching within a changing landscape of contracts, prices, risks, claims and institutions. It operates through habit, defaults, attention, search and switching costs, learning about insurer quality, changing household needs and strategic insurer responses. These mechanisms create observationally similar retention but have different welfare implications.
The reviewed evidence establishes persistence in several major insurance settings and demonstrates that usable comparative information can sometimes alter behavior substantially. It also establishes important limits: information interventions are heterogeneous; customers may learn and improve over time; life insurance lapse reflects evolving circumstances and shocks; claims can influence switching through several channels; and competition with adverse selection can cause an individually beneficial change in choices to have an adverse aggregate effect.
Consequently, a renewal rate is an operational statistic whose meaning depends on the contracts and alternatives behind it. The scientifically relevant assessment connects observed decisions to feasible options, benefit adequacy, price trajectories, claim experience, consumers' understanding, and the market response. Insurance policy should preserve valuable continuity while making comparison and exit practicable and preventing unjustified disadvantages for established customers.
The monograph's analytical contribution is its cross-line synthesis, distinction between observable retention and underlying causal mechanisms, and explicit specification of the evidence required for welfare interpretation. It makes no claim of an original behavioral theory or a universal empirical coefficient. The scientific conclusion is that renewal persistence is a context-dependent outcome, and that reliable judgments about its benefits and costs require institutionally grounded causal identification.
References
Abaluck, Jason, and Jonathan Gruber. 2011. “Choice Inconsistencies among the Elderly: Evidence from Plan Choice in the Medicare Part D Program.” American Economic Review 101 (4): 1180–1210. https://swlb2.aeaweb.org/articles?id=10.1257%2Faer.101.4.1180
Abaluck, Jason, and Jonathan Gruber. 2016. “Evolving Choice Inconsistencies in Choice of Prescription Drug Insurance.” American Economic Review 106 (8): 2145–2184. https://benny.aeaweb.org/articles?id=10.1257%2Faer.20130778
Adams, Paul, Robert Baker, Stefan Hunt, Darragh Kelly, and Alessandro Nava. 2015. “Encouraging Consumers to Act at Renewal: Evidence from Field Trials in the Home and Motor Insurance Markets.” Financial Conduct Authority Occasional Paper No. 12. https://www.fca.org.uk/publications/occasional-papers/occasional-paper-no-12-encouraging-consumers-act-renewal-evidence
Bhargava, Saurabh, George Loewenstein, and Justin Sydnor. 2017. “Choose to Lose: Health Plan Choices from a Menu with Dominated Option.” Quarterly Journal of Economics 132 (3): 1319–1372. https://doi.org/10.1093/QJE/QJX011
Ericson, Keith M. Marzilli. 2014. “Consumer Inertia and Firm Pricing in the Medicare Part D Prescription Drug Insurance Exchange.” American Economic Journal: Economic Policy 6 (1): 38–64. https://www.aeaweb.org/articles?id=10.1257%2Fpol.6.1.38
Ericson, Keith M. Marzilli, and Amanda Starc. 2016. “How Product Standardization Affects Choice: Evidence from the Massachusetts Health Insurance Exchange.” Journal of Health Economics 50: 71–85. https://pubmed.ncbi.nlm.nih.gov/27693893/
European Parliament and Council of the European Union. 2016. “Directive (EU) 2016/97 of 20 January 2016 on Insurance Distribution,” consolidated text as of January 9, 2024. Official Journal of the European Union. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A02016L0097-20240109
European Insurance and Occupational Pensions Authority (EIOPA). 2025. “Consumer Trends in Insurance and Pensions 2025: Heatmap with Key Findings.” December 17. https://www.eiopa.europa.eu/publications/consumer-trends-insurance-and-pensions-2025-heatmap-key-findings_en
Fang, Hanming, and Edward Kung. 2021. “Why Do Life Insurance Policyholders Lapse? The Roles of Income, Health, and Bequest Motive Shocks.” Journal of Risk and Insurance 88 (4): 937–970. https://onlinelibrary.wiley.com/doi/10.1111/jori.12332
Financial Conduct Authority (FCA). 2017. “Transparency in Insurance Renewals.” Regulatory guidance and implementation information. https://www.fca.org.uk/firms/transparency-insurance-renewals
Financial Conduct Authority (FCA). 2017. “Firms Falling Short of Renewal Expectations.” Multi-firm supervisory review; historical review with subsequent website update. https://www.fca.org.uk/publications/multi-firm-reviews/firms-falling-short-renewal-expectations
Financial Conduct Authority (FCA). 2021. “General Insurance Pricing Practices Market Study: Feedback to CP20/19 and Final Rules.” Policy Statement PS21/5. May. https://www.fca.org.uk/publication/policy/ps21-5.pdf
Financial Conduct Authority (FCA). 2022–2026. “Insurance: Conduct of Business Sourcebook, ICOBS 6B.2: Setting Renewal Prices.” Official FCA Handbook, timeline version reviewed November 2026. https://handbook.fca.org.uk/handbook/icobs6b/icobs6bs2?timeline=true
Financial Services Consumer Panel. 2024. “Consumer Panel Research on Insurance Renewal Experiences for Consumers.” FCA-hosted press release, August 27. https://www.fca.org.uk/panels/consumer-panel/news/press-release-consumer-panel-research-insurance-renewal-experiences-consumers
Gallagher, Justin. 2014. “Learning about an Infrequent Event: Evidence from Flood Insurance Take-Up in the United States.” American Economic Journal: Applied Economics 6 (3): 206–233. https://benny.aeaweb.org/articles?id=10.1257%2Fapp.6.3.206
Handel, Benjamin R. 2013. “Adverse Selection and Inertia in Health Insurance Markets: When Nudging Hurts.” American Economic Review 103 (7): 2643–2682. https://pubs.aeaweb.org/doi/10.1257/aer.103.7.2643
Handel, Benjamin R., and Jonathan T. Kolstad. 2015. “Health Insurance for ‘Humans’: Information Frictions, Plan Choice, and Consumer Welfare.” American Economic Review 105 (8): 2449–2500. https://www.aeaweb.org/articles?id=10.1257%2Faer.20131126
Jõeveer, Karin, and Kaido Kepp. 2023. “What Drives Drivers? Switching, Learning, and the Impact of Claims in Car Insurance.” Journal of Behavioral and Experimental Economics 103: 101993. https://www.sciencedirect.com/science/article/abs/pii/S2214804323000198
Ketcham, Jonathan D., Claudio Lucarelli, Eugenio J. Miravete, and M. Christopher Roebuck. 2012. “Sinking, Swimming, or Learning to Swim in Medicare Part D.” American Economic Review 102 (6): 2639–2673. https://www.aeaweb.org/articles?id=10.1257%2Faer.102.6.2639
Kling, Jeffrey R., Sendhil Mullainathan, Eldar Shafir, Lee C. Vermeulen, and Marian V. Wrobel. 2012. “Comparison Friction: Experimental Evidence from Medicare Drug Plans.” Quarterly Journal of Economics 127 (1): 199–235. https://pmc.ncbi.nlm.nih.gov/articles/PMC3314343/
Perrone, Helena, and Fabricio Valiati. 2025. “Drivers of Switching in Autoinsurance: Evidence from Observable and Exogenous Consideration Sets.” CEPR Discussion Paper No. 20707. October 4. https://cepr.org/publications/dp20707
Polyakova, Maria. 2016. “Regulation of Insurance with Adverse Selection and Switching Costs: Evidence from Medicare Part D.” American Economic Journal: Applied Economics 8 (3): 165–195. https://www.aeaweb.org/articles?id=10.1257%2Fapp.20150004
Samuelson, William, and Richard Zeckhauser. 1988. “Status Quo Bias in Decision Making.” Journal of Risk and Uncertainty 1: 7–59. https://scholar.harvard.edu/sites/scholar.harvard.edu/files/rzeckhauser/files/status_quo_bias_in_decision_making.pdf