Disability insurance: what are the issues, what are countries doing and how can they do better?

ByEstelle James

World Bank Pension Mini-Course, 2011

The economic rationale for mandatory disability insurance

• Most people reach old age—when young we can plan and save for the loss in earnings when old.

• A smaller number lose earning capacity when young due to major health problem (disability). Small risk of large loss. Creates need for risk-pooling through insurance.

• But problems with voluntary insurance– people may be myopic, underestimate risk, don’t purchase

enough insurance voluntarily– Adverse selection—only those with health problems insure

voluntarily; this raises premium for average person– Cherry-picking by insurance companies if allowed to exclude

• Mandatory insurance avoids these problems– prevents poverty, maintains living standard of disabled

• But many pitfalls—1) disability difficult to measure, often inaccurate; 2) moral hazard widespread; 3) systems can be very costly; 4) labor supply and GDP reduced by system incentives 2

Typical plan before recent reforms• Usually handled by public system, even in countries with

multi-pillar systems (except Latin America)• PAYG, so long-term costs are hidden• Defined benefit, based on recent wages, even when old

age system is defined contribution• Replacement rate often high in contributory plans >70%• Worker presents case with evidence from own GP • Pressure on agency to be sympathetic, so low denial rate• Criterion: inability to do regular line of work• Eligibility often stops if recipient earns wage—so

beneficiaries almost never go back to work• Costs 1-6% of GDP, 1-10% of wages, 10-90% of OA costs– Larger plans are in middle and high-income countries– Low-income countries should be cautious, because of

problems and complexities of running disability plans

3

Public and Mandatory Private Spending on Incapacity as % of GDP, 1995-2005

(Source: OECD Social Expenditure Database 2008 (SOCX))

4

Australia Denmark Germany Mexico Norway Sweden Switzerland UK US Netherlands Poland OECD av.0

1

2

3

4

5

6

7

199520002005

Public + Mandatory Private Spending on Disability/Old Age Spending, 2005

(Source: OECD Social Expenditure Database 2008 (SOCX))

5

Mexico Poland US Germany UK Switzerland Denmark Sweden AustraliaNetherlands Norway OECD av.0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

Dis/OA

Problems: 1. Disability is ambiguous, subjective, difficult to measure

• Intended for cases where health issues severely limit work– but little direct connection between observable medical diagnosis

and unobservable functional capacity• Majority of new claims are due to psychiatric or muscular-

skeletal condition (arthritis, pain)-difficult to prove/disprove • Limited information—traditionally, worker’s own GP • So very prone to errors, high administrative costs

– U.S. audit found 48% false denials, 19% false approvals – self-assessed disability compared with SSA decisions: 20% false

positives, 60% false negatives (Benitez-Silva et al 2004)– In other high-income countries errors are around 40%

• What is “right” mix of false positives vs. false negatives?• Should disability be treated as separate category or not?• Well-trained examiners & administrators needed for

accurate assessment—scarce in low-income countries 6

2. Moral hazard and agency problems (insurance increases risky behavior)

• People who dislike their job are less likely to continue working, more likely to collect benefit, when they have health problem that is insured

• Doctors and gatekeepers may apply lax definitions of disability to keep clients happy

• Employers use disability program to get rid of low productivity workers

• Politicians use it to cut unemployment rate• Unions use it as early retirement pathway• Local governments use it to get people off welfare rolls• Opportunities for moral hazard are increased by

ambiguous definition • Important object of system design—keeping moral

hazard low by better information and incentives that align interests of system and key decision-makers 7

3. Systems are costly, costs are rising• Costs <1% of GDP in low-income countries, 2-6% of GDP

(4-10% of wages) in high-and middle-income countries (except LAC), highest in Nordic countries

• Rising costs in most countries, but dramatic cuts in some (Netherlands, Poland)

• Why the rise and difference across countries and time?• Population health and aging don’t play large role• Many studies show applications and recipiency rates rise with

generous benefits, high replacement rates, lax screening (when benefit goes up 10%, awards go up 3-4%, B&B 1999)

• High UnE rate, low UnE benefit increase disability recipiency (workers lose jobs, can’t find new one, claim disability, keep it)

• Ambiguous definition and moral hazard add to problem• Recent reforms have tried to tighten eligibility

conditions, increase information to screeners, reassess more frequently, change incentives facing workers

8

Disability benefit recipients as % of population age 20-64, 1995 & 2008

(Source: www.oecd.org/els/disability)

9

0

2

4

6

8

10

12

1995

2008

4. Systems reduce labor supply, GDP• Studies show growth of disability system is a major

reason for declining lfpr among older men. Why?– Benefit is for people who can’t work so applicants don’t work – Benefit provides income that enables beneficiaries to work less

—income effect– Beneficiaries face high implicit tax--benefit stops if they try to

work so they don’t try--work disincentive effect– Skills depreciate while out of work so once people start

disability benefits they practically never return to work– Many studies show higher benefits and lax screening reduce LFP

of men over age 45 (when benefit rises 10%, non-employment rises 3%, Gruber 2000; when denial rate rises 10%, LFNP falls 2.7%, Gruber & Kubic 1997)

• Big policy push now is how to reduce disability inflow, increase outflow, keep partially disabled on the job

Parametric reforms that increase information & accuracy, decrease

moral hazard and costs• Eligibility--moving toward any job, not regular job• Benefit amount—often 70% RR, >80% for low

earners. Lower benefit decreases recipients, increases work (but means less insurance)• Screening—denial rate around 45% in OECD – Individual’s own GP or system doctors?– Who else participates? Employers? Voc. couns.? Ins. co.?– How to encourage self-selection (applicant bears part cost)– Stop easy entry through generous sick leave

• Avoid permanent disability, require reassessment 11

Ex: Poland—temporary benefits and better information

• In 1995 disability cost 5.7% of GDP, by 2005 only 2.7%. Disability inflows fell 2/3. How did they accomplish that?

• Since 1999, almost all benefits are temporary• System doctors can reassess and over-rule GP certifications• Any job instead of own-job criterion• Replacement rates reduced• But—many successful appeals and re-applications, UnE

benefit recipients and early retirement pensioners increased• Beneficiaries still lose benefits+child supplements if they work

—high implicit tax on work so few beneficiaries work• Step in right direction but may reverse if ER pensions tighten• Need better incentives, party with direct interest in cutting

costs, lower implicit tax12

Structural reforms:

• 1) large informational role to private parties who have a pecuniary interest in controlling costs (employers, insurance companies, pension funds, voc. counselors)

• 2) eliminate implicit tax on work: beneficiaries work, contribute to economy, keep benefit but get part of income from wage (but this makes it espec. important to label disabled correctly ex ante)

• Examples: Netherlands, Denmark, Australia, Chile13

Netherlands—employer responsibility• In 1995 Netherlands had incapacity cost of 5.7% of

GDP (10% of payroll)– Employers used dis. system to get rid of excess workers,

politicians to keep UnE low, unions for early retirement• By 2005 cost fell to 4.3% of GDP. What did they do?• Major responsibility for disability was shifted to

employers and private service providers– employer bears full cost for 2 years of sick leave– must use private occupational health services to

monitor absent workers and get them back to work– this provides information about health of worker,

gives employer incentive to keep worker on job by rehab and workplace modification

14

Netherlands 2: wage supplements

• After 2 years of sickness leave, worker is assessed for disability benefit– if capacity loss is <35%, no disability benefit– if capacity loss is 35-79%, worker gets wage

supplement if he continues to work partially (wage subsidy instead of implicit tax)

– if he doesn’t work he gets lower flat benefit– Employer finances this for first 10 years (by self-

insurance or experience rated external insurance)– If capacity loss >79%, public system pays full benefit

• Assessment by social insurance doctors & voc expert• Criterion: ability to do any job (not one’s regular job)• Replacement rate reduced slightly• 2004-9: stock of beneficiaries < age 45 reassessed

15

Netherlands 3: success and problems• Results—inflow into sick leave and disability fell sharply,

outflow rose, costs fell. Employers have incentive to monitor worker, find job for him, provide information. We don’t know if partially disabled work more.

• Potential problems: firms shift to temporary contract workers who are not employer’s responsibility

• Employers try to avoid hiring high-risk and older workers• Small and medium sized firms at a disadvantage• High reporting and administrative costs for employer• Reassessed workers may shift to other welfare programs• Govt. has offset this with “no risk” policy—public system

covers temporary contract workers, small firms, newly hired disabled workers (>50% of total in public scheme)

• Employer responsibility would not work well in countries with much casual temporary employment, small firms 16

Denmark—wage subsidies & flex jobs• Concerned by rising absenteeism and costs,

reformed mid-2000’s• Partially disabled get access to subsidized flex-jobs

(part time job for full pay) instead of benefits, hoping to keep more disabled at work for partial benefit

• Administered by municipalities but center pays most • Problems: – many workers applied—but half of all flex jobs go to

workers already employed by firm on ordinary terms– Not enough flex jobs to meet demand so long waiting list

developed (on full benefit)– Number of full beneficiaries did not fall– Shift from full time to subsidized part time workers, not

from full to partial beneficiaries. Costs rose. • Need party with good information and strong

incentives to counter increased applications ong17

Australia—no benefits for partial disabled; outcome-based fee for private voc

counselors• Concerned by rise in beneficiaries and costs 1985-95,

so reformed system; costs began to fall• Tightened criteria: Full disability benefits go only to

those unable to work >15 hours per week• No dis benefits for partial disability; they get

temporary UnE benefit, retraining, job search • Voc rehab by private companies, fee based on

employment outcomes—for 13 & 26 weeks of work• Potential problems: do workers keep job for long

term, are they good jobs, do providers try to cream healthy workers, implicit tax on work remains high, espec. for potential low earners. Evaluation needed.

18

Chile—private pension funds provide information, control costs

• Unlike most countries, disability benefits are partially pre-funded and run by private sector (but regulated)

• Since 1981 old age pensions based on individual accounts

• IA’s also provide benefit if worker becomes disabled– worker is promised replacement rate of 70% (50% if partial dis.)– if money in account isn’t enough to cover benefit, it is topped

up by group insurance policy purchased by pension funds• Pension funds and insurance companies have incentive

to keep approved claims and costs low—increases profits• Assessments are made by Medical Board chosen by

public agency, but pension funds and insurance companies participate in hearings, provide information, ask questions, bring appeals

19

Chile: no implicit tax on work• If beneficiary works he keeps own-wage + benefit• Results: low inflow into disability, lower costs, higher

accuracy than other countries or old system. In new system (but not in OS), benefits go to most severely disabled with highest mortality rates, yet work propensities of dis. pensioners did not fall

• Prefunding through individual accounts further reduces costs & claims in long run; higher cost due to population aging is offset by more funds in account

• Problems: too many false negatives? pension funds may try to cream low-risk workers?– since 2008 all workers put into 1 insurance pool 20

Lessons: 1) information and incentives matter

• Keeping replacement rates low is most sure-fire way to keep disability recipiency rates and costs low and employment high—but this also means low insurance so not done much

• Careful screening is second important mechanism– object should be accuracy– should be based on capacity to do any job (not regular job)– should be done by system’s medical experts, not by own GP– two-sided appeals should be permitted– prepare for higher denial rate when early retirement is tightened

• Mobilize self-selection whenever possible– allow sick and disabled workers to bear some cost—waiting

period, benefit <wage, retraining & job search required– in IA systems, apply balance in account toward disability benefit – smaller net gain will discourage healthier workers from applying

21

Lessons: 2) avoid high implicit tax on work for the partially disabled

• Total and permanently disabled should be a small group, can be more objectively defined, given large benefit

• Majority of disabled should be partial and temporary: can get transitional benefit, job search help, retraining

• If stay out of work for extended period, they will never return, become dependent on benefits

• Important to avoid high implicit tax--give modest benefit that they keep for several years, even if they work

• Maybe subsidize wage to offset lower productivity (but employers & workers may overuse—careful targeting needed)

• Periodic reassessment of beneficiaries is essential (but this conflicts with elimination of implicit tax)

• Difficult to get it right 22

3) Give key party financial incentives to provide information, control costs

• Involvement of private sector– Employers in Netherlands, pension funds in Chile,

private service providers in Australia– This provides detailed information, targeting– But leads to selection/creaming/hiring issues

• Coordinate disability benefit with unemployment and early retirement benefit, to minimize spillover effects and benefit shopping.

23

4) Disability in low and middle-income countries

• Disability systems are complex to design, require skilled personnel. Low and middle-income countries with scarce human capital should be cautious: small system with modest benefits based on clear-cut medical conditions.

• They should set up structures and procedures that will avoid poor information, perverse incentives, high costs and low LFP in the future.

24