Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Partial Disability and Residual Earnings Benefit Design

Abstract. Partial disability benefits transfer part of the financial risk associated with continuing but impaired ability to earn. Their design connects clinical and occupational evidence, predisability earnings measurement, benefit calculation, work incentives, claims administration, and the duration of insurer obligations. This monograph examines these relationships through the original 2018 standards for individual disability income policies adopted by the Interstate Insurance Product Regulation Commission, a published specimen insurance contract, American, Canadian, and British public-benefit provisions, international employment-injury standards, industry actuarial experience, and peer-reviewed research on labor supply and return to work. Its original contribution is an integrated contract-to-cash-flow interpretation of existing designs, with independently calculated examples and a source-specific comparison of incentives. A proportional residual formula can preserve a positive return to additional earnings when its replacement benefit is below the insured earnings base; dollar-for-dollar offsets can eliminate that return over a range; and benefit eligibility thresholds can reverse the local incentive altogether. Predisability pay construction and changing earnings introduce measurement risk distinct from medical eligibility. Claim termination experience is also distinct from sustained occupational recovery. The analysis establishes which conclusions follow mathematically, which depend on insurance wording and jurisdiction, and which need linked claims and employment data for causal testing. It proposes a reproducible assessment agenda without establishing a new theory, doctrine, or proprietary rating model.

Keywords: Partial disability; residual disability; disability income insurance; predisability earnings; income replacement; partial employment; work incentives; occupational incapacity; claims duration; actuarial experience; disability policy wording; return to work; benefit offsets; earnings volatility; insurance regulation.

Introduction

A worker can remain employed while losing the ability to perform usual duties at the former level of productivity, hours, or earnings. The resulting economic exposure is a distribution of income losses across time, with possible spells of recovery, relapse, employer accommodation, occupational change, and separation from employment. Residual or partial disability insurance addresses a subset of these losses. Its operation depends on the promise embodied in a policy, the insured earnings base, the evidence of disability, and the applicable claims process. Analysis begins by preserving the distinction between a functional limitation, a reduction in actual pay, a contractual covered event, and the amount that becomes payable.

The ICF framework of functioning developed by the World Health Organization (WHO 2001) situates functioning within health conditions, activities, participation, and environmental factors. This is an indispensable descriptive perspective for work capacity: similar impairments can coexist with very different work opportunities when accommodations, transportation, digital accessibility, task design, and labor demand differ. The ICF does not determine entitlement under a private insurance policy. Contractual definitions, occupation requirements, causation provisions, waiting periods, benefit riders, offsets, and proof of loss remain decisive in that setting.

This study combines primary contract and regulatory interpretation with actuarial measurement, elementary financial mathematics, and appraisal of research designs. The original analytical work is a synthesis: it identifies incentive gradients and contractual discontinuities, tests benefit calculations against explicit assumptions, and specifies the minimum data structure necessary to distinguish sustained return to work from a payment termination. It introduces no new scientific theory or claims of universal causal effectiveness. Every named external study, official instrument, and policy document used to support a material factual claim appears in References and in the accompanying EVIDENCE.json.

Conceptual and Historical Foundations

Partial disability as a specified insured state

Definitions differ across the individual-policy, group-policy, public social-insurance, and employment-injury settings. An individual disability income contract may treat partial disability as inability to perform some substantial duties, inability to perform all duties for the usual duration, or an injury- or sickness-caused reduction in working time or earnings. A contract can also require medical supervision or a period of total disability before residual benefits attach. The Insurance Compact individual-policy standards (IIPRC 2018) expressly describe different permissible partial-disability triggers and proportional payment mechanisms. The instrument applies to the specified individually underwritten forms filed under the Compact, including qualifying multi-life products. It is not a general statute governing every group disability or public social-insurance program, and opt-out and filing-effective-date rules matter.

For analytic purposes, a condition of partial disability comprises three separate observations: an impairment and its consequences; an occupational task-and-hours capacity; and an earnings outcome. These may move together, but none is a universal measure of either of the others. The functional quantity determines what tasks and schedules are feasible; labor-market arrangements determine what work is offered; the contract determines whether actual earnings losses are a covered consequence of sickness or injury. A worker who earns less after changing jobs voluntarily may not satisfy the required causation language. A worker who remains at full pay under a temporary employer guarantee can have substantial clinical restrictions without any immediately measurable income reduction.

A publicly filed Standard Insurance policy specimen (Standard Insurance Company 2024, SEC exhibit) illustrates the distinction between policy sections that define total disability and attached residual disability riders. Its basic text recognizes occupation-specific duties, a transition from regular occupation to any occupation under specified conditions, and a rehabilitation program; it separately lists a basic residual disability rider. A document filed as an exhibit is evidence of the identified policy wording, not evidence that identical wording governs all contracts sold by that insurer or that any particular claimant qualified for a benefit.

The historical logic is older than a particular product form. Public employment-injury systems have long recognized degrees of permanent earning-capacity loss. The ILO Employment Injury Benefits Convention (International Labour Organization 1964, Convention No. 121) distinguishes total from prescribed degrees of partial loss of earning capacity and specifies different permissible cash-benefit forms in Article 14. The treaty concerns employment injury and national implementing systems. Its partial-loss language therefore establishes a useful institutional parallel, not the interpretation of a modern private residual earnings rider.

Four separate benefit environments

Individual disability policies typically insure an underwritten person against qualifying disability as specified in the original policy and riders. Group long-term disability (LTD) plans usually involve employer or association coverage and may use different insured earnings bases, coordination offsets, definitions and appeal systems. Statutory disability programs confer entitlements by national legislative criteria. Workers' compensation or employment-injury regimes may determine partial benefits according to scheduled impairment, wage loss, earning capacity or other national rules. Cross-system comparison is coherent only when the product type, payor, beneficiary, claim condition, legal basis, and unit of observation are stated.

In the United States, Social Security Act section 223 defines disability insurance entitlement by an inability to engage in substantial gainful activity due to a medically determinable impairment meeting statutory duration and other requirements. This threshold is distinct from the individual private-policy concept of residual earnings loss. The SSA trial-work guidance also establishes employment-testing rules for beneficiaries, including a nine-month trial work period and a subsequent 36-month period of extended eligibility; the published 2026 dollar thresholds are administrative year-specific figures. One must therefore avoid comparing a private proportional payment and an SSDI entitlement rule as though they were variations of one benefit equation.

Employment policies and income protection interact. A contract may provide an incentive to resume work while an employer's ability to offer reduced hours determines whether that incentive can be exercised. The UK government's fit-note employer guidance recognizes graded duties, phased return, altered hours, and workplace adaptations as practical arrangements. Their availability varies by occupation and workplace and influences observed outcomes independently of an insurer's residual formula.

Scope of the evidence

The inquiry uses adopted product standards, original policy text, public benefit regulations, official actuarial study summaries, and published empirical economics. These sources answer different questions. A contract tells us what the parties' defined obligations are; a regulation identifies operative legal requirements within its jurisdiction; an experience study describes specified insured populations; and a research paper can estimate behavior under its particular identification design. None is a substitute for the others. This division structures all further analysis.

Residual Earnings Formula

Predisability earnings as the denominator

Let P denote the policy-defined predisability monthly earnings base, E the policy-defined earnings during the claimed month, D the monthly total-disability benefit otherwise available, and B the payable residual benefit before any additional policy limits or offsets. Define the measured earnings-loss ratio as L = max(0,(P-E)/P) for P greater than zero. A common proportional construction is B = D L, subject to the rider's terms. At E = P the measurable loss is zero, while at E = 0 the formula returns D. The mathematical resemblance of this expression to a wage-loss indemnity must not conceal the role of eligibility, waiting periods, and exclusions: multiplication produces an amount only after the insured state has been established.

The IIPRC proportional-benefit standard describes a loss-of-earnings fraction multiplied by the full-disability periodic benefit. It also discusses alternative time-worked triggers, specified 20 percent and 80 percent thresholds, treatment of substantial losses, and the conditions under which a prior total-disability qualification period may be included. It provides special rules for the definition of prior earnings, including a five-year lookback boundary in the described standard and protection for a highest earnings year when the period considered extends beyond one year. These distinctions reveal why one cannot take an insurer's headline replacement percentage and mechanically infer a residual benefit for every claimant.

For a fully hypothetical insured, let P = $8,000 monthly, D = $4,800, and E = $4,000. The measured loss is (8,000 - 4,000)/8,000 = 0.50. Under the unmodified proportional expression, B = $2,400 and gross earned-plus-insured income is $6,400. At E = $6,000 the loss is 0.25, B is $1,200, and the combined amount is $7,200. Both calculations assume that the remaining loss satisfies the medical and causal conditions and that there are no additional offsets, caps, tax effects, or threshold exclusions. They are transparent numerical demonstrations, not quotations of one insurer's exact payout or market observations.

Earnings-loss measurement versus amount replacement

The ratio D/P is the policy's full-benefit-to-prior-earnings fraction. Where it is 60 percent, a 50 percent earnings loss produces a payment equal to 30 percent of the prior earnings level. Hence proportional residual insurance can reduce the variability of income without fully restoring predisability income. Write the combined amount as T = E + D(1-E/P) = D + E(1-D/P) for 0 <= E <= P. The marginal combined-income return to one additional dollar of measured wages is dT/dE = 1-D/P, holding the base and eligibility constant. When D/P = 0.60, the insured retains 40 cents of additional gross combined income for each dollar of earnings before tax and other program interactions. This is an algebraic property of the specified hypothetical formula, not an observed elasticity or actual net-of-tax reward.

By contrast, a benefit construction B = max(0,D-E) within its operating range makes T = D when E is between zero and D. The entire extra dollar reduces the benefit, so the gross combined income does not increase in that interval. A partial offset B = max(0,D-0.5E) produces a 50-cent marginal gain while benefits remain payable. This comparison isolates the financial gradient. It does not incorporate medical effort, travel expenses, time costs, wages that may reflect preexisting productivity, or discontinuities at benefit caps.

The exact wording matters. The SEC-filed Standard Insurance residual provisions specify a proportionate loss-of-income calculation under stated loss brackets, including payment of the basic benefit for losses above an 80 percent condition, zero residual benefit for loss below 20 percent, and a first-six-month minimum in the exhibited form. Accordingly, two contracts with the same stated total monthly benefit can produce different realized trajectories because of their loss boundaries, special minimums, rider definitions, and duration provisions.

Thresholds, notches, and discontinuities

A smooth benefit curve and an eligibility rule need to be read together. Suppose a hypothetical rider pays B = D L only when L >= 0.20, and zero below the threshold. If eligibility starts precisely at a 20 percent loss, the benefit can jump from zero to 0.20D when earnings decline across the boundary. With D = $4,800, that is a $960 payment discontinuity around earnings of $6,400 in the earlier example. Whether the threshold is strict, inclusive, averaged, deemed, subject to minimum payments or otherwise moderated is contract-specific.

At the other end, a deeming rule can replace a calculated 80 percent loss with a 100 percent loss. A transition from 0.80D to D adds 0.20D, which equals $960 in the example. A policy may define precisely when the condition operates, how it treats mixed earnings and whether the full payment applies for specified months. Economically, a notch can generate locally nonmonotonic combined income. Analytically, the policy must be represented as a piecewise function before estimating its incentive effects. A smooth ratio alone will miss the very behavior of interest.

Take a claimed monthly income of $6,399 compared with $6,401 near a hypothetical $6,400 boundary. If an idealized 20 percent minimum-loss trigger applies with a proportional benefit and the base is $8,000, the difference in earned income is only $2, but the first state pays approximately $960.60 in gross insurance benefits and the second state pays none. One should not transfer this calculation to an actual insured unless the rider contains that exact discontinuity and the relevant earnings definition treats the two months as stated. Its role is to demonstrate a structural sensitivity requiring precise legal interpretation.

The numerator is a constructed measure

Pre-disability compensation can contain salary, overtime, bonuses, commissions, partnership draws, deferred compensation, and fluctuating self-employment income. Many contracts define some of these categories differently; the calendar month in which a payment arrives may not correspond to the work that produced it. A worker receiving a bonus in the residual period could appear to have recovered earnings capacity even though the bonus remunerates labor undertaken before disability. Conversely, a temporary earnings drop unrelated to sickness could overstate disability-related loss.

The measurement protocol should therefore record the earnings reference period, the type of pay, timing of performance and payment, accounting convention, cost-of-living adjustment if provided, employment status, leave pay, and causal explanation for changes. The IIPRC standard's explicit definition of prior earnings protects against unrestricted cherry-picking of distant high-income years while imposing specific conditions on longer lookbacks. The reference window is a material part of the insured promise, not an administrative afterthought.

Consider monthly earnings that vary between $4,000 and $10,000 seasonally. If the selected reference is a peak month, a subsequent $6,000 month appears as a 40 percent loss. Against a $7,000 seasonal average, the same actual month indicates a 14.3 percent loss. The difference could determine eligibility under a 20 percent trigger. Neither denominator is inherently universally correct: the governing contract and the correspondence between work and cash flow decide which comparison has legal meaning. Economically, both reference measures also have different bias and variance.

Benefit coordination and net protection

Where several benefits overlap, the financial outcome depends on the order in which insurers apply statutory or contractual offsets. Let B_raw be the residual benefit and O the applicable offset after any coordination exemptions. The final benefit may be represented schematically as B_final = max(0,B_raw-O) when this is the wording, but other contracts apply caps on combined replacement income or define a deductible-source list. The resulting worker income should be calculated as an explicit sequence rather than by indiscriminately summing every theoretical benefit. Taxes, means-tested programs, employee benefits, employer salary continuation, and benefit waiting periods can materially change the result. Every calculation in an actual comparison must disclose which entities' payments and which taxes are included.

Work Capacity and Variable Pay

From functional limitation to compensated work

A diagnosis provides evidence about health; work capacity requires a functional and occupational translation. Relevant dimensions include physical tolerance, cognitive concentration, stamina, shift predictability, environmental exposure, travel demands, task safety and capacity to sustain an agreed schedule. The WHO's ICF environmental perspective explains why a worker's capacity under one arrangement may be different under another. A job with assistive technology and adjusted hours may preserve earnings where an otherwise comparable job cannot accommodate the same limitations.

The UK fit-note guidance on phased work explicitly identifies alterations to hours, duties, and work setting. The US Equal Employment Opportunity Commission's ADA accommodation guidance likewise discusses modified or part-time schedules and reassignment in the scope of US employment law, subject to conditions including undue hardship. Neither employment instrument automatically modifies a private disability insurance contract. The connection is empirical: feasible accommodation affects residual earnings and may change the duration and value of an insurance claim.

This implies that a worker's observed wages cannot be treated as a clinical test with perfect specificity. Actual earnings depend on employer decisions, labor demand, bargaining arrangements and occupational mobility. Work availability is distinct from work capacity. If an employer declines to offer reduced hours, the resulting absence of earnings is a mixture of health-related limitation and institutional constraint. In economic research the distinction is central when attributing effects to insurance incentives.

Labor-supply incentives and their limits

Kostol and Mogstad's (2014) Norwegian work-incentive study uses a policy discontinuity to identify behavioral responses to disability benefit incentives. The authors report a meaningful return-to-work response and potential improvements in recipients' income and public finances for relevant groups. Their findings concern the studied Norwegian institutional setting and policy exposure. They do not supply an insurer-neutral elasticity that can be inserted into an American individual-policy rating basis.

Koning, Muller, and Prudon (2022) examine Dutch benefit applicants and report that disability benefits reduce work resumption after mental-health improvements; the reported empirical comparison is tied to a disability eligibility threshold and to the selected population. Their estimate that approximately half of the work-resumption response to recovery is offset by benefits is a finding of that study, not a general multiplier for all disability claims. Together, these investigations support treating policy incentives as relevant, while also showing why contract architecture, beneficiary selection and health dynamics must enter the interpretation.

The estimated response of hours to a payout gradient is not the same as an incentive gradient itself. The derivative 1-D/P describes a statutory or contractual payoff relationship under fixed assumptions. Labor supply also depends on preferences, health, available hours, wages, access to assistance, and uncertainty about benefit reinstatement. A shallow positive gradient may coexist with a strong return-to-work response if workers value professional continuity and health stability. A steep gradient may have little effect when jobs are unavailable or occupational risks are high.

Variable pay, self-employment, and occupational inequality

Workers paid hourly wages have a comparatively direct relation between reported hours and earned pay. Commission-based workers may work intensively while receiving little contemporaneous cash, and may receive payments long after work is completed. Self-employed claimants may report business receipts that incorporate subcontractor efforts, capital earnings and seasonality. A residual earnings measure must therefore specify whether it tracks personal labor income, business revenue, net profit, or another defined contract variable. This classification can change both claims adjudication and observed behavioral incentives.

Suppose a consultant had a three-year average monthly personal labor income of $9,000 but earned $5,000 during the disability period while an unrelated delayed invoice of $3,000 was paid. Counting receipts makes E = $8,000, a nominal 11.1 percent loss; assigning the delayed invoice to work already performed makes E = $5,000, a 44.4 percent loss. The difference does not establish the correct insured payment. It isolates why an audit trail of the invoice date, service period, and contract's definition of earnings is necessary. Record keeping can resolve facts but cannot override a clear policy clause.

Comparative equity also raises occupational questions. Two equally limited individuals may exhibit different observable earnings effects because one works in an accommodating office occupation and another in manual employment with fixed shifts. Design and adjudication need to distinguish labor-market opportunity from insured disability. An earnings-only trigger may understate limitation for a worker with temporary salary protection; a time-only trigger may pay despite unchanged income. Permitting both types of evidence can recognize more situations but expands the need for precise anti-duplication rules and transparent claims procedures. These are consequences of comparing contractual mechanisms, not a proposal for a new universal insurance doctrine.

Employment support as a complementary intervention

Derbyshire and colleagues' (2024) systematic review of employer interventions synthesizes 71 studies in high-income countries and reports comparatively consistent associations between part-time sick leave or graded return-to-work arrangements and improved participation, while results for several other employer instruments are mixed. This evidence concerns employer-focused employment measures and does not identify the causal impact of a specific residual rider. Its practical significance for insurance is that the availability of a phased position may mediate how the claimant responds to a residual benefit.

In the Standard Insurance specimen, the described rehabilitation program contemplates written agreement and potentially supported training, work adaptations, and related expenses. The provision is useful as evidence that an insurer's role can include services and an employment trajectory, rather than payment only. Its wording also makes program approval and participation conditions material. The amount spent on vocational services is a separate claim cost and should not be confused with the contractual residual benefit to the claimant.

Claim Duration and Incentives

Disability duration, payment duration, and employment duration

A disability episode may continue after the insurer stops paying because a waiting period, coverage expiration, earnings threshold, statutory termination, exhaustion, or documentation rule has changed. Conversely, a worker may resume some employment while remaining eligible for a partial benefit. Thus clinical episode duration, duration in insured partial-disability status, duration of cash payment, and duration until stable employment are four different outcomes. A claims system that treats every termination as recovery will confound events of different economic and medical meaning.

The 2026 SOA/LIMRA group LTD termination study covers claim termination experience from 2015 to 2024. The publicly described data aggregate more than 43 million life-months exposed, approximately 937,000 recoveries, and over 181,000 deaths, contributed by 18 companies. These are aggregate study descriptors, not a calculation of an individual claimant's transition probability or evidence that all terminations are durable workforce returns. Detailed segmentation and underlying standard data package access are subject to the study's published availability conditions.

The 2025 SOA/LIMRA group LTD incidence study covers 2015 through June 2022, reporting approximately 294 million life-years exposed and 1.2 million claims for fully insured group LTD in its defined sample, with specified exclusions including individual disability policies and certain short-duration plans. The magnitude and representativeness of these exposures make the study valuable for group-level incidence benchmarking, yet incidence is a flow into insured claim status. It provides neither a claimant's exact residual payment nor a direct estimate of the effect of a rider on hours worked.

State-transition specification

For measurement, define an initial employed state W, total disability T, partial disability with employment P, paid residual status R, unpaid ongoing impairment U, recovered employment F, death M, and coverage termination C. These are labels for an observational classification, not a new theory of disability. State R may overlap employment; accordingly, the data implementation should either use mutually exclusive combinations or track work status and payment status on separate axes. A monthly panel can record start and stop dates, benefit amount, work hours, earned income and claim decisions independently.

A time-to-event analysis should define time zero consistently: onset of health limitation, end of elimination period, first partial work, or start of residual payment answer different questions. Right censoring arises at the data cutoff; delayed entry occurs when only continuing claims appear; competing events include death, age-based expiry, contract termination, recovery and other eligibility changes. A Kaplan-Meier curve of benefit persistence is interpretable only after such events are treated consistently. Competing-risk cumulative incidence may be preferable when estimating the absolute probability of a named termination cause, while a multi-state model can distinguish total-to-partial and partial-to-full employment transitions.

Suppose 1,000 new insured residual-benefit episodes are followed for one year. If 200 recorded payment terminations are coded as recovered, a simplistic report says 20 percent recovered. If chart review shows 120 stable employment returns, 30 coverage expirations, 20 administrative closures and 30 unresolved records, then the directly observed stable-return fraction is 12 percent within this artificial complete cohort. Further adjustment is required when follow-up is incomplete or individuals re-enter disability. This example contains deliberately invented observations to illustrate classification error and cannot be read as an estimate from SOA, LIMRA, or any insurer.

Public systems as incentive comparators

Canada's Working While on Claim rules allow qualifying Employment Insurance recipients, including sickness-benefit recipients, to retain earnings and part of their EI benefit under a specified deduction schedule: generally a 50-cent reduction for each dollar earned up to the stated prior-earnings limit, followed by a different reduction above it; a full week worked ends benefit eligibility for that week. This is a public employment-insurance rule, not an individually underwritten LTD rider. It nevertheless provides an actual institutional example of a partial offset that leaves an initial positive marginal earnings return.

In the United Kingdom, ESA permitted-work conditions (updated April 21, 2026) specify that a qualifying claimant can generally work less than 16 hours per week and earn no more than the stated weekly amount while retaining the normal ESA benefit, with distinct conditions for supported permitted work. The official factsheet's April 2026 threshold is GBP 203.50 per week after tax and National Insurance. These are named-program, date-specific rules: a person receiving another benefit, or subject to a changed later policy, may be governed by different conditions. The policy mechanism is qualitatively distinct from Canada's proportional EI earnings deduction and from private insurance's loss-of-earnings fraction.

The United States' SSDI work rules for 2026 instead include a trial-work regime with full payment during qualifying months and a subsequent earnings-level eligibility test. During the nine trial months, the threshold design is different from a continuous residual earnings benefit. When studying the combined income of an individual with employer LTD, SSI or SSDI, one must establish program interaction and possible private-plan offsets before calculating the incentive curve. There is no jurisdiction-free rate of benefit reduction.

Timing and uncertainty of recovery

Benefit duration depends on initial severity, functional progression, occupation, age, comorbidities, accommodations, workplace conditions and policy design. Conditional comparisons between partial workers and totally absent workers face selection: people who resume partial employment may already have better recovery prospects, more flexible employers or different diagnoses. A favorable subsequent outcome among part-time workers can arise partly from these prior differences. Moreover, a benefit that stabilizes a worker's income may support treatment adherence and a gradual return, even when its nominal earnings offset dampens the cash reward. A credible evaluation must test these channels rather than infer causality from an unadjusted duration difference.

The insurer has another time horizon: present value of future insured payments and associated expense flows. For a simplified monthly reserve, the conditional expected present value can be represented as V = sum over future months t of v(t) times the probability of qualifying residual status at t times the expected contractual benefit at t given that status. The payment amount and the probability of remaining eligible may both depend on time-varying earnings. Combining incidence, claim termination, wage recovery and contractual caps into a single constant factor obscures both actuarial risk and behavioral response.

Evidence Validation and Sensitivity

Source hierarchy and claim verification

Original contracts and adopted standards are authoritative for the wording they actually contain. Their jurisdiction, edition and effective dates limit interpretation. The IIPRC 2018 standards establish an individually underwritten US Compact product-filing framework, not a generalized international rule. The SEC policy filing supplies a particular specimen text rather than market prevalence. The US Department of Labor claims guidance explains that 29 CFR 2560.503-1 sets claim and appeal procedures for ERISA-covered plans, with disability-specific requirements. The claims procedure framework is not a formula for benefit magnitude.

Official actuarial study descriptions define data years, exposure units, insured business covered, sampling frames, and reported aggregate results. The SOA incidence publication is therefore appropriate support for its reported group data scope but cannot validate a private residual benefit prevalence rate. The SOA termination publication supports reported study totals but, without cross-tabulations and individual trajectories, is insufficient to isolate causal return-to-work effects. Empirical scholarly articles supply research designs and conclusions within their populations; the present monograph does not pretend to re-estimate their coefficients from inaccessible microdata.

Sensitivity of the denominator

For proportional B = D(1-E/P) with E fixed and positive, an increase in P raises the residual payment: dB/dP = D E/P squared. In a hypothetical case D = $4,800, E = $4,000, P = $8,000, a 5 percent increase in the denominator to $8,400 changes B from $2,400 to approximately $2,514.29, an increase of about $114.29. A corresponding decrease of P to $7,600 reduces the payment to about $2,273.68. The sensitivity is asymmetric for equal percentage changes around the original base because the formula is nonlinear in P. In actual policies, indexing and benefits may interact, so one must verify whether D also changes with P before applying this derivative.

A separate discontinuity appears if a revised P changes the measured loss across a minimum threshold. Even a small denominator difference can alter whether the rider's medical and earnings eligibility test is met. For this reason, actuarial audits should assess formula sensitivity and threshold sensitivity separately. A payroll-data-quality improvement can cause a larger legal payment difference than its apparent 2 or 3 percent change in the earnings base suggests.

Sensitivity to earnings timing

For volatile compensation, run at least three pre-specified comparisons: payment-date cash earnings, accrual-based earnings attributable to work during the month, and a permitted rolling average under actual policy terms. Under the proportional expression, every additional $1,000 of counted E reduces B by D/P times $1,000, assuming there is no threshold or cap. With D/P = 0.60 the difference is $600. When a benefit threshold is crossed, the discrete impact can be more substantial. The analysis should document the accounting convention before examining which version improves a claim, to avoid outcome-contingent reconstruction.

Selection, censoring, and competing explanations

An observational return-to-work study requires a clear eligible denominator. Relevant contrasts may compare cohorts at benefit entry rather than only claims still open a year later; conditioning on survival in claims can induce selection bias. Calendar-time labor markets, shifts in medical management, modifications to eligibility rules, employer leave practices, changes in plan mix, and wage inflation are candidate confounders. Censoring must be identified; recurrent disability should be coded rather than silently treated as permanent recovery. A program threshold can support quasi-experimental inference, as in the Kostol and Mogstad study, but results transport only to populations and institutional margins that are substantively comparable.

A research package should retain the original policy edition, rider versions, wage records, health-related restrictions at decision time, occupation descriptors, reported hours, earnings by service period, work accommodations, insurer payment determinations, public offsets, death and exit codes, re-entry dates and adjudication appeals. Observed state must be distinguished from imputed state. Analyses that use algorithmic classifiers should report validation and uncertainty because an administrative code for 'claim closed' is not a clinical finding.

Legal and ethical review of benefit decisions

The IIPRC and ERISA materials must remain tied to their applicable form and plan scopes. ERISA disability-benefit procedures include disclosure and review protections; the Department of Labor's disability claim guidance is a starting point for checking claims-administration compliance within covered employee plans. An individually underwritten insurance policy can be governed by different state law and claims-handling instruments. Cross-border studies should record substantive entitlement rules and procedural review separately.

Disputes over the proper earnings base can affect claimants with volatile pay or intermittent work. Measurement choices can reproduce occupational or employer inequalities when they do not account for work-related timing, contractually permitted averages and genuine health limitations. Evaluating fairness requires matching the legal promise with reliable evidence, documenting alternative calculations where the policy permits them, and providing an understandable explanation of why one approach was selected. Those requirements follow from scientific traceability and contractual administration; this manuscript does not claim that a proposed universal earnings measure has legal priority over policy text.

Verification ledger and methodological limits

The companion evidence ledger links each externally sourced proposition to its primary institutional or scholarly location. All examples in this monograph are deliberately illustrative and were calculated from the stated numbers. The review accessed public regulatory and policy text, open institutional webpages, and the publicly accessible metadata and abstracts of peer-reviewed work; it did not obtain insurer-specific microdata, restricted SOA standard data packages, a representative international library of filed policies, or a country-by-country judicial case-law census. No original clinical study, actuarial claim-rate estimation, causal identification, meta-analysis, or peer review was performed. These limits confine the conclusions in a precise and scientifically meaningful way.

Scientific Synthesis and Research Agenda

An integrated reading of existing benefit design

Residual disability coverage contains at least four analytically separable decisions. Eligibility determines whether the worker's current limitations and earnings satisfy a contract; measurement determines P and E; the payout rule maps the measured loss to B; duration and coordination rules determine how long and how much is actually paid. Work incentives are the combined outcome of all four decisions plus taxation, alternative benefits and employer opportunity. Studying only the headline replacement ratio omits the variables most likely to account for sharp outcome differences near thresholds.

This synthesis leads to an important comparison. A proportional formula can preserve a positive return to extra earned income whenever D/P is below one; at the same time it reduces the marginal gross return relative to uninsured earnings. A dollar-for-dollar offset can produce a flat combined-income region. A stepped eligibility or deeming rule can cause a cash-income jump when an earnings or hours boundary is crossed. These are logical consequences of each stated rule, independent of assumptions about claimant motivations. Their actual behavioral importance is an empirical question.

The loss denominator also mediates the incentive. A higher P increases measured loss for the same E and thus can affect both benefit amount and threshold status. If a policy uses the highest qualifying prior earnings year, claimant incentives cannot be inferred from a single recent paycheck. If a contract uses indexed prior earnings, nominal wage recovery can coexist with continuing real-income loss. The proper earnings base is therefore part of the insurance's economic transfer function as well as its legal definition.

What the comparative record establishes

The IIPRC text supplies a concrete example of regulated individual-policy residual calculations and earnings definitions. The exhibited private policy shows how those calculations appear in specific product wording. Canada's EI provisions demonstrate partial public offset of earnings; UK's ESA permitted work shows an hour-and-earnings allowance; and US SSDI shows trial-work and subsequent eligibility rules. Their common analytical feature is an interaction between work and benefit continuation. Their legal and financial mechanisms are substantially different, and the comparison is valuable precisely because these differences are explicit.

The empirical evidence further establishes that behavioral incentives can matter: Norwegian and Dutch studies identify labor-market responses in specific social-insurance arrangements. The 2024 employer-intervention review documents the importance of workplace structures and phased work. Actuarial group LTD research provides extensive descriptive incidence and termination experience but does not, by itself, estimate the effect of residual-payment formulae. A sound international synthesis respects this division of labor among sources rather than constructing a fictitious pooled effect.

Research questions capable of adjudication

A priority study would compare contract designs on the same outcome definitions: elapsed months until sustained work, total work earnings over two years, net combined income, residual payment cost, recurrent disability, employment retention and claimant understanding of rules. For each cohort, the policy's own loss thresholds and pay treatment should be coded before any statistical estimation. Analyses can then assess whether observed bunching occurs near exact earnings boundaries and whether it is explained by reported earnings timing, employer scheduling, or behavioral adjustment.

A second study would investigate earnings reference windows in commission-intensive and self-employed occupations. It should evaluate both variance of calculated loss and frequency of threshold crossings, while using predefined rules for seasonal compensation, delayed payments and income attributable to work before disability. A third study should examine whether phased-work access modifies the effect of contractual gradients across occupations and firm sizes. These are research designs and open questions, not assertions that any particular rider is already demonstrated superior.

For causal interpretation, promising opportunities include changes in contractual provisions, staggered employer plan adjustments, exogenous program thresholds, or administrative differences that satisfy explicit identification assumptions. Difference-in-differences would require assessment of parallel pretrends; regression discontinuity would require scrutiny of manipulation around the running variable; linked claimant-employer panels would improve attribution to actual work. Findings should report external validity limits, substitution across benefits, administrative costs, and outcomes for claimants whose health deteriorates again.

Implications for actuarial, contractual, and public analysis

An actuary evaluating a residual rider should distinguish entry incidence, time in each qualifying state, earnings conditioned on the state, the distribution of payment amounts, and transitions back to total disability or out of coverage. A contract specialist should verify the earnings base, rider applicability, required causal nexus, minimum-loss threshold, benefit floor or cap, and claims-review rules before proposing an interpretation. A public economist should examine cumulative household resources, labor opportunities and work incentives after all overlapping programs are mapped. The three approaches share data requirements but ask different questions.

The main editorial conclusion is that an evidence-based residual benefit analysis begins with the contract and statutory text, moves through auditable individual cash flows, and only then draws conclusions about claims experience and labor-market response. The sequence prevents a common category error: treating a favorable-looking payout formula, a claim closure statistic, and a successful return-to-work outcome as equivalent measures of benefit design quality.

Conclusion

Partial disability and residual earnings benefits are mechanisms for protecting income while some ability to work remains. Their economic results depend on the precise interaction of eligibility, measured predisability earnings, current earnings, payment formula, thresholds, offsets, duration, and workplace opportunity. A proportional benefit can preserve a positive marginal return to earnings; a dollar-for-dollar offset can neutralize that return; and threshold discontinuities can generate substantial local financial differences. Those conclusions follow directly from the stated mathematical structures.

Original policy and regulatory documents confirm that residual benefit designs vary and that earnings measurement and partial-work definitions are substantive contractual terms. The comparative public schemes in Canada, the United Kingdom and the United States illustrate distinct administrative ways to manage work during disability support. Peer-reviewed evidence from Norway and the Netherlands supports the relevance of financial incentives to work resumption, while the international employer-intervention literature emphasizes feasible phased employment. SOA/LIMRA experience studies provide important insured claims context whose denominators and event classifications must be preserved.

The resulting scientific conclusion is operationally precise: benefit adequacy, claimant incentives, legal entitlement, and sustained occupational recovery should be evaluated as connected but independently measured dimensions. The monograph has provided original analytic comparison and sensitivity calculations, not a new theory or unverified priority claim. Future empirical evaluation requires original rider versions, documented pay histories, medical and occupational measures, complete transition records, and transparent jurisdictional mapping.

References