Hernan Winkler, Elizabeth Ruppert Bulmer and Hilma Mote

Expanding Social Insurance Coverage to

Informal Workers

JOBS WORKING

PAPERIssue No. 6

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

©2017InternationalBankforReconstructionandDevelopment/TheWorldBank.

1818HStreetNW,Washington,DC20433,USA.Telephone:202-473-1000;Internet:www.worldbank.org.

Somerightsreserved

ThisworkisaproductofthestaffofTheWorldBankwithexternalcontributions.Thefindings,interpretations,andconclusionsexpressedinthisworkdonotnecessarilyreflecttheviewsofTheWorldBank,itsBoardofExecutiveDirectors,orthegovernmentstheyrepresent.TheWorldBankdoesnotguaranteetheaccuracyofthedataincludedinthiswork.Theboundaries,colors,denominations,andotherinformationshownonanymapinthisworkdonotimplyanyjudgmentonthepartofTheWorldBankconcerningthelegalstatusofanyterritoryortheendorsementoracceptanceofsuchboundaries.

NothinghereinshallconstituteorbeconsideredtobealimitationuponorwaiveroftheprivilegesandimmunitiesofTheWorldBank,allofwhicharespecificallyreserved.

RightsandPermissions

ThisworkisavailableundertheCreativeCommonsAttribution3.0IGOlicense(CCBY3.0IGO)http://creativecommons.org/licenses/by/3.0/igo.UndertheCreativeCommonsAttributionlicense,youarefreetocopy,distribute,transmit,andadaptthiswork,includingforcommercialpurposes,underthefollowingconditions:

Attribution—Pleasecitetheworkasfollows:Winkler,Hernan,ElizabethRuppertBulmerandHilmaMote.2017.“Expandingsocialinsurancecoveragetoinformalworkers”.WorldBank,WashingtonDC.

Translations—Ifyoucreateatranslationofthiswork,pleaseaddthefollowingdisclaimeralongwiththeattribution:ThistranslationwasnotcreatedbyTheWorldBankandshouldnotbeconsideredanofficialWorldBanktranslation.TheWorldBankshallnotbeliableforanycontentorerrorinthistranslation.

Adaptations—Ifyoucreateanadaptationofthiswork,pleaseaddthefollowingdisclaimeralongwiththeattribution:ThisisanadaptationofanoriginalworkbyTheWorldBank.ViewsandopinionsexpressedintheadaptationarethesoleresponsibilityoftheauthororauthorsoftheadaptationandarenotendorsedbyTheWorldBank.

Third-partycontent—TheWorldBankdoesnotnecessarilyowneachcomponentofthecontentcontainedwithinthework.TheWorldBankthereforedoesnotwarrantthattheuseofanythird-party-ownedindividualcomponentorpartcontainedintheworkwillnotinfringeontherightsofthosethirdparties.Theriskofclaimsresultingfromsuchinfringementrestssolelywithyou.Ifyouwishtore-useacomponentofthework,itisyourresponsibilitytodeterminewhetherpermissionisneededforthatre-useandtoobtainpermissionfromthecopyrightowner.Examplesofcomponentscaninclude,butarenotlimitedto,tables,figures,orimages.

AllqueriesonrightsandlicensesshouldbeaddressedtoWorldBankPublications,TheWorldBankGroup,1818HStreetNW,Washington,DC20433,USA;fax:202-522-2625;e-mail:pubrights@worldbank.org.

ProjectNumber:P164623�

Images:©WorldBank.Furtherpermissionrequiredforreuse.

1

ABSTRACT

The high incidence of informality in developing countries implies thatmanyworkers are not coveredagainst important risks, suchasunemployment, illnessandold-agepoverty.Given thatexpanding theBismarckian system to include informal workers presents many challenges, several countriesimplemented non-contributory social insurance programs to expand coverage. However, thesecontributed to labor market segmentation and are unlikely to be financially sustainable. This notereviews the economic literature dealing with the expansion of social insurance programs andsummarizes themain policy insights. It draws on international evidence on social insurance systemdesignandinnovations,andtheresultingimpactoncoverage.Italsoprovidesgeneraldesignprinciplesthatcanapplytounemploymentbenefits,healthinsurance,andpensions.

2

ACKNOWLEDGEMENTS

ThisnotewaspreparedbytheWorldBankGroup’s(WBG)JobsGroup.TheprincipalauthorsareHernanWinkler, Elizabeth Ruppert Bulmer and Hilma Mote. The report was prepared under the generaldirectionandongoingsupportofAlvaroGonzalezandDavidRobalino.TheauthorsacknowledgetherichcommentsprovidedbyTrumanPackard,peerreviewerofthisdocument.

The publication of this report has been made possible through a grant from theWorld Bank’s JobsUmbrellaTrustFund,whichissupportedbytheDepartmentforInternationalDevelopment/UKAID,andthe Governments of Norway, Germany, Austria, the Austrian Development Agency, and the SwedishInternationalDevelopmentCooperationAgency.

ThereportwaspreparedunderprojectFirms,AccesstoMarkets,ValueChainsandJobs(P164623).

3

CONTENTS ABSTRACT............................................................................................................................................1

ACKNOWLEDGEMENTS.........................................................................................................................2

1. INTRODUCTION............................................................................................................................4

2.INTERNATIONALEXPERIENCEINEXTENDINGSOCIALINSURANCETOINFORMALWORKERS.............8

3.DESIGN,IMPLEMENTATIONANDPOLICYOPTIONS.........................................................................12

3.1 Registeringandenrollingworkers.................................................................................................12

3.2 Linkingcontributionstobenefits..................................................................................................14

3.3 Designingredistributivearrangements.........................................................................................15

3.4 Providinginformationandincentivestocontribute/save............................................................16

3.5 Monitoringandenforcement........................................................................................................18

4 CONCLUSIONS............................................................................................................................20

REFERENCES.......................................................................................................................................21

4

1. INTRODUCTION

The high incidence of informality in developing countries implies thatmanyworkers are not coveredagainst important risks, suchasunemployment, illnessandold-agepoverty (Figure1). InSub-SaharanAfrica, for example, typically notmore than 5 to 10 percent of theworkforce is in formalworkwithaccess to social insurance (Van Ginneken, 2010). Inmany Latin American countries, informality ratesexceed60percent (Tornarolli et al., 2014). This reality of thepredominanceof informal employmenttypicallytranslatesintolowandunstableearningsinlowproductivityactivitiesandahighincidenceofpoverty.Notallinformalworkersarepoor,particularlyamonghighlyskilledworkersinundeclaredself-employmentandinformalwagework.Buteventhesenon-poorarevulnerableto incomeshocksfromunemployment,illnessandlongevity.

Figure1SnapshotofInformalEmploymentaroundtheWorld(%oftotalemployment)

Source:ILOSTAT.Informalemploymentcomprisesallworkersoftheinformalsector(inunregisteredfirms)andinformalworkersoutsidetheinformalsector(own-accountworkers,andemployeesnotcontributingtosocialsecurity,payingincometaxes,orwholackcertainbenefitssuchasfamilyleaveorsickleave).

Observed employment trends suggest that formalwage employment is unlikely to become themainsource of jobs. The development process that in the past brought economies dominated bydecentralized primary production tomodern industrial sectors comprised ofmostly formal jobs is nolongertheprevalentpathforeconomicdevelopmenttoday(Choetal.,2012;GindlingandNewhouse,2014). Rather, the structural transformations currently underway in the developing world arecharacterized by transitions out of agricultural activities and other primary production into mainlyservice activities, much of which in non-wage employment. Stable and formal wage employment isbecoming lesscommoneveninmiddle-andhigh-incomecountries,withtheemergenceofalternativeworkarrangementssuchasself-employment,part-timeworkortemporarycontracts–allofwhichtendtolacksocialinsurancecoverageandotherlaborprotections(Kellyetal.,2017).

5

The coverage of current Bismarckian social insurance systems is unlikely to expand considerably.Indeed, these systems are financed by employer and employee contributions, and designed for atraditionalemploymentrelationshipwhereworkershold formalsalaried jobswiththesameemployerthroughouttheirworkinglives.1ExpandingtheBismarckiansystemtoincludeinformalworkerspresentsmany challenges. First, Informal workers are unlikely to be registered, especially in developingeconomies, thereby governments struggle to identify, locate andmonitor individuals in this category.Second,enforcingsocialinsurancecontributionsformicro-,small-andmedium-sizedfirmsorfromsmallagriculturalproducersisdifficult,asinspectioncostsarehigh,andcanleadtoadministrativelyinefficientsystemsthatadddistortionstothe labormarket(AlmeidaandCarneiro,2012).Third,theproductivitylevels of informal workers and/or firms are often too low for them to afford social insurancecontributionsandtheassociatedtransactioncosts.

Several countries implemented non-contributory social insurance programs to expand coverage butthesecontributedtoincreaselabormarketsegmentation,andarelikelytobefinanciallyunsustainable.Socialpensionsornon-contributoryhealthinsuranceareeffectivemechanismstoreachouttoworkerswho are “outside” the mandatory contributory systems. The problem is that these programs canbecomeanimplicittaxonformaljobsandreduceincentivestoenrollandcontribute;particularlywhere

1 This systemdesignwaseffective in settingsof youngpopulationage structureswheremostworkers inaneconomymadepayroll contributions throughout a 40-year career, and retirement periods averaged less than 10 years due to shorter lifeexpectancies.

BOX1:SOCIALINSURANCESYSTEMS:SOMECONCEPTS

Socialinsurancesystemshavetwogoals:consumptionsmoothingandpovertyprevention.Asocialinsurancesystemseekstoprovidepeoplewithresourcestosmooththeimpactsofnegativeincomeshocks on their consumption. To prevent poverty, social insurance aims not only to protectindividualsfromnegativeshocks,butalsotoguaranteethemwithaminimumlevelofconsumptionthat is socially acceptable. These goals are many times indistinguishable from each other,particularlyin lowincomecountriesandincaseswhereanegativeshockpushesthenon-poorintopoverty.

To achieve these goals, social insurance systems have two instruments. The first one is a risk-poolingmechanism,whereindividualsandemployerscontributetoacollectivefundtofinancethetransfers to those who face a negative shock. While it is not always the case in practice, thesecontributionsaresupposedtobeactuariallyfairtoincentivizeparticipation.Thesecondinstrumentincludessavingsarrangements,whereindividualssavemoneyin individualsavingsaccountstopayfortheirconsumptionwhentheyfaceanegativeshock.

Becauseof liquidityconstraints, social insurancesystemsalso require redistributivearrangementsthat can operate in two ways: by subsidizing contributions (either in the form of subsidizedpremiums or topping-up savings), or by subsidizing benefits (for instance, by guaranteeing auniversalminimumpensionlevel).

Anotherimportantaspectofasocial insurancesystemis itsfinancing.Inadditionto individualandemployer contributions, the sources of finance could include general government revenues andearmarkedtaxes.

6

transitionsbetween formal and informal jobsare frequent. Therearealsoquestionsabout the fiscalsustainabilityoftheprogramsandwhetheritmakesensetosubsidizebenefitsforinformalworkerswhohave,atleastpartially,thecapacitytopay.Thisnotereviewstheeconomicliteraturedealingwiththeexpansionofsocialinsuranceprogramsandsummarizes themain policy insights. It draws on international evidence on social insurance systemdesign and innovations, and the resulting impact on coverage of and participation by formal andinformalworkers.The focus isongeneraldesignprinciples that canapply tounemploymentbenefits,healthinsurance,andpensions(old-age,disability,andsurvivorships).

Therestof thisnoteproceedsas follows.Section2discusses threedifferent typesof social insuranceschemes and argues that well-designed integrated programs are the best policy option. Section 3discusses the common implementation challenges and outlines alternative approaches to overcomethem, highlighting the role of digital technologies to promote take-up and retention, and the designoptionsthatwouldminimizepotentialdistortionaryeffects.Finally,Section4concludes.

7

BOX2:EFFICIENCYANDEQUITYGROUNDSFORGOVERNMENT-PROVIDEDSOCIALINSURANCE

Thereare important efficiencyand equityconsiderations to justify thepublicprovisionof social insurance.Theexistenceofsignificantmarketfailuresandindividuals’diverseriskaversionprofilesarekeyreasonsforgovernmentinvolvement.

InsuranceMarket Failures. Unemployment and health insurance are affected by adverse selection, whicharises because insurers cannot determine theunemployment andhealth riskswith sufficient accuracy, asworkorhealthhistoriesareimperfectandinsomecasesnonexistentpredictors(Vodopivec,2004).Low-riskindividuals do not have strong incentives to obtain insurance, skewing the pool of insured individuals tothose with higher average risk. Adverse selection therefore leads to under-provision of insurance.Unemployment andhealth insurance are also affectedbymoral hazard,which leads to over-utilizationofunemploymentbenefitsorhealthservices.Forexample,unemployment insurancemay reducework effortwhen employed and job search efforts when receiving benefits, and health insurance may lead to riskierbehaviors. Finally, someof the risks in question arehighly correlated across awide rangeof individuals.Morespecifically, somerisks, suchasunemployment,arecorrelatedacross individualsduring thebusinesscycle,making it difficult for private insurers to diversify risk. A similar issue arises in termsof uninsurablerisks,wherebysomesocio-demographicgroupsfaceriskssohigh that theyarepricedoutoftheinsurancemarket.Forinstance,thechancesofre-employmentforworkersclosetoretirementagemaybesolowthatprivateinsurancemaynotbeviable.Similarly,pre-existinghealthconditionsmaypreventsomepeoplefromqualifyingforprivatehealthinsurance.

Capitalmarketfailures.Imperfectcapitalmarketsrepresentanotherimportantimpedimentforindividualstoaccessprivatefinancingtocopewithashort-termshock,evenwhentheprobabilityofrepaymentishigh.Forinstance, individualswhobecomeunemployedbecause their skillsbecomeobsoletemay find itdifficult toobtainaloantofinanceeducationortraining,giventhedelayedreturnstohumancapitalinvestments.

Myopia. The existence of individual barriers to participation may result in insufficient demand and sub-optimallevelsofsocialinsurancecoverage. Inadequatelong-termplanningonthepartofindividualscanbeattributedtomanyfactors,includinglackofinformation,awarenessorfinancialskills,behavioralfactorsthatimpede individuals from accurately perceiving and weighing risks, or outright psychological barriers forlonger term planning (Holzmann, 2014). There is a large body of literature on the individual barriers toparticipation in pension schemes, where individuals may rationally postpone deciding on whether tocontributetoapensionfundbecausethecostofgatheringtheinformationneededtodecidecoupledwiththe complexity involved in evaluating the information may exceed the short-run benefit from doing so(Madrian and Shea, 2001). There are similar examples for unemployment and health insurance take-updecisions(Madrian,2014).

Poverty and liquidity constraints. Someof the failures that characterize insurancemarkets are particularlyburdensome for informalworkers, especially those forwhom informal employment is a last resort due tolack of alternatives. For example, unskilledworkers in precarious or irregularwork aremore likely to facecredit constraints, lack the resources tomakecomplexdecisions,orbeaffectedbyuninsurable risks.Theyarealsomorelikelytolive inimpoverishedcommunitiesorlaggingregionswithlimitedworkopportunities,andthereforearelessabletoself-insureagainstrisks.Moreover,problemsofadverseselectionarelikelytobeheightenedcomparedtoformalcounterpartswhosehealthor jobrecordsaremorereadilyavailableorverifiableforpotentialinsuranceproviderstomakeriskassessments.

8

2. INTERNATIONAL EXPERIENCE IN EXTENDING SOCIAL INSURANCE TO INFORMAL WORKERS

The performance of Bismarckian contributory social insurance programs has been disappointing indevelopingcountries.Giventhechallengesofextending itscoverageto informalworkers,several low-andmiddle-incomecountries introducednewsocial insuranceschemes tocover individualswhowork(permanentlyor intermittently) in the informalsectorandhave lowor limitedcapacity topay regularcontributions.Whilesomeof thesenewsystemshaveauniversaldesigntargetingworkers’economy-wide, such asSeguroPopular inMexico, others focuson a specific segmentof the economy, such asEcuador’sefforts toexpandcoverage in rural areas. Someprogramsare fully subsidized, requiringnoindividual contribution – for example, the “30 Baht” in Thailand – whereas others take a mixedapproach incorporatinga targeted subsidy for vulnerableworkers, andanactuarially fair contributionsystem for formal workers, such as the Chilean pension system.2 Social insurance systems thatincorporate informalworkers can be broadly sorted into three categories: non-contributory universalprograms,parallelschemes,andnationallyintegratedprogramswithexplicitsubsidies.

Non-contributory universal programs include social pensions,3 universal health insurance andunemploymentassistancewhereeligibilityisnotbasedonpastcontributionsandisinsteadfinancedoutof general taxation. While social pensions have been in place for a long time, they expandedsubstantially since the 1990s (Palacios and Knox-Vydmanov, 2014). According to the HelpAgeInternational Social Pensions Database, 110 countries have a social pension, albeit some with lowcoveragerates(Figure2).Forinstance,NewZealand’spensionsystemisbasedaroundasocialpensionforthoseaged65yearsorolder,whichisfinancedfromgeneralrevenues.Anotherexamplecomesfromthe National Health Service (NHS) of the United Kingdom, which provides public healthcare to allpermanentresidents,anditispaidoutofgeneraltaxation.

2 While the contributory part of the Chilean pension system is fully funded by workers’ contributions, the government isresponsible for financing the Basic Solidarity Pension for persons who did not contribute, and the Pension SolidarityComplementforpersonswhosecontributionsarebelowathreshold.Thegovernmentalsoguaranteespensionbenefitsincaseswhereapensionfundisshutdownorbecomesinsolvent(Shelton,2012).3 Broadlydefined, social pensions are cash transfers not linked to contributions thatoccur after retirementor after a giveneligibilityage(Holzmannetal.2009).

9

Figure2SocialPensionCoverageacrosstheWorld

Source:BasedondatafromHelpAgeInternational,SocialPensionsDatabase

Parallel schemes include those where several fragmented contributory systems coexist, such as inTunisia,wheretherearedifferentpensionprogramsforpublicandprivatesectorworkers.Evenwithintheprivate sector,workers havemultiple schemes that dependonwhere theywork (Robalino et al.,2012).ParallelprogramsalsoincludecaseswhereaBismarckiansystemcoexistswithnon-contributoryprogram, where eligibility for the former depends on labor market status and workers cannot beincludedinbothsimultaneously,asitisthecasewithMexico’shealthinsurancesystem.Mexicoin2002introduced Seguro Popular, a large program designed to provide health insurance to all workers notaffiliatedwithsocialsecurityandthatcoversapre-determinedsetofproceduresandtreatments(Pages,RigoliniandRobalino,2014).WhereasMexico’sformalsocialsecuritysystemprovidesbettercoverage,thedifferenceinbenefitsbetweenthetwoprogramshasnarrowedovertime.Moreover,theprogramisvirtually freeof charge tousers,with costsborneby thegovernment.Overall, theprogramhasbeenveryeffectiveat increasinghealthcoverage,andreducinghealthexpendituresforMexicanfamiliesbyan amount equivalent to 4.2 percent of the household budget (Barros, 2009). The program has alsoreducedinfantmortalityratesbyabout0.5percentagepoints,asignificantimprovement(Pfutze,2014).

10

Integrated national programs. Some governments have reinforced the solidarity mechanism byintegrating a Bismarckian system and new social insurance schemes aimed at including informalworkers, intoasinglesystemforbeneficiariesacrosstheincomespectrum.Chiledidthisbyreformingitspensionsystemin2008 inresponsetothefact thatmany individualsdidnotcontributefrequentlyenoughtothesystem,particularlyamonggroupswithlowformallabormarketattachment(Attanasioetal.,2011).Toincentivizeworkerstojointheformalsectorandraisecontributions,thereformreplacedthe redistributive component of the pension system with two new components: one consists of ameans-testedpensionforthosenotentitledtothecontributorypensionbenefit;theotherisawelfarepension complement to sustain consumption by topping-up the contributory pension for those withinsufficient contributions. In addition, the new system introduced incentives to encourage theparticipationinthecontributorysystemforgroupswithhistoricallylowattachmenttotheformallabormarket,suchaswomen,youngworkersandtheself-employed.Thereformincludedsubsidiestowomenforeachchildtheyhave,awageandcontributionsubsidytoyoungworkers,andpensioneligibilityfortheself-employed(includingeligibilityfortheredistributivecomponent).

Parallel and universal non-contributory systems face different sets of challenges. Universal non-contributory systems can add important fiscal pressures because benefits are typically not linked torevenues. Parallel systems can create implicit subsidies or taxes asworkersmove between jobs thathave different formality status and are thus covered by different social insurance schemes. They aremorepronetoberegressiveanddistortivewhentheyarehighlyfragmented.

An integrated systemcan, in principle, address the limitations inherent toparallel anduniversal non-contributory schemes by being more financially sustainable and not exacerbating labor marketdistortions that constrain labor mobility. Integrated systems do have an income effect, sincebeneficiaries receive larger or smaller subsidies depending on their earnings; and they may still bedistortionary, if the income threshold to qualify for subsidies is high enough to incentivize earlyretirement or reduced job-search effort. The next section provides some guidelines for an effectiveimplementationofanintegratedsocialinsurancesystem.

11

BOX 3: THE TRADE-OFFS BETWEEN UNIVERSAL SOCIAL INSURANCE AND INCREASING FORMALWORK

Despite the equity and efficiency arguments in favor of extending social insurance to informalworkers, legitimate concerns arise around the potentially distorting effects on labor supply andlabordemand.

Social insuranceprogramscanaffect labor supplyanddemanddecisionswhen theyare linked tothe labor market, either through eligibility requirements or financing mechanisms. When socialinsurance is financed through payroll taxes (paid by employers) and workers’ contributions, thewedge between the cost of labor and take-home pay increases. This could reduce formalemployment,especiallyinlowandmiddleincomecountrieswithlowenforcementcapacity.

Themagnitudeoftheimpactonformallabordemandandsupplywilldependontheperceivedsizeofthiswedge,not justtheactualsizeofthewedge.Perceivedtax-wedgesare largerwhenpeopledonotvaluethebenefitsofferedbytheschemeandtherebywouldprefernottoparticipateintheprogram,orprefer tocontribute less.Theseperceptions regarding thevalueofa social insuranceprogramarepartiallydrivenbya rationaland informedprocess, suchaswhen individualsdonotvalue the systematic and often implicit redistribution mechanisms included in risk-poolingarrangements.Theperceivedwedgemayalsoincreaseiftheservicesprovidedareofpoorquality,ifthere is excessive“bundling”ofbenefits thatmanyparticipantsdonot value,and if contributionsare not linked to the expected benefits. The perceived wedge is also driven by myopia. Myopiaincludescaseswhereindividualsfaceinformation,awarenessand/orpsychologicalbarriers,or lackthefinancialskillsforlong-termplanning.

Alargetax-wedgemayprovideincentivesforemployerstohireworkersinformallyortohirefewerworkers,becauseitbecomesmoredifficulttopassonthecostofinsurancetoworkersintheformoflowerwages.Alargetax-wedge–perceivedorreal–canalsoprovideincentivesforworkerstotake informal jobs. Moreover, it can interact with implicit or explicit subsidies offered by socialinsurancetoinformalworkers.Explicitsubsidiesgiventoworkersintheinformalsectorintheformof non-contributory benefitsmayprovide incentives towork informally, becausewhereas formalworkerspayforthesebenefits,informalworkersgetthemforfree. Implicitsubsidies–intheformof “above market” rates of return on contributions, minimum pension guarantees, minimumunemploymentbenefits,orearlyretirementprovisionsthatarenotactuariallyfair–canalsodistortincentivesandencourageearlyretirement,shorterenrollmentperiods insocialsecurityorreducejob-searcheffort.

The literature provides significant evidence that extending social insurance to informal workersdistorts labor supply decisionsand laboroutcomes in theaggregate (seeFrölich etal.2014). It isneverthelessimportanttorecognizethatitisimpossibletodesignsocialinsurancemechanismsandprotectworkerswithoutchanging economic incentives (BarrandDiamond, 2006).Thepotentiallydistortive effects of extending social insurance to informalworkersmust beweighed against theassociatedwelfareandeconomicbenefits.

Source:Pages,RigoliniandRobalino(2014)

12

3. DESIGN, IMPLEMENTATION AND POLICY OPTIONS

Introducing a social insurance system for informal workers is fraught with challenges, ranging frompractical issuesofenrollingandregisteringparticipants,togettingtheincentivesrighttocontributeorsave, and not exacerbating labor market distortions (see Box 3). Identifying eligible participants iscomplicated by the fact that informal work is mainly unorganized and takes place in a variety ofsometimes changing locations or within homes. In the case of self-employed or seasonal workers,maintaining up-to-date and accurate records is administratively complicated due to intermittent andirregularworkfordifferentemployers,andirregularityofincomes.Andgivenworkers’myopiaandtimepreferences for spending versus saving or self-insuring, incentivizing informal workers to contributevoluntarily isdifficult.Thesefactorsareobstaclesforregisteringbeneficiaries,attractingenrolleesandcollecting contributions (VanGinneken, 1999,HolmesandScott, 2016). Furthermore, establishing thelevelofbenefitsasareplacementrateofincomecannotbeeasilydeterminedinmostcases.Andunlikehealth status,employment status isnotdirectlyverifiableas individuals cannoteasilyprove that theyareunemployedwhentheymightbeoutofthelaborforceorworkinginformally.Toovercomethesechallenges,itiscriticaltodelinkaccesstosocialinsuranceprogramsfromthelaborcontract.Allworkerswouldhaveaccesstothesamesocialinsuranceprogramregardlessofworkstatusor sector.Withina single integrated system, rather than in twoparallel systems, contributionswouldremainmandatoryforformalworkers(andtheiremployers,whenavailable),andthesystemwouldbeopen to informal workers on a voluntary basis. To this end, informal workers would need to beidentified and enrolled, and there would be a system of financial and non-financial incentives topromote contributions and savings. Within this broad framework, this note summarizes someof themainissuesrelatedtodesignandimplementation(seesummaryinTable2below).

3.1 REGISTERINGANDENROLLINGWORKERSAkeyfactordrivingworkersintoinformalityaretheburdensomerequirementsforparticipatinginsocialinsurancesystems.Inadditiontothehighcost,paymentprocessesandbureaucraticprocedurescanbeverycomplex.Advancesintechnologycanbeeffectiveinspurringtheenrollmentofinformalworkers,startingwithrobustIDsystems.3.1.1DigitalIDsystemsIdentifying informalworkers is a pre-condition for enrollment in social insurance programs. New IT-basedsystemscangreatlyfacilitatetheprocess.Forinstance,India’smassiveAadhaarprogram,whichhassofarenrolledover830millionpeople,replacedthephysicalIDcardaltogether(WorldBank,2016).OthercountriesdevelopingdigitalIDsystemsincludeBangladesh,KenyaandGuinea.

13

DigitalIDsareinprincipleuniversalandtherebycoverbothformalandinformalworkers.Inadditiontofacilitatingservicedelivery,digital IDsystemswereeffective inreducing leakages inbenefits forsocialprotection programs, health insurance and pension schemes stemming from, for example, duplicateclaims,ghost-workers,quasi-ghosts,andcorruption(WorldBank,2016).Digital technologies can also help improve coordination and consistency across concurrent socialprotectionsystems.Since2005,themainsocialsecurityinstitutionsinthePhilippinesreliedonaunifiedsmart card that can store information and perform transactions in all these institutions. The systemenables authorities to identify participants, and facilitates and streamlines implementation ofproceduresandmonitoring(Duran-Valverdeetal.,2013).3.1.2FacilitatingenrollmentandfinancialtransactionsReducing the time required to enroll in a social insurance scheme is an important way to expandcoveragetoinformalworkers,forwhomtimeawayfromworkcanmeansignificantforegoneearnings.Nicaraguaintroducedaprogramin2007thathighlightstheimportanceoftimetoinformalworkers.Thisprogram extended voluntary health insurance to informal workers by randomly allocating monetarysubsidies for enrollment, implementing an aggressive dissemination strategy and reducing the timerequiredtoenrollbyallowingenrollmentonthespot.Theresults indicate that timeandconveniencecosts matter almost as much as monetary subsidies (Hatt et al. 2009). Whereas these factors weresuccessful inpromotingtake-up,theydidnotcontributeto long-termretention.Otherfactorssuchasthe inconvenienceofmakingpayments, thepricingofpremiumsandthehigherperceivedqualityandconvenience of private medical providers were key reasons behind low retention rates. Addressingtheseconsiderationswillbecrucialforsustainingparticipationovertime.

Newadvancesintechnologycanalsohelpfirmsformalizetheirworkers.Forinstance,theMbaomicro-pension scheme in Kenya,which covers small- andmedium-sized firms, usesmobile transfer servicessuchasM-PESAandAirteltofacilitatepayments(HolmesandScott,2016).

Encouraging the enrollment of informal workers may require establishing new institutions andinfrastructure, but relying on existing infrastructure to the extent possible is likely to be both moreeffectiveandlesscostly.HuandStewart(2009)citethecaseofIndia,wherethegovernmentplannedtoextend the existingmandatory pension scheme for government officials into a newNational PensionScheme(NPS)thatwouldbevoluntaryforinformalworkers.Thefinancialsectorinfrastructurerelatedto theNPSwas designed to be aswide-ranging as possible by utilizing existing infrastructure (banks,postoffices,etc.),tofacilitateaccessbyinformalworkers,particularlyinremoteruralareas.

Given the differences in characteristics of informal and formal workers and their distinct earningpatterns, any attempt to foster the enrollment of informal workers will require new institutionalarrangements (Ribe, Robalino andWalker, 2012). In the first place, becausemandatory contributionscannot be imposed or enforced on most informal workers, participation must be voluntary andthereforeattractive.Secondly,governments introducingnewsocial insuranceschemeswillneedtobeproactiveinmarketingandincollectingcontributions,suchasthroughusingmobileagenciesoperatinginstreetmarketsorremoteareas.Thirdly,transactioncostsmustbelowgiventhelimitedresourcesofinformalworkers.Someoftheseproblemscouldbeaddressed,inpart,byoperatingthroughfacilitating

14

agenciesthataggregatethecontributionsoflow-incomeworkersintheinformalsector.Explicitfinancialincentivesmightberequiredtoattractindividualswithlimiteddisposableincomeorsavingcapacity.

Relying on financial agencies such asmicrofinance institutions (MFIs)which aremore focusedon theinformalsector,couldbeagoodstrategy.Forexample,theGrameenBankinBangladeshstartedtoofferaproductforold-ageprotectionbeginningin2000,wherebyborrowerswererequiredtomakeasmalldepositeachmonth intoapersonal savingsaccount.A similarplanwas introduced in thePhilippines,where individuals canmake contributions at a level as low as USD 0.12 perweek. Counter evidencecomes fromaprogram inNicaragua thataimed toextendhealth insurance to informalworkersusingsubsidiesandMFIs;itshowedthatthelatterwerelesssuccessfulthanthecentralagencyatencouragingenrollment(Hattetal.2009).4

3.2 LINKINGCONTRIBUTIONSTOBENEFITSThelevelofcontributionsandbenefitswillvaryacrosscountries,reflectingsocialpreferencesregardingthe balance between individuals’ and governments’ responsibilities (i.e., the social contract). Threeessentialcriteriaforsettingbenefitsare:adequacy,efficiency,andaffordability(Ribeetal.2012).Morespecifically,benefitswouldbehighenoughtoallowadecentstandardoflivingandpreventindividualsfromfallingintopoverty(adequacy).However,theywouldnotbesetsohighthatindividualswouldbediscouragedfromsaving,workingorlookingforajob(efficiency),orthatwouldplaceanunsustainablestrainonpublicfinances(sustainability).Theselevelsaredifficulttodefineandwilldependoncertainkeyfactorssuchasthelevelofeconomicdevelopment, demographic structure, labor productivity, income distribution, efficiency of the taxsystem, and level of informality. In the case of pensions and unemployment insurance, the keyparametersfordefiningthelevelofbenefitsarethereplacementrate,theminimumbenefitlevel(awaytoincreasethereplacementrateforlowincomeworkers),andaceilingoncoveredearnings(toreducethe replacement rate for high earners). In the case of health insurance, it is important to find thebalance between the goal of improving health outcomes and protecting people against the financialconsequencesofillness.Themainparameterstobedefinedarethelistofhealthinterventionsthatarecoveredandnotcoveredbythesystem,andthe levelofcost-sharingthroughhealthplandeductiblesandcopaymentsfortreatmentstobepaidbythe individual.Thesecost-shareratioscouldvarybythelevelofthebeneficiary’sincome.

4Hattetal.(2009)suggestthatpeoplehaddoubtsabouttheexpertiseofMFIstodealwithhealth-relatedissues,andothersweresuspiciousthattheMFIswouldtrytoprofitfromthearrangement.

15

Table1Keyparametersofold-ageandunemploymentinsurancesystems

Typeofbenefits ParametersOld-age Replacementrates:inarisk-poolingscheme,pensionswouldbe

higherthan40to50percentofpreviousearningsafter30yearsofcontributions(ILO,2012).Theminimumwouldbebetween20to25percentofeconomy-wideaverageearnings.Theceilingwouldbe60percentofearningsforthesystemtobefinanciallyviable(HolzmannandHinz,2005),or2to2.5timestheeconomy-wideaverageearnings.

Unemployment Replacementrates:50to70percentofcoveredearnings.Theminimumcouldbeequaltotheminimumwage.Theceilingcouldbeequalto2.5to3.5timestheeconomy-wideaverageearnings(RobalinoandWeber,2013;Kuddoetal.,2015)

Itisalsoimportanttoreduceuncertaintyabouthowbenefitswillevolveovertime(Ribeetal.2012).Toavoidanydistortionary impactsof inflation, for instance,minimumpensionsandcontribution ceilingscanbeestablishedasafractionofeconomy-wideaverageearnings,andbenefitscanbeautomaticallyindexedtoinflation.Percapitabenefitsneedtohaveanexplicitfinancingmechanism;acombinationofindividual contributions (or contributions from employers when available) and government transfersfinancedfromgeneraltaxation.In thecaseofpensions, thecontribution rate (expressedasa shareof coveredearnings)needs tobelinked to the accrual rate and life expectancy at retirement. For unemployment benefits thecontributionrateisusuallysetatalevelwhereitgeneratessufficientrevenuestopaybenefitsduringagivenperiod(e.g.,ayear).5Forhealthinsurance,contrarytostandardpractice,thecontributionwouldtaketheformofapremiumthatreflectstheexpectedcostofthehealthinsurancepackage.

3.3 DESIGNINGREDISTRIBUTIVEARRANGEMENTSRedistributioniscriticalwithinthesocialinsurancesystemtofundcontributionsand/ortopupbenefits(e.g., minimum pension) of low income individuals. A proper redistributive arrangement involvesexplicit subsidies/transfers and an eligibility criteria, as well as dedicated taxes. Subsidies would bebased on earnings or savings capacity, and not on workers’ occupation, economic sector, oremployment in the formalor informal sector (Ribeet al, 2012). Theamountof the subsidywouldbekeptbelowacertainthresholdtoavoidanyadversebehavioraleffectssuchasearlyretirementorlowjobeffort.Eligibilityforsubsidiescanbeuniversalormeans-tested.Universalbenefitshavemorepredictablecosts,andarelessexpensivetoadminister.Ontheotherhand,giventhemagnitudeofinformalemploymentindevelopingcountries,extendingsocialinsurancethroughuniversalbenefitsrisksunsustainablefiscalcosts,orbenefit levels thataretoo lowtomakeasignificantdent inpoverty (GroshandLeite,2009).

5Robalinoetal.(2009)

16

TheexperienceofBolivia’suniversal income transfer illustrates thepotential forunintendednegativeconsequences.Theprogramisexpensiveandbenefitsmanymiddle-orhigh-incomeindividuals,sincenorestrictionswere imposed on those already receiving benefits from the contributory pension system,whilemanyofthepoorestindividualsareexcludedduetoadministrativeproblems.Anotherdownsideofuniversalbenefitsisthepotentialtocrowdoutcontributionsbyindividualswithsavingcapacity.Thetotalcostsofmeans-testedbenefitsmaybelowerinitially,buttheycouldincreaseinthefaceofpoliticalpressureforlessstringenteligibilitycriteriaormoregenerousbenefits.Evenmeans-testedbenefitscanaffectincentivesforworkersclosetotheeligibilitythresholdwhomaydecidetoworkinformallyorworkfewerhourstoavoidthecontributionand/orretainthesubsidy.Ribe,RobalinoandWalker(2012)recommendmeans-testingforanysubsidieswithinasocialinsurancesystem, and downward adjustment to subsidy levels as beneficiary income rises, with the subsidycomponentfinancedbygeneraltaxrevenues.Tominimizethepotentialdistortionaryeffectsofmeans-tested benefits, itwould be important to keep the size of the transfermodest and avoid setting theincome threshold for eligibility too high (to minimize the number of potential beneficiaries). Thisarrangementwouldleadtoamoreprogressiveincomeredistributionandavoiddistortingtheincentivesfacingworkersandemployers.

In terms of financing mechanisms, it is recommended not to rely on labor taxes that can reduceincentives tocreateor take formal jobs. Thealternative is to relyongeneral revenues fundedoutofincome taxes, consumption taxes, or property taxes. Minimumpensions, for instance,would have aseparatefundingmechanism;theircostwouldnotbeincludedinthecalculationofthecontributionratedescribedintheprevioussection.Thesameappliestotheshareofthehealthinsurancepremiumthatissubsidizeforagivencategoryofworkers.

In thecaseofunemploymentbenefits thingscanbemorecomplicated. In traditionalunemploymentinsurance systems redistribution is financed, implicitly, through taxes on the savings of “low risk”workers;i.e.thosewhoareatalowriskofunemployment(RobalinoandWeber,2013).Ifthistaxwaseliminated, theUI systemwouldbecomeapureunemploymentsavingsaccount system(UISA)whereredistributionwouldneedtobefundedoutofgeneralrevenues.Essentially,thesewouldbetransferstotop-upthecontributionsof“high-risk”workerswhoareunabletosaveenoughontheirowntofundthebenefitstheyreceive.Inpractice,theoptimalfinancingarrangementsarelikelytoinvolvebothtaxesonsavingsandgeneralrevenues.

3.4 PROVIDINGINFORMATIONANDINCENTIVESTOCONTRIBUTE/SAVEWhensocialinsurancesystemshaveavoluntarycomponent,itiscrucialtoremoveinformationbarriersand strengthen the incentives to contribute. These goals can be achieved through communicationscampaigns, financial literacy programs, and by introducing financial and non-financial incentives tocontribute.

17

3.4.1CommunicationscampaignsandfinancialliteracyLow levelsofeducationor literacy canbea fundamentalbarrier toparticipation ina social insurancesystem.InGhana,forexample,educationalattainmentisasignificantdeterminantofenrollmentintheexpanded health insurance program. This suggests that program information would have to bedisseminated ina varietyofways that can reach thosewith littleornoeducation (HolmesandScott,2016). The Self-Employed Women’s Association (SEWA) in India has adopted several innovativeapproachestohelpmemberssubmitclaims,whichisachallengeforpoorindividualswithloweducationorwheresubmissionrequirestraveltoanothercity(HolmesandScott,2016).Theseinnovationsincludetheuseofextensionagentsaswellasabarcodescannersystemthatallowsilliteratememberstomakea claim by attaching a sticker to a prepaid envelope and sending it to themicro-insurer. Education-related barriers can be large in developed countries as well. In the United States, information andadministrative costs are estimated to be significant obstacles to enrollment in the public healthinsurance system (Medicaid), especially for Hispanics and Asians (Aizer, 2007), implying the need fortargetedoutreach.

Targetingthepopulationofinterestcanhavelargeimpactsoncoverage.ThisstrategywaskeyforCapeVerdetoincreasethesocialsecuritycoverageofindependentworkersfromzeroto9percentinaboutayear (Duran-Valverde, 2013). The government took a pro-active approach consisting of targetedcommunications campaigns, together with a set of education and awareness-raising activities at thelocal level, both in urban and rural areas. To get closer to independent workers, the social securityinstituteopenedservicecentersinlocationswithahighconcentrationofindependentworkers.

Bhargava andManoli (2012) provide empirical evidence that better dissemination of information oneligibility–suchasthroughimprovingtheexplanationofpotentialbenefits–canhaveapositiveimpactontake-up.Simplification isanotherapproachthathasbeensuccessfullyappliedto increasecoverageandcontributionlevels;Madrian(2014)suggeststhatreducingthenumberofprogramoptions(e.g.,setcontributionlevels,limitedchoiceofinvestmentvehicles),andincorporatingdecisionaids,personalizedinformation and standardized optionswould help encourage the participation of informalworkers involuntarysocialinsuranceprograms.

3.4.2Financialincentives

Transfers to subsidize workers with limited saving capacity can take the form of ex ante matchingcontributionsorofadditionalexpostnoncontributorybenefits.Socialpensionsareanexampleofanexpostbenefit.Unemploymentinsurancecantaketheformofanexantetransfer(ifgovernmentmatchesindividual contributions to a savings account) or ex post (if the government pays subsidizedunemploymentbenefitswhen the savings account runsout). The costs of serviceswouldbepre-paidratherthanreimbursedtotheuser,toencouragecontributions.Ex-ante transfers tend to produce better incentives to contribute and may cost less than ex posttransfers, at least in the case of unemployment insurance andpensions (Ribe et al., 2012). Themainadvantageofmatchingcontributionsforunemployment insuranceisthat itgivesworkersanincentiveto work. While there is empirical evidence for the United States on the effectiveness of matchingpensioncontributions,thereisnosolidevidencefordevelopingeconomies.

18

3.4.3Non-financialincentivesNon-financial nudges such as sending personal reminders can be effective in incentivizing savingbehavior. In a recent example from Kenya, Akbas et al. (2016) implemented an experiment amongparticipants of theMbao Savings Plan, which resembles a regular bank account with a commitmentdevice,thatis,athree-yearrestrictiononwithdrawalsafterregistering.Amongotherinterventions,theyrandomly sent a “coin” to the treated group at the start of the intervention and each week, andparticipantswereaskedtokeeptrackoftheirownweeklydepositstothesavingsaccount.Theresultsshowthat this intervention incentivized themostsavings,at levelsdouble those in thecontrolgroup.This type of intervention was also the least costly, especially when compared to a less effectivealternative involvingmatching contributions. Kast et al. (2012) find that simply announcing a savingsgoalandthenregularlyreportingtopeersthelevelsofachievedsavingsalsoincreasesavingrates.Thisevidence highlights the potential for interventions based on behavioral economics to promoteindividuals’contributionsatalowercostcomparedtofinancialincentives.

3.5 MONITORINGANDENFORCEMENTAdequate monitoring and enforcement capacity is crucial for successful implementation of a socialinsuranceprogram.Insomecases,poorcapacitymeantthatrelativelycomplexprogramcomponentsdonotget implemented,often leadingtotheexclusionof thepooresthouseholds.For instance,Ghana’shealth insurance program exempts poor people from paying premiums, but administrators facechallengesinimplementingthemeanstest,suchthatinpractice,theysetthepremiumataflatrateatthedistrictlevel,whichmayexceedthelevelaffordablebythepoorest(AlatingaandWilliams,2014andBrugiaviniandPace,2016,citedinHolmesandScott,2016).

Increasing enforcement may improve the incentives for firms and individuals to contribute to thesystem.However,itmayalsohavesomeunintendedconsequencesbyraisingtheincentivesforworkerstobeintheinformalsectorifearningsaretoolowtoaffordcontributionstothesystem(AlmeidaandCarneiro, 2012). Moreover, if inspections are inefficient or affected by corruption, firms andindependent contractors may find new ways to remain undetected by authorities. Lessons fromtransitioneconomiesindicatethatmakingthelaborinspectoratesandtaxauthoritiesmoreefficientandprofessionalcanfosterhighercontributionstothesystem(Cordova-NovionandSahovic,2010).Thiscanbeachievedthrough,forexample,makinginspectorsmoreaccountable,increasingthetransparencyofinspections(byestablishingtimelimitsonvisitsandrequiringwrittenvisitsummaries),andrandomizingthe assignment of inspectors to firms. This could be complemented by efforts to improve the hiringprocessofinspectors,andinvestingintrainingtodevelopamoreclient-orientedservicemindset,whichwouldenhancefirms’willingnesstocooperate.

19

Table2SummaryofDesignOptionsforSocialInsuranceSystemParametersDesignOptions Opportunities Risks

Registeringworkersandfacilitatingpayments

Digitaltechnologies - Improveaccessandadministrativeefficiency - Greaterexclusionifthesystemrequiresdigitalliteracy

Streamliningregistrationprocedures - Encouragingtake-upamongworkerswhosetimeawayfromworkmeansforegoneearnings

- Costsavingswhencomparedtoothermethods(e.g.,matchingcontributions)

- Toomuchfocusonregistrationmaydistractfromretentionefforts.

Relyingonnon-governmentagenciesforoutreachorpaymentcollection

- Moretake-upamonghard-to-reachgroups - Highermonitoringcosts- Noteffectiveifpotentialbeneficiariesmistrustthese

organizationsDefiningBenefitsandCosts

Levelofbenefitsandcosts - Findtherightbalancebetweenadequacy,efficiencyandsustainability

- Benefitswouldbelinkedtocontributions

- Excessivelyhighbenefitscanadddistortionsandmakethesystemunsustainable

- Toolowbenefitsmaynotpreventpoverty

DesigningSubsidiesUniversalbenefits - Morepredictablecostsandlessexpensivetoadminister - Risksofleakagetoricherhouseholds

Mean-testedbenefits - Moreprogressiveredistributionandlessdistortiveincentives - Politicalpressuretoexpandcoveragecouldresultinfiscalpressures

- Higherdistortiveeffectsifthetransferandthresholdsarehigh

Implicitsubsidies - Labormarketdistortions

ProvidingInformationandIncentivestoContribute/Save

Communicationscampaignsandfinancialliteracy - Greaterawarenessaboutthecostsandbenefitsoftake-up- Improvingtake-upamonghard-to-reachgroups

- Thecostsandbenefitsoftargetedoutreachtosmallgroupsofpotentialbeneficiariesshouldbeweighedagainstthoseofuniversaloutreach.

Financialincentives - Promotecontributionsamonglowerincomegroups - Financiallyunsustainable

Non-financialincentives - Improvingtake-up- Costsavingswhencomparedtofinancialincentives

MonitoringandEnforcement

Bettermonitoring - Increasinginclusionandefficiency

Greaterenforcement - Increasingregistrationandretention - Increasedincentivesforfirmstoremainintheshadows,ifinspectionsarecorruptorinefficient;oriftheyareunabletoaffordcontributions.

20

4 CONCLUSIONS

Social insurance can be an effective tool for addressing income shocks to workers by providing for

consumption smoothing througha risk-poolingapproach.Whereasmost countrieshave some typeof

social insurancesysteminplace,fewareefficientorequitablebecauseoftheirpartialcoverageofthe

labor force. This is particularly true in developing countries with a large shadow economy. The

traditional contributory design of these programs was appropriate fifty years ago in the highly

industrializedmiddle-andupper-incomeeconomieswheremostworkerswereformallyemployed.But

this is not the reality in developing economies today.Moreover, theongoing transformationof labor

marketsinadvancedeconomieswhereoccupationalmobilityandinformalityarebothrisingmeansthat

traditionalsocialinsuranceprogramsarelosingrelevance.

Aspolicymakersconsiderhowtoextendthecoverageofsocialinsurancetoinformalworkers,thereare

key lessonstobedrawnfromexperience. Integratedsocial insurancesystems(suchasChile’spension

system) that combine an actuarially fair contributory systemwith explicit subsidies for the poor and

informal seem to be more financially sustainable than universal non-contributory systems, and less

distortionary than fragmented parallel schemes. Within an integrated system, there are key choice

parameterstominimizeunintendedimpactsofsocialinsuranceonlaborsupplyanddemand,providean

adequatelevelofbenefitsandimprovefinancialsustainability.Theseparameters includetheeligibility

criteria,thelevelofcontributionsandbenefits,andthemixoffundingsources(contributionsvs.general

taxation).

Topromotetake-upandefficiency,contributionswouldbelinkedtotheindividualandnottothejob.

Beneficiaries need to be aware of the contribution requirements for each benefit level. And if social

insurance includesany redistributionmechanism, itwouldbe transparent; that is, theprogramwould

makeclear theamountof the subsidy, theeligibility criteria, andhow it is financed. Finally, itwill be

important to adapt this general framework to the specific needs of a country’s informal workers to

ensureasocialinsurancesystemthatisinclusive,fair,efficientandeffectivelyimplemented.

21

REFERENCES

Aizer,Anna."Publichealthinsurance,programtake-up,andchildhealth."TheReviewofEconomicsand

Statistics89.3(2007):400-415.

Akbas,M.,Ariely,D.,Robalino,D.A.,&Weber,M.(2016).HowtoHelpPoorInformalWorkerstoSavea

Bit:EvidencefromaFieldExperimentinKenya.

Alatinga,K.A.andWilliams,J.J.(2014)‘DevelopmentpolicyplanninginGhana:Thecaseofhealthcare

provision’,EuropeanScientificJournal10(3):359–82.

Almeida,Rita,andPedroCarneiro."Enforcementoflaborregulationandinformality."AmericanEconomicJournal:AppliedEconomics4.3(2012):64-89.

Aterido,R.,M.Hallward-DriemeierandC.Pages.(2011).“Doesexpandinghealthinsurancebeyond

formal-sectorworkersencourageinformality?MeasuringtheimpactofMexico’sSeguro

Popular”,PolicyResearchWorkingPaperSeries5785,TheWorldBank.

Attanasio,Orazio,CostasMeghir,andAndresOtero.FormalLaborMarketandPensionWealth:Evaluatingthe2008ChileanPensionReform.Mimeo.DepartmentofEconomics,UCL,2011.

Barr,A.,andT.Packard.2004.“Preferences,constraints,andsubstitutesforcoverageunderPeru's

pensionsystem”,backgroundpaperforregionalstudyonsocialsecurityreform.Washington,

DC:WorldBank(mimeo).

Barr,Nicholas,andPeterDiamond."Theeconomicsofpensions."Oxfordreviewofeconomicpolicy22.1(2006):15-39.

Bertrand,Marianne,SenhilMullainathan,andDouglasMiller."PublicPolicyandExtendedFamilies:

EvidencefromPensionsinSouthAfrica."WorldBankEconomicReview17.1(2003):27-50.

Bhargava,Saurabh,andDayManoli."Whyarebenefitsleftonthetable?Assessingtheroleof

information,complexity,andstigmaontake-upwithanIRSfieldexperiment."NA-AdvancesinConsumerResearchVolume40(2012).

Bosch,Mariano,andWilliamMaloney."CyclicalMovementsinUnemploymentandInformalityin

DevelopingCountries."(2008).

Bosch,Mariano,AnaBelenCobachoandCarmenPages.“EffectsofNon-ContributorySystemson

Informality:TakingStockofEightYearsofImplementationofMexico'sSeguroPopular”,chapter

inSocialInsurance,Informality,andLaborMarkets:HowtoProtectWorkersWhileCreating

GoodJobs,byMarkusFrölich,DavidKaplan,CarmenPages,JameleRigolini,andDavidRobalino

(eds),OxfordUniversityPress,2014.

Brugiavini,A.andPace,N.(2016)‘ExtendinghealthinsuranceinGhana:EffectsoftheNationalHealth

InsuranceSchemeonmaternitycare’,HealthEconomicsReview6:7.doi:10.1186/s13561-016-0083-9

22

Cho,Y.,Margolis,D.N.,Newhouse,D.,&Robalino,D.(2012).Labormarketsinlowandmiddleincome

countries:trendsandimplicationsforsocialprotectionandlaborpolicies.WashingtonDC:WorldBank,WorldBankSocialProtectionDiscussionPapers,(67613).

Cordova-Novion,Cesar,andTarikSahovic,"InspectionReform-DoModelsExist?".TheWorldBank

Group,2010.

DuránValverde,F.,FloresAguilar,J.,OrtizVindas,J.F.,MuñozCorea,D.,deLimaVieira,A.C.,&Tessier,

L.(2013).Innovationsinextendingsocialinsurancecoveragetoindependentworkers:

ExperiencesfromBrazil,CapeVerde,Colombia,CostaRica,Ecuador,Philippines,Franceand

Uruguay.ESSDocument,(42).

Encina,Jenny."PensionreforminChile:Adifferenceindifferencematchingestimation."Estudiosdeeconomía40.1(2013):81-95.

Galiani,Sebastian,PaulGertler,andRosangelaBando."Non-contributorypensions."LabourEconomics38(2016):47-58.

Gasparini,Leonardo,andLeopoldoTornarolli."LaborinformalityinLatinAmericaandtheCaribbean:

Patternsandtrendsfromhouseholdsurveymicrodata."DesarrolloySociedad63(2009):13-80.

Gindling,T.H.,andDavidNewhouse."Self-employmentinthedevelopingworld."WorldDevelopment56(2014):313-331.

Grosh,Margaret,andPhilippeLeite."Definingeligibilityforsocialpensions:Aviewfromasocial

assistanceperspective."ClosingtheCoverageGap:TheRoleofSocialPensionsandOtherRetirementIncomeTransfers(2009):161-86.

Hatt,L.,Thornton,R.,Magnoni,B.,&Islam,M.(2009).Extendingsocialinsurancetoinformalsector

workersinNicaraguaviamicrofinanceinstitutions:Resultsfromarandomizedevaluation.

HelpAgeInternational.2017.Socialpensionsdatabase—http://www.pension-watch.net/about-social-

pensions/about-social-pensions/social-pensions-database/[AccessedonJune122017]

Hirose,KenichiandMiloslavHettes.“Extendingsocialsecuritytotheinformaleconomy-Evidencefrom

BosniaandHerzegovinaandtheRepublicofMoldova”(2016).ILO.

Holmes,Rebecca,andLucyScott."Extendingsocialinsurancetoinformalworkers:Agenderanalysis."

(2016).ILO.

Holzmann,RobertandRichardHinz(2005)Old-ageincomesupportinthe21stcentury:Aninternational

perspectiveonpensionsystemsandreform.WashingtonD.C.:TheWorldBank.

Holzmann,Robert,DavidA.Robalino,andNoriyukiTakayama,eds.Closingthecoveragegap:Roleof

socialpensionsandotherretirementincometransfers.WorldBankPublications,2009.

Holzmann,Robert."Participationinmandatedandvoluntarysocialriskmanagementarrangements:The

roleandlimitsoffinancialeducationandotherinterventions."SocialInsurance,Informality,andLabourMarkets:HowtoProtectWorkersWhileCreatingGoodJobs7879(2014):191.

Hu,Yu-Wei,andFionaStewart."PensionCoverageandinformalsectorworkers."(2009).OECD.

23

ILO.INCOMESECURITYINOLD-AGE,DISABILITYANDDEATHOFABREADWINNER.2012.

ILO.TheInformalEconomyandDecentWork:APolicyResourceGuidesupportingtransitionsto

formality.2013.

Kelly,Tim,AleksandraLiaplina,ShawnW.Tan,andHernanWinkler."ReapingDigitalDividends."(2017).

Kuddo,Arvo,D.Robalino,andMichaelWeber."Balancingregulationstopromotejobs:from

employmentcontractstounemploymentbenefits."Washington,DC:WorldBankGroup(2015).

Madrian,BrigitteC.,andDennisF.Shea."THEPOWEROFSUGGESTION:INERTIAIN401(k)

PARTICIPATIONANDSAVINGSBEHAVIOR."QuarterlyJournalofEconomics116.4(2001).

Madrian,Brigitte.Forthcoming,2014.ApplyingInsightsfromBehavioralEconomicstoPolicyDesign.

AnnualReviewofEconomics.

Packard,T.2002.“Pooling,Savings,andPrevention:MitigatingtheRiskofOldAgePovertyinChile”,

PolicyResearchWorkingPaper2849.Washington,DC:WorldBank.

Pagés,Carmen,JameleRigolini,andDavidRobalino."SocialInsurance,Informality,andLaborMarkets:

HowtoProtectWorkersWhileCreatingGoodJobs."SocialInsurance,Informality,andLabourMarkets:HowtoProtectWorkersWhileCreatingGoodJobs7879(2014):9.

Palacios,Robert,andCharlesKnox-Vydmanov."Thegrowingroleofsocialpensions:history,taxonomy

andkeyperformanceindicators."PublicAdministrationandDevelopment34.4(2014):251-264.

Perry,G.,W.Maloney,O.Arias,P.Fajnzylber,A.MasonandJ.Saavedra-Chanduvi.2007.Informality:ExitandExclusion.InternationalBankforReconstructionandDevelopment.Washington,DC:

WorldBank.

Pfutze,Tobias."TheeffectsofMexico’sSeguroPopularhealthinsuranceoninfantmortality:An

estimationwithselectionontheoutcomevariable."WorldDevelopment59(2014):475-486.

Ribe,Helena,DavidRobalino,andIanWalker.Fromrighttoreality:incentives,labormarkets,andthechallengeofuniversalsocialprotectioninLatinAmericaandtheCaribbean.WorldBank

Publications,2012.

Robalino,DavidA.,MilanVodopivec,andAndrásBodor."Savingsforunemploymentingoodorbad

times:optionsfordevelopingcountries."(2009).

Robalino,DavidA.,LauraRawlings,andIanWalker."BuildingSocialProtectionandLaborSystems:

ConceptsandOperationalImplications."(2012).

Robalino,DavidA.,andMichaelWeber."Designingandimplementingunemploymentbenefitsystemsin

middleandlowincomecountries:beyondrisk-poolingvssavings."IZAJournalofLaborPolicy

2.1(2013):12.

RuppertBulmer,E.,A.KuddoandM.Weber.“Firms’PerspectivesonReducingtheCostsandEnhancing

theBenefitsofFormality”,WorldBankJobsCCSA(2017).

24

Tornarolli,L.,Battistón,D.,Gasparini,L.,&Gluzmann,P.(2014).ExploringtrendsinlaborinformalityinLatinAmerica,1990-2010(No.159).DocumentodeTrabajo.

VanGinneken,Wouter."Socialsecurityfortheinformalsector:Anewchallengeforthedeveloping

countries."InternationalSocialSecurityReview52.1(1999):49-69.

VanGinneken,Wouter."Socialsecuritycoverageextension:Areviewofrecentevidence."InternationalSocialSecurityReview63.1(2010):57-76.

Vodopivec,Milan.Incomesupportfortheunemployed:Issuesandoptions.WorldBankPublications,

2004.

Wagstaff,Adam,andWanwiphangManachotphong."Thehealtheffectsofuniversalhealthcare:

evidencefromThailand."(2012a).

Wagstaff,Adam,andWanwiphangManachotphong."Universalhealthcareandinformallabormarkets:

thecaseofThailand."(2012b).

WorldBank.2016.WorldDevelopmentReport2016:DigitalDividends.Washington,DC:WorldBank.

doi:10.1596/978-1-4648-0671-1.License:CreativeCommonsAttributionCCBY3.0IGO

25

MoreJobsWorkingPapers:

1. FosteringQualityofEmploymentforWomen

2. WheretoCreateJobs:CitiesorTowns?

3. TargetedSMEFinancingandEmploymentEffects:WhatDoWeKnowandWhatCanWeDo

Differently?

4. Zambia:AReviewoftheWorldBankGroupJobsPortfolio

5. JobCreationinthePrivateSector-AnExploratoryAssessmentofPatternsandDeterminantsat

theMacro,Sector,andFirmLevels

Address:1776GSt,NW,Washington,DC20006Website:http://www.worldbank.org/en/topic/jobsanddevelopment

Twitter:@WBG_JobsBlog:https://blogs.worldbank.org/jobs/