control 3 - fisher gonzalez serra wd 2006_pdf-notes_flattened_201212172252

World Development Vol. 34, No. 4, pp. 647–664, 2006� 2006 Elsevier Ltd. All rights reserved

0305-750X/$ - see front matterdoi:10.1016/j.worlddev.2005.09.008

www.elsevier.com/locate/worlddev

Does Competition in Privatized

Social Services Work? The Chilean Experience

RONALD FISCHER, PABLO GONZALEZ and PABLO SERRA *

Universidad de Chile, Santiago, Chile

Summary. — This paper examines the premises under which the privatization of Chilean social ser-vices of 1981 was carried out. Reformers expected that (i) competition between providers wouldensure a more efficient supply of services, and (ii) shifting decisions to households would guaranteea better satisfaction of household needs. Although some of the benefits of competition are lostthrough rent dissipation, especially in the providers’ search for the more profitable customers,we conclude that the reform has benefited society by providing competition to public providersand reducing the risk of political capture. The major lesson, however, is that the full benefits fromprivatization-cum-competition are slow to arrive and require able regulators. Moreover, the bene-fits of privatization may depend to a large extent on implementation fine tuning.
� 2006 Elsevier Ltd. All rights reserved.
Key words — Latin America, Chile, social services, privatization, regulation

* We thank the three anonymous referees for the very

helpful comments. Final revision accepted: September 8,2005.

1. INTRODUCTION

This paper evaluates the privatization of so-cial services undertaken in 1981 by the Chileangovernment. 1 Our assessment of these reformsconsiders their original intent, which was to in-crease the efficiency of markets through liberal-ization as well as to increase freedom of choice.The government expected that competition be-tween providers would ensure a more efficientsupply of social services. Moreover, shiftingdecisions concerning social services to house-holds would guarantee a better match betweensupply of services and household needs. Thepaper derives the lessons from this quarter-century experience by taking advantage of thesimilarities and differences among the socialservices.

A private pension fund system and a privatehealth insurance system that link benefits tocontributions started to operate in 1981. 2 Thegovernment also introduced a voucher systemthat does not discriminate between public andprivate schools, and established non-discrimi-natory rules for the entry of new institutionsinto tertiary education, which previously hadbeen tightly restricted. In addition, in a bidfor decentralization, local governments becameresponsible for the primary level of the public

647

health care system as well as for public school-ing. As many countries are considering givingthe private sector a more substantial role inthe provision of social services, the Chileanexperience is of interest because of its wide-range and early implementation.

There were limits to the liberalization pro-cess, however, because the government estab-lished restrictions on consumer sovereignty. 3

For instance, dependent workers must con-tribute 10% of their gross wage income to theprivately administered pension fund of theirchoice, and must spend at least an additional7% of wage income in purchasing a healthinsurance plan. The government subsidizes edu-cation directly, rather than providing cash-transfers to families. The reasons for theserestrictions can only be guessed, as they werenot made public by the government. One possi-ble explanation is moral hazard in decisionsinvolving social services. Senior citizens with-out pension plans or severely ill citizens with-out health plans would demand and probablyreceive State support even without having

648 WORLD DEVELOPMENT

contributed to the system. A second reasonwould be the self-control problems recently de-scribed by the behavioral economic literature(Mullainathan & Thaler, 2001). Even whenindividuals realize their self-control problems,they could still undersave. In this context, man-datory savings could be welfare enhancing(Diamond & Koszegi, 2003). Similarly, familiescould underinsure in health and underinvest ineducation. Externalities, at least in educationand health, might also account for these restric-tions on household sovereignty.

The reformers were also concerned about thefunctioning of these industries, as they providecomplex services, which are characterized byasymmetric information and require knowledgeto make the correct choices. There are at leasttwo restrictions that consumers face whenmaking decisions: first, the limited availabilityof relevant information and, second, theirrestricted capacity to process the informationwhen available. Obtaining relevant informationis costly and individual consumers might preferto use proxies that they believe are correlatedwith the ‘‘unobservable’’ relevant characteris-tics of the services. 4

It is probable that the decision maker did nothave complete confidence in the ability of fami-lies to decide correctly on complex issues requir-ing specific knowledge. Hence, the State retainedits regulatory and supervisory powers, and setlimited trust on unregulated market forces.First, it kept a close eye on the financial solvencyof social security institutions. It set limits on thetypes of financial instruments in which the pen-sion funds could invest and caps on investmentin the different instruments. Other rules tendedto protect consumers. Since 1991, the healthinsurance firms are required to renew their con-tracts to any affiliate who desires renewal, andyearly raises in plan costs are capped. Schoolsare restricted by the curricula defined by theMinistry of Education. 5 The State also providesminimum levels of information about social ser-vices or forces social service providers to do so.For instance, it runs a national standardized testfor primary and secondary students, whichenables comparisons between schools.

Competition among private providers wasexpected to result in better quality or lowerprices for services and more variety. 6 However,although individuals value freedom of choice,there is increasing evidence that this is expen-sive in social services. The administrative costsof competitive service suppliers are high be-cause they lose the scale economies of a single

compulsory system and they incur additionalcosts to attract customers (advertising, sales-people, and the like). Another reason why indi-vidual choice is expensive is that regulations ormarket conditions tend to limit price discrimi-nation, providing incentives to spend resources(salespeople) to attract clients whose compul-sory contributions exceed the costs of serviceprovision (Diamond & Valdes-Prieto, 1994).

The Chilean pension funds administrators(AFP) receive a fee that is, in the main, a frac-tion of the affiliate’s contribution to the pensionfund. However, since the cost of an AFPconsists of a large fixed cost (the research de-partment) and variable costs that depend onaccumulated funds, younger and higher incomeindividuals are more attractive to AFPs. Anal-ogously, private health insurers try to attractclients with lower than average health risksand, unless forced to do so, do not renew con-tracts to families whose members developchronic or catastrophic illness. Schools com-pete for better-endowed students and forceout underachievers in order to perform betteron national standardized tests. Since parentstend to prefer those schools that perform wellon national tests, school owners devote re-sources to attract skilled students.

A further possibly undesired effect of individ-ual choice in privately provided social servicesis that it generates market segregation whenprivate providers search for ‘‘good’’ customers.In addition, segregation by suppliers might bereinforced by segregation by consumers. Forinstance, upper class or more educated parentsmight prefer schools attended by children ofsimilar families (either due to peer effects, socialsegregation, or social networks). Similarly,health insurance customers might prefer insur-ers with a low risk portfolio that cater to house-holds with fewer health risks, as they are lesslikely to default. Social stratification could bereinforced in a world of imperfect informationbecause poor and less educated householdshave fewer social networks, less information,and a smaller ability to use the information(for school choice, see references in Schneider,1999, pp. 6–7). Hence, policies that provide ac-cess to relevant information and train consum-ers on exercising choice are likely to improvemarket efficiency as well as reduce segregation.

The rest of the paper is organized in the fol-lowing way. Section 2 analyzes the pensionfund system, while Section 3 studies the healthinsurance market. Section 4 deals with educa-tion, and Section 5 concludes.

DOES COMPETITION IN PRIVATIZED SOCIAL SERVICES WORK? 649

2. THE PRIVATIZATION OF THEPENSION SYSTEM

(a) Description 7

In 1980 the government introduced a law thatcreated the private pension fund system, whichbegan operating in 1981. This new systemintroduced compulsory savings accounts forretirement. Dependent workers are required tosave 10% of their gross wages (with a maximumof around US$ 150 per month) in the AFP oftheir choice (the contribution is deducted fromthe personal income tax base). 8 At retirement,workers make phased withdrawals from theirindividual accounts, and at any time they canpull out the remaining funds to purchase anannuity which provides a fixed monthly incomein real terms. Workers contributing to the pre-vious pension system at the time of the reformwere allowed to choose between keeping theirold system and switching to the new system. 9

In addition, workers pay a percentage oftheir monthly deposits as a fee to the pensionfund administrators. 10 Although AFPs are freeto set their commissions, they have to chargethe same rates to all affiliates independently ofcosts (including the cost of the insurance thatcovers a disability pension and a pension forsurvivors in case of death). As the cost of AFPsconsists of a large fixed cost and variable coststhat depends on accumulation while their in-come is derived mostly from the variable feeon wages, they prefer to attract young andhigh-income workers.

The State remains responsible for some as-pects of social security. First, it regulates andsupervises the AFPs. In this role, its main taskis to determine the types of instruments in

Table 1. Types o

Type offund

Legal limits on variableincome instruments

Distrb

Maximum(%)

Minimum(%)

Defaultassingmenta (%

A 80 40B 60 25 53C 40 15 42D 20 5 5E 0 0

Source: Superintendencia de AFPs and Asociacion de AFPa Corresponds to the default assignment at the inception ob Figures corresponds to 43.3% of affiliates that made an ec Annualized.

which the funds can invest and to set limits toinvestment in the different instruments. Second,it provides state guarantees to the funds inAFP’s. Third, the State continues to pay thepensions of workers retired on the previouspay-as-you-go system and receives contribu-tions from the workers that chose to remainin that system. Fourth, it acknowledges thecontribution to the old public system of work-ers who switched to the private system bygranting them a ‘‘recognition bond,’’ whichadds to the accumulated funds at the time ofretirement. 11 Fifth, it guarantees a minimumpension to workers that have contributed forat least 20 years to the pension system. Finally,it pays a minimum pension, based on need, toindividuals over age 65 and the disabled overage 18.

Until the year 2000, AFPs were allowed tooperate only a single fund. In that year, theywere authorized to create a second fund con-ceived for workers close to retirement that pre-fer less risk, since this fund could only investin fixed income instruments. In August 2002,three additional types of funds were intro-duced, with the object of increasing the optionsfor affiliates. Each of the five funds has differentmaximum and minimum limits for their invest-ments in variable income instruments, as shownin the first two columns of Table 1. Workersnow choose both type of funds as well as thepension fund manager.

(b) Evaluation of the private pensionfund system

We focus on the impact of the new pensionsystem on AFPs and workers. Table 2 showsthat pension funds administrators have been

f pension funds

ibution of affiliatesy type of fund

Real rate of returnsc

September 2002–December 2003 (%)

)Choiceb

(%)December2003 (%)

12.2 3.1 22.025.7 42.0 12.548.7 43.8 7.96.7 9.7 6.56.7 1.3 1.8

s.f the multi-fund system.xplicit choice up to October 2003.

Table 2. Profitability of the fund administrators andof pension funds

Year Administratorsprofit rate (%)

Funds type C realrates of return (%)

1985 17.9 13.41990 56.7 15.61995 21.7 �2.51997 17.5 4.71999 30.3 13.32000 50.1 4.42001 33.8 6.72002 27.0 3.02003 25.6 10.6

Source: Superintendencia de AFPs.


very profitable for their owners. By the mid-1990s, the AFPs became less profitable due inpart to rent dissipation caused by an increasein their selling efforts. The salaries of salesper-sons reached 36% of operational costs of AFPsin 1997 (45.3% of the total when the insurancepremium is excluded). At the time, it was com-mon for affiliates to receive gifts from salespeo-ple for switching AFPs. Since the profitabilityof the funds in the different AFPs was similar,5% of affiliates switched every month. In an at-tempt to reduce the sales effort and hence lowercommissions charged to affiliates, the govern-ment made it more difficult to change AFPs.By the year 2000, sales effort represented a mere14.9% of operational costs and had declinedfurther to 12.9% by 2002.

Although affiliates were helped by the fall innet commissions, the major beneficiaries ofthe regulatory change were the AFPs. The re-duced efforts to attract affiliates increased theirprofitability (see Table 2). The government isnow attempting to increase competition byopening the system to other financial institu-tions. In these last two years the profitabilityof AFPs has declined due to an increase inthe survivors and disability insurance premiumswhich has not been transferred to prices. Thispremium increased from 41.7% of AFP’s oper-ational costs in 2000 to 44.0% in 2002. This isdue to aging in the affiliates of the new systemand also to increased disability claims due tohigher unemployment during these years (Cas-tro, 2005).

The biggest advantage for workers broughtby the private system is that they have moresecurity about the destiny of their pension con-tributions. 12 Diamond (1996) believes that theChilean privatized mandatory pension systemprovides a high degree of insulation against

political risks. As individual accounts are con-sidered private property, they are entitled tothe same protection as other assets. The abovedoes not imply that the system is fully protectedfrom government expropriation, as stressed byOrszag and Stiglitz (2001) and Mesa-Lago(2002). As shown by the recent case of Argen-tina, during a severe crisis the government canrequire pension funds administrators to buypublic bonds that represent a sure loss for pen-sion funds, in effect expropriating the savings ofworkers.

However, the old system was prone to polit-ical risk even under normal conditions. Duringthe first years of operations the old system col-lected funds in excess of withdrawals, as theratio of pensioners to contributing workers waslow. This led to a reduction in the requirementsfor entitlement to a pension and to the use ofthe funds to pay for other benefits or services.For instance, some pension funds provided25–30 years mortgage loans to affiliates, withnominal interest rates of 5% when averageinflation was close to 20%. There also was thelatent risk that the government could reducepension benefits discretionally to finance a bud-get deficit, as occurred in 1985. 13

The old system was also notorious by its lackof fairness. Fifty-two different pension systemsfor different industries and occupations notonly multiplied bureaucracies but also led touneven retirement benefits. Pension benefitsdepended on the ability of affiliates to a specificfund to exert political pressure (Arellano,1985). The system redistributed from the poorto the well to do and full indexation to wageswas only granted to selected high-income work-ers. White-collar workers could retire youngerthan the blue-collar workers: blue-collar work-ers retired at the age of 65, while white-collarworkers could receive pensions after 35 yearsof work. In the case of the public sector,employees could retire after 30 years, and after24 years of work in the case of bank employees.

The private pension fund system seems to becostly. 14 In 1984, on average workers paidcommissions of 47.6% of the pension contribu-tion (see Table 3), a sharp rise from the 35.7%of 1982. However, by 2002, average commis-sions had fallen to 24.3% of contributions. Asa percentage of accumulated funds, commis-sions fell from 11.8% in 1984 to 1.3% in 2002,due to the increase in the accumulated funds.These numbers are high when compared tothe administrative costs of the Chilean publicpension system, which represented 1.4% of pay-

Table 3. Commissions of pension fund administrators

Year As % of wage income As % of accumulatedfunda

Gross Insurance premium Netb Gross Netb

1982 3.57 2.50 1.071984 4.76 3.50 1.26 11.82 3.131985 4.51 2.52 1.99 4.46 1.971990 3.15 1.03 2.12 4.20 2.831995 3.06 0.76 2.30 2.12 1.591997 2.96 0.71 2.25 2.01 1.532000 2.49 0.93 1.56 1.44 0.902002 2.43 0.91 1.52 1.28 0.87

Source: Superintendencia de AFP (Documento de trabajo No. 4).a Own calculations based on information of the Superintendencia’s bulletins. Computed by dividing yearly totalcommissions by average accumulated funds.b Excludes the survivors and disability insurance premium.


outs and 7% of worker’s contributions in theyear 2000. 15 One of the explanations for thelower costs is that the public pension systemmanager does not invest funds but works on apay-as-you-go basis, using government trans-fers to cover the deficits.

Net commissions, that is, those that excludethe cost of the survivors and disability insur-ance premium are a better reflection of admin-istrative costs. On average, workers paid netcommissions of 21.2% of the pension contribu-tion in 1990 (see Table 3) and 15.2% in 2002. Inturn, net commissions as a percentage of accu-mulated funds declined from 2.83% to 0.87% inthe same period. Unfortunately, informationon insurance premiums is unreliable in the firstyears of the system, as AFPs overstated insur-ance while understating commissions (in 1983,only 43% of their income came from affiliates’net commissions, with the remainder consistingof rebates from insurers).

Diamond (1996) claims that the issue is therelative administrative efficiency of the privatemarket, not particular features of the Chileansystem. He claims that the cost per person inChile is not very different from the costs ofother privately managed pension systems, andthese are usually more expensive than well-run unified government-managed systems.Note, however, that commission charges havefallen substantially since 1994 and could falleven further in the future. In particular, AFPscharge annual commissions of only 0.47–0.64% of the accumulated funds in the case ofthe voluntary retirement savings funds thatwere created in 2002, a much lower rate thanthe 0.87% for mandatory pension savings. A

possible explanation for this difference is thatcompetition for voluntary pension savings isstronger than in the case of mandatory pensionsavings, since other financial institutions canparticipate in this market. It appears that thereare barriers to entry into the pension fund mar-ket, and it has become more concentrated overtime as the three main AFPs concentrate 78%of all affiliates. 16

The average annual real rates of return sinceinception till 2003 of the different pension fundsvary from 10.1% to 10.6%. This extraordinaryperformance of pension funds reflects high rateof returns to capital in the Chilean economy ingeneral, which are in turn explained by the col-lapse of the financial system in the early 1980sthat led to soaring rates of return on govern-ment bonds and extremely high interest rates,and the colossal climb of equity prices in theearly 1990s after return to democracy. Further-more, the AFPs were highly constrained in thetypes and the amounts of instruments theycould invest, which makes it difficult to judgetheir investment performance. Rates of returnshave been lower in recent years. From January1998 up to December 2003, the average rate ofreturn on pension funds fell to 6.7%.

Consequently, in spite of the high cost of thesystem, affiliates have had high effective realrates of return, that is, real rates of return oncecommissions are discounted from the fund.From its inception in July 1981 up to December2003 the average annual effective real rate re-turn for fund type C was 7.0% for an individualearning around US$ 120 per month (the mini-mum legal wage) and 7.3% for an individualearning around US$ 1.400. 17 Rates of returns


have moderated in recent years. The averageeffective real rate of return was 5.7% for thelowest income workers and 5.9% for those withmonthly salaries above US$ 1.400 during 1998–2003. Again, it is more appropriate to excludesurvivors and disability insurance premium,but information on the latter is not reliable(see Table 3). A rule of thumb, however, wouldbe to increase the above effective rates of returnby 1%.

An oft-touted advantage of the AFP systemis that it allows for freedom of choice amongadministrators. 18 However, workers did nothave much actual choice, as the compositionand consequently the results of fund portfoliosof different AFPs were quite similar. Moreover,since the law sets penalties for AFPs whosefund returns are more than 2% below theaverage of the industry, while there is no com-pensating benefit when the returns are higherthan average (except for marketing possibili-ties), AFPs followed a herd-like investmentbehaviour. The limits on investment have beenrelaxed over time, and hopefully investmentability will become more important in the fu-ture. The multi-fund system will generate moreinvestment diversity, one of the reasons for itsintroduction. Table 1 shows that the profitabil-ity of the new funds can vary substantially.

Diamond and Valdes-Prieto (1994) believethat affiliates seem to choose among AFPsaccording to past returns, and that the publicseems to be unaware of the trade-off betweenrisks and return. 19 However, it appears fromTable 4 that affiliates chose the AFP thatcharges the lowest commission. The two AFPs

Table 4. AFPs market parti

AFP Affiliates CommissionsJuly 2003

Cuprum 312,122 24.90Habitat 811,280 24.01Magister 66,002 28.22Planvital 97,931 29.44Provida 1,454,441 24.03Suma 431,823 25.64Sta. Marıa 283,488 26.02System 3,457,087

Source: Superintendencia de AFPs.a Excludes the survivors and disability insurance premium c254,100 in April 2003.

with the lowest commissions are the ones withthe largest number of affiliates, while the mostexpensive ones are the smallest. An alternativeexplanation is that the larger AFPs can affordto charge lower commissions because of scaleeconomies. There are other explanations, andwithout a more detailed analysis, it is difficultto choose among these hypotheses.

When the recent reform was introduced, affil-iates could choose among the five funds thateach AFP was allowed to administer. Therewas a default assignment that depended onthe age of the affiliate. The fourth column inTable 1 shows the distribution of affiliates ifthey had all been assigned by default. Beforethe new funds started operating, 35.0% of affil-iates, including both active and retired workersactively exercised their choice and by October2003 the figure was 43.3%. The fifth columnin the table shows the choices of affiliates whomade an explicit option and the sixth columnshows the distribution of affiliates by fund inApril 2003. This seems to indicate that affiliatesare willing to make choices when they havemore at stake.

However, affiliates still seem to have littleinformation on the working of the pension fundsystem. A sample survey of 17,000 beneficiariesof the APF system conducted during the secondhalf of year 2002 found that 30% of the inter-viewed read the quarterly account statementthey receive. 20 Moreover, only 10% said thatthey could understand it, and 1% used it fordecision making. About 95% did not knowhow much fund managers charged for theservice.

cipation and performance

Real rates of returnMay 2000–April

2003 (%)

Real rates of returnMay 2002–April

2003 (%)

Fund Neta Fund Neta

5.07 4.19 2.74 2.015.80 4.95 4.16 3.435.83 4.73 4.45 3.535.68 4.58 3.76 2.805.31 4.46 3.20 2.475.54 4.59 3.52 2.725.48 4.54 3.48 2.675.46 4.57 3.48 2.72

ost and corresponds to an affiliate with an income of Ch$


3. THE HEALTH INSURANCE SYSTEM 21

(a) Description

A second innovation, introduced in 1981, wasthe partial privatization of the health insurancesystem. Before 1981, all workers had to contrib-ute compulsorily to the public health insurancesystem (Fonasa) even if they did not use its ser-vices. 22 Under the new system, private healthinsurance firms, known as Isapres, were created.All active and retired workers must contribute7% of their wage (or pension) to a health insur-ance system, with a maximum compulsory con-tribution of US$ 120 per month. Workers canchoose between one of the 10 open privatehealth insurers (Isapres) or Fonasa. Currently,the private system covers about 18% of the pop-ulation, while Fonasa covers almost 70%, andthe armed forces health systems and the privatesector cover the rest of the population. Peoplethat are self-employed can pay into eithersystem (Isapres or Fonasa).

Fonasa offers the same array of health plansto all affiliates, independently of their contribu-tions and of the number of dependants of theaffiliate. Affiliates can choose between twoforms of health provision: free choice or institu-tional. Under free choice, the affiliate and itsdependants can select a private health provider(which has a contract with Fonasa at specifiedrates), but they must pay a co-payment. In theinstitutional mode, co-payments are lower andinversely related to the beneficiary’s income(the co-payment is zero for low-income individ-uals), but beneficiaries get health provision fromthe public system, without choice. Since Fonasaserves the destitute, provides public goods (vac-cination programs and health campaigns, etc.),and finances the primary health centers thatare supervised by the municipalities since1981, 54% of its funding comes from the State.

The private system is run on a totally differentbasis: the affiliate signs an annual contract withan Isapre that specifies the benefits she will re-ceive, and which depends on her contribution,age, sex, the number of dependants and theirage, sex, etc. 23 Affiliates can improve their basicplan by paying additional, voluntary contri-butions. In December 2003, voluntary contri-butions amounted to 34.1% of compulsorycontributions. The clients of Isapres can use pri-vate health providers in two ways: first, they canbuy a voucher previous to receiving service froma provider that has a contract with the Isapre (atpre-specified rates). The cost of the voucher is

the client’s co-payment. Alternatively, they canchoose any provider and get a reimbursementfrom the Isapre afterwards. In general, the re-imbursement does not cover the full cost of thevisit, so there is an implicit co-payment. Theamount of the co-payment or reimbursementdepends on the specific plan that the affiliatehas contracted with the Isapre. On average, Isa-pres pay 68% of medical costs, the remainderbeing the co-payment by affiliates.

The private system is too expensive for mostChileans. The compulsory contribution is notsufficient for low-income individuals or forpotential affiliates with high health risks to buyinto a good plan. The nationally representativeCASEN survey of the year 2000 showed thatonly 3.1% of the members of the lowest incomequintile are beneficiaries of Isapres (while54.2% of the households in the highest incomequintile were beneficiaries). In September2003, only 5.1% of beneficiaries of the Isapressystem were older than 60, while their presencein the population is more than double thatpercentage according to the 2002 census.

The number of beneficiaries of the Isapre sys-tem grew every year until 1997, when it came torepresent 26.5% of the population. Since then,the percentage of the population that is abeneficiary of the system has been decreasing,reaching only 17.8% of the population in 2003as shown in Table 5. Several reasons explainthis decline: (i) increased unemployment sincethe late 1990s, (ii) increased funding for Fonasa,making it relatively more attractive (expen-diture per affiliate increased by 300% in the1990s), (iii) the decline in (less expensive) collec-tive plans in Isapres, partly explained by theelimination of the tax incentive for the corpo-rate contribution to the collective plans of theiremployees, (iv) better supervision by Fonasa tobar Isapre affiliates from getting free services(as indigents) from the public system, and (v)the increase in the costs of the Isapres.

There has been a general increase in the costof the medical services, as shown in Table 6,due to an increase in both the number of healthvisits by beneficiary and in the cost of thesevisits. Rodrıguez and Tokman (2000) have con-structed a quantitative index of the cost ofhealth procedures. They have estimated that thenumber of procedures just about doubled inthe period 1990–99 in the Isapre system, whiletotal costs (excluding sickness compensation)increased by 165%, showing that there has beenan increase in the unit costs of medical servicesof about 30%. 24 These increases continued in

Table 5. The private health insurance system

Year Number ofbeneficiaries

% ofpopulation

Administrativecosts

Profitratea

Profit/sales Health medicalvisits

Expenditure perbeneficiaryb

Thousands ofDecember 2000 Ch$

Isapres Fonasa

1985 545,587 4.5 29.0 39.9 6.01 8.36 118.9 –1990 2,108,308 16.0 21.4 26.8 7.67 9.04 104.9 37.61995 3,763,649 26.5 20.0 23.1 4.83 9.41 147.4 88.31997 3,882,572 26.6 19.0 15.3 3.00 10.18 162.7 103.62000 3,092,195 20.3 17.5 9.2 1.82 13.12 212.5 118.3c

2002 2,828,228 18.7 14.9 18.0 1.40 14.00d 197.4 –2003e 2,729,088 17.8 14.4 21.6 2.50 13.70d – –

Source: Series Estadısticas, Superintendencia de Isapres. Rodrıguez and Tokman (2000).a Profits over equity.b Excludes the subsidy of medically certified absence from work. Co-payments included, using data for the year 2000.The expenditure in the Program for Complementary food is excluded from Fonasa.c Corresponds to 1999.d Provisional numbers for 2001, 2002, respectively.e Data to September 2003.

Table 6. Health services provision per beneficiary, year2000

Type of service Fonasa Isapres Difference (%)

Doctor visits 3.65 3.80 4.1Lab. exams 4.56 4.12 �10.0Surgery 0.08 0.11 37.5

Expenditureper beneficiary(Ch$ of 2000)a

87,339 137,525 57.1

Sources: Fonasa Statistics, Superintendencia de Isapres,and authors’ calculations.a Excludes co-payments by Isapre patients and medicallicenses.


the next decade: comparing 2002 with 2003,both health visits and unit costs increased by6% (Resultados Sistema Isapre, April 2004).A large fraction of the increase in unit costs isdue to newer and more sophisticated medicaltreatments and better service quality. 25 Thestricter regulation of Isapres has also contrib-uted to their costs and hence to the higher priceof their plans. The 1990 law forced Isapres torenew contracts to expensive affiliates and it in-creased the minimum required coverage.

(b) Evaluation

The Isapre system was very profitable untilthe late 1990s. Even in 1995, when the systemwas already mature, the profit rate on equitywas 30%. In recent years, the profitability of

the system has been variable, as new regula-tions have increased costs. In 1999, for in-stance, the average profit rate was 2.2%,bouncing back to 21.6% in the year 2003 asIsapres were able to raise rates (at the cost ofa reduction in the number of affiliates) asshown in Table 5. Nevertheless, the marginsare becoming thinner: the earnings to sales ra-tio fell from almost 8% in 1990 to 2.5% in 2003.

Contrary to the AFP market, the privatehealth insurance market remained relativelyunconcentrated until recently, though thisseems to be changing. The three largest Isapreshave around 23% each of all beneficiaries, andthere is an additional Isapre with more than10% of the market. While the Herfindahl indexhas increased to only 1800, it is set to becomemore concentrated in the future as margins de-crease and the risk reducing effect of a largermass of clients becomes more important. Thelow concentration does not necessarily implycompetition among Isapres, since a large frac-tion of affiliates are captive in their Isapresdue to pre-existing illnesses among membersof their families, which ensures that otherIsapres will not accept them as clients.

The system has been beneficial to higher in-come families, since, under the previous system,their contributions were just another form oftaxation, as they did not use the public system.An overall welfare assessment is difficult, sincethe public health system lost the compulsorycontributions of these same households. Theexpense per beneficiary in the Isapre system is


almost 60% higher than in the public system,even though this difference has decreased overtime. This comparison underestimates the costof the private health system, because it omitsthe direct payments of Isapre affiliates for thepart of their treatment that is not covered bytheir plan, which represents 33% of the totalexpenditure. Notice, however, that the numberof health visits does not differ substantiallybetween the two systems, except in the case ofsurgery, as shown in Table 6.

A cursory analysis might suggest that thepublic system is more efficient. The problemwith that interpretation is that there is a signif-icant difference in the quality of care betweenthe two systems. Economic principles suggestthat private, individual health care insurancewith free choice of services and providers ismore expensive than public insurance withoutfree choice. First, because there is a tendencyto overprovide services: the emblematic exam-ple is the fact that 65% of all pregnancies inthe private sector ended in a caesarean sectionin 2003, while the average for the public sectorwithout free choice was less than half thatrate. 26 Second, the administrative cost of indi-vidual insurance contracts is higher, amongother things, because Isapres evaluate thehealth risks of each new affiliate and must en-sure that reimbursements are appropriate fortheir particular plans. 27 In the year 2003, theadministrative and marketing expenses were17.5% of total expenses, while the profit ratewas 6.9%. The administrative cost in Fonasais only 1.5% of the total expense. On the otherhand, other economic principles indicate thatpublic systems are less efficient due to lack ofcompetition. Public hospitals financed by Fo-nasa are probably less efficient than the privateclinics that provide services for Isapres. Thereare no serious estimates of the extent of theseinefficiencies, but anecdotal evidence suggeststhat they might be large.

One of the main problems of the Isapre sys-tem used to be that plans offered a good cover-age for routine health care, but a poor coverageof catastrophic illness, which should be themain object of compulsory health insurance.Strong criticism forced the Isapres to introducecatastrophic illness insurance in the year 2000.This program covers all expenses beyond apre-specified yearly expenditure by the benefi-ciary. The system does not allow free choiceand works as a ‘‘managed care’’ system forthese patients. Preliminary evidence seems toshow that this approach works well, but the

system has only operated for a few years. Anec-dotically, the system had covered 50 transplantsby March 2003, including six heart transplantsand 11 liver transplants. Furthermore, expendi-tures of the catastrophic insurance systemexceeded its premiums by 5%.

In any case, it is interesting to speculate as tothe reasons why clients would choose plans thatlack good coverage for catastrophic illness. Oneexplanation is that affiliates are myopic and donot evaluate the cost and/or probability of rarebut costly illnesses. Second, affiliates are able toswitch to the public system if they acquire anillness with little coverage under their Isapreplan. Third, the system is not transparent, sinceplans have limits on payments that depend on astandard devised by each Isapre that is hard touncover for each treatment.

Another problem is that Isapres try toexclude beneficiaries who develop expensiveillnesses. 28 In an attempt to end this problem,since 1991, the Isapres are required to renewtheir contracts to any affiliate who wishes to re-new. However, the Isapres found a way aroundthis obligation by raising the price of existingplans and offering new plans with similar bene-fits but at the original price only to affiliatesthat do not represent a high risk. The Superin-tendencia that supervises Isapres has institutedrules that try to reduce this type of risk selec-tion, by limiting the yearly price raises in plans.However, this approach runs counter to theinherent instability of the private health insur-ance system. Since low cost affiliates in a givenplan are attractive to other Isapres, there is atendency to poach them with a plan with simi-lar characteristics but without the expensiveindividuals. Even if this last problem might besolved, affiliates that (or whose dependants) ac-quire an expensive illness will still be unable toswitch between Isapres, thus losing one of themain advantages of the system: the freedomof choice between Isapres.

For a young healthy single agent, her definedcontribution should be higher than her healthcosts, so she is a profitable affiliate and Isapreswill compete for her by offering unneededhealth services. 29 This type of inefficient rentdissipation replicates the phenomenon that oc-curs in AFPs, and is caused by the compulsorycontribution. When these same clients growold, their fees will increase and they may haveto switch to plans with lower coverage, just asthey start needing the better coverage. 30

Hence, there are strong pressures to change toa system in which at least a minimum level of


health insurance can be bought at a fixed priceindependently of age, sex, and health status,financed via a compensation fund.

Most of the problems arise from the seriousinformation asymmetries in private health sys-tems (see Fischer & Serra, 1996). There are waysof reducing these problems, but they remainintrinsic to private health insurance systems,so they cannot be totally eliminated from a sys-tem that combines free choice of providers andasymmetric information. Recent legal reformshave tried to address the problems of theChilean Health system by creating a minimumhealth plan called AUGE. Isapres and Fonasaare now required to offer coverage for the 24most important health problems, with themaximum annual co-payment being a multipleof the affiliates’ monthly contributions. Fonasawill have to abandon its strategy of promisinguniversal coverage while rationing health careby making patients wait (at least for the 24 diag-nostics covered by the new plan), sometimes foryears, to receive treatment. Isapres, which havetimely care, will have to limit the co-paymentsfor these diseases in all their plans.

The progress in regulating Isapres since 1990has solved some of the major problems of thesystem. 31 Other problems remain: clients whocannot change Isapres after developing chronicdiseases (captive clients), the rise in medicalcosts caused by competition between Isapresfor new clients and those caused by moralhazard in the doctor–patient relation. Finally,note that catastrophic illness insurance and theAUGE plan both require Isapres to collaboratein creating a unified plan to avoid creamskimming of the healthier affiliates, but thisfacilitates cartelization. The basic dilemmaremains: the productive efficiency of competi-tion versus its inability to insure patients dueto cream skimming: the legal changes have justchanged the weight given to these problems,arguably for the better.

4. PUBLIC EDUCATION ANDSCHOOL VOUCHERS

(a) The reform

The three main elements of the educationalreform were (i) the shift of public resourcesfrom tertiary to primary and secondary educa-tion, (ii) the transfer of state-owned schools tolocal municipalities, and (iii) the establishmentof a non-discriminatory subsidy (an implicit

voucher) per enrolled student both at publicand private schools. The role of the central gov-ernment was reduced to supporting and con-trolling schools (Parry, 1997). The financingof the university system changed radically asthe responsibility for financing tertiary educa-tion shifted from the state to families. This deci-sion was based on two premises. First, spendingin tertiary education is regressive, since themajority of tertiary students belong to middleand upper income households. Second, the pri-vate return to tertiary education is high andexternalities are less likely to exist than in pri-mary education, which means that there is lessneed to subsidize tertiary education since thebenefits are internalized.

Policy makers believed that state financed pri-mary and secondary education would be im-proved by having education providers competefor students, as public fund transfers to institu-tions would depend on the number of studentsattending classes, and having parents decidingfor their children’s education. The government’sstrategy was based on the premise that the com-petition for students would be based on aca-demic quality, which in turn assumed parentalinvolvement in children’s education and thecapability of evaluating alternative educationaloffers. They also trusted that another benefitof the reform would be an increased diversityin the educational offer, reflecting the heteroge-neous preferences and needs of families. More-over, policy makers believed that choice itselfwas a positive good, above and beyond theeffects on school and student performance.

Non-discriminatory transfer rules were estab-lished to insulate the system from the influenceof specific interest groups. Vouchers differaccording to the cost of providing education.The value of the monthly voucher is aroundUS$ 50 for primary schools and US$ 60 for sec-ondary non-vocational education. Subsidies forvocational secondary schools and differentialeducation (for children with learning disabili-ties) are higher because classes are smaller andrequire more equipment. Schools located inrural areas receive an additional per capita sub-sidy that decreases with the number of students.Although it has been recognized that the under-privileged students are costlier to educate, thevoucher system still does not differentiate alongthis dimension.

The country has gone a long way to decentral-ize the finances of the publicly funded educationsystem and to shift resources from tertiary edu-cation to secondary and primary education. In


1980, just before the policy reform, the Educa-tion Ministry transferred 37.5% of its budgetto higher education institutions and 7.3% toschools. In the year 2003, these figures were11.2% and 64.0%, respectively (see Table 7).The amount transferred through vouchers isroughly 82% of total school subsidies, and ofvoucher transfers, 40% went to private schools.

The number of private subsidized schoolsmore than doubled during 1980–2003, 32 whilethe number of public schools fell slightly in thesame period (see Table 8). The growth in theprivate subsidized sector might have beengreater if the value of the voucher had not beenreduced substantially in the mid-1980s. Accord-ing to figures of the Ministry of Education, ifthe real value of the subsidy per student is setat 100 in 1982, by 1985 the value in real termshad fallen to 75, and even by 1990 it stood at76. This led to stagnation in the number of pri-vate schools by the mid-1980s after the rapidgrowth of the first half of the decade. The pri-vate, subsidized schools survived the declinein the value of the voucher in the 1985–94 per-iod by increasing enrolment by 20%. During

Table 7. Budget of the Education M

Year Total School s

Amount

1980 717,332 52,0871981 806,865 194,1341985 727,930 329,0631990 586,300 374,1011994 876,400 528,7471995 1,011,302 640,1332000 1,654,419 1,041,9112003b 1,986,839 1,271,270

Source: Ministry of Education, document in process.a Excludes capital transfers, transfers in kind, and especialb Preliminary figures.

Table 8. Number of primar

Type of school 1980

Public, centralized 6,370(%) 72.4Municipal 0(%) 0Private, subsidized 1,627(%) 19Private, no subsidy 802(%) 9.1

Total 8,799

Source: Estadısticas de la educacion 2003, Ministry of Edu

the 1990s the fall in the per capita subsidywas reversed, and the real value of the averagevoucher recovered to 104 by 1994 and to 202 in2002. In 1994, the number of private schoolsstarted increasing once again (see Table 8).

Since 1993, the private subsidized schoolshave been allowed to charge fees to their stu-dents subject to some conditions, which includean upper limit on the fee (US$ 68 per month), aspecial tax favoring the Education Ministry, andthe availability of scholarships benefiting fami-lies that cannot afford these fees. Even thoughthe secondary municipal schools can also chargefees, these are in practice much lower and arecharged on a lower percentage of their enrol-ment. In the year 2002, the 1,663 private subsi-dized schools charging fees received Ch$ 126billion from students’ fees, while the corre-sponding 110 municipal schools received onlyCh$ 2.3 billion. The additional resources in theprivate subsidized schools may widen the exist-ing gap between private and municipal schools(discussed below). However, in the year 2003,municipal schools received additional fund-ing amounting to Ch$ 71 billion from the

inistry (millions of Ch$ of 2002)

ubsidiesa Tertiary education

% Amount %

7.3 268,695 37.524.1 197,676 24.545.2 200,777 27.663.8 110,484 18.860.3 144,304 16.563.3 156,984 15.563.0 209,378 12.764.0 222,042 11.2

programs.

y and secondary schools

1985 1994 2003

808 0 08.2 0 0

5,668 6,221 6,13857.8 63.6 54.72,667 2,707 4,155

27 27.7 37.0668 860 9306.8 8.8 8.3

9,811 9,788 11,223

cation.


municipalities as well as Ch$ 53 billion from re-gional governments for infrastructure improve-ments.

(b) Evaluation

The overall results of the reform of 1981 arestill unclear. There is no question that parentsvalue choice, and this is one of the reasonsfor the wide acceptance of private subsidizedschools. Another fact is that private schools re-quire less public funds than municipal schools.Indeed, private schools have financed their owninfrastructure, while municipal schools usethe previously existing infrastructure of stateschools. Another positive effect of the reformis the decrease in the rates of truancy, whichis explained by the inherent characteristics ofthe voucher scheme, though it is probable thatthe incentives in the system have led to looserstandards for passing grades. Although profit-ability rates are not available, the rapid expan-sion of private subsidized schools reveals this tobe an attractive business for private investors, ifthe second half of the 1980s is excluded.

A more complex question is whether thereform has added value to education or not.Unfortunately, the scores of the nationalstandardized tests that measure educationalachievement (SIMCE) are not comparable withthose previous to 1998, and cannot be used toassess the impact of the reform. Moreover,the large reduction in the voucher value duringthe second half of the 1980s is likely to havehad a negative impact on the quality of educa-tion. Data shown in Table 9 are consistent withthis hypothesis. Using the private non-subsi-dized schools as control group, we can see thatthe relative performance of private subsidizedschools deteriorated in the 1980s and improvedin the 1990s, mimicking the fluctuations in thevalue of vouchers.

The decrease in teachers’ salaries worsened thepool of applicants to teaching schools whileexisting teachers with the best outside oppor-tunities left the educational system, and this

Table 9. National standardized test

1988 1990 1

Type of school

Municipal 49.25 56.7 6Private subsidized 56.35 58.8 7Private non-subsidized 76.15 80.05 8

Source: Ministry of Education.a Before 1998 scores represent percentage of success. From

change in the quality of the stock of teachershas long-lasting effects. After decades of decline,the quality of new students at teaching schoolsbegan to improve in 1996, following steadyimprovements in teacher salaries since 1990.Teachers’ compensation in the municipal sectormore than doubled during the last decade, froman average monthly salary of $272,109 in 1990 to$699,104 in 2003, both in Chilean pesos of 2003.

There has been much disagreement about theimpact of the reform on the educational systemas a whole. Subsidized private schools on aver-age have performed better than municipalschools on standardized tests. However, the per-formance gap between the two types of schoolcould be attributed to differences in students’socioeconomic characteristics, including theirparents’ education. Studies using individualstudent data for tenth grade in the 1998 SIMCEresults show that private subsidized schools per-form better than municipal schools in nationalstandardized tests, even after accounting forsocioeconomic variables (Mizala & Romaguera,2001; Sapelli & Vial, 2001). Hence, socioeco-nomic differences of the student body explainpart, but not the whole of the performancegap between private subsidized and municipalschools. Possible explanations for the residualgap are (i) that private schools have managedto do better because they are more flexible, effi-cient, and have better incentives, (ii) that subsi-dized private schools spend more per studentthan municipal schools due to the additionalfee they charge, and (iii) private subsidizedschools can select students, whereas municipalschools cannot reject students unless they haveno openings.

Parry (1996) reported that 15% of municipalschools and 63% of private subsidized schoolsin Santiago (where selection might be morewidespread) applied some method of selection.Elacqua and Fabrega (2004) figures are 24%for municipal schools and 60% for privateschools, respectively. Moreover, it is easier toexpel students from private than from munici-pal schools. Student performance does not only

results (SIMCE), fourth grade

992 1994 1996 1999a 2002

3.85 64.43 68.00 238 2370.15 70.66 73.65 257 2576.05 85.07 85.85 298 299

then onwards, scores are normalized around 250.


depend on the quality of the education providedby the school, but also on personal abilities,parental educational level, peers’ skills, and soon. Hence, due to peer effects, it is rational forparents to send their children to schools that at-tract the best students. Therefore, prestigiousschools can choose their students based on aca-demic achievement prospects, so that standard-ized tests magnify the contribution of selectiveschools to the results.

Perhaps the comparison between municipaland private subsidized schools is beside thepoint, since one of the main benefits of the vou-cher system has been to increase the awarenessof school quality and to make schools behavemore competitively. However, the voucherreform of 1981 has not led to a dramaticimprovement in school quality and achieve-ment, as shown by the performance in interna-tional tests (PISA, TIMSS). There are severalpotential explanations for the poor interna-tional performance: factors limiting competi-tion, the decade of low funding that followedthe reforms, and the low quality of teachers.Nevertheless, the latest research results in Chileand other countries seem to show a positiveeffect of competition on school performance. 33

Several circumstances have dampened com-petition, the mechanism through which thequality of schooling was supposed to improve.First, parents did not have objective measuresof school quality: tests equivalent to theSIMCE have been used since the 1980s, butonly in 1995 were they published at the schoollevel. 34 Moreover, parents—even those consid-ering several alternatives—do not use test scoreinformation when selecting the school for theirchildren (Elacqua & Fabrega, 2004). Practicalreasons (such as proximity) and school valuesappear to be the most important factorsexplaining school choice. Second, the TeachersStatute has reduced the municipalities’ flexibil-ity in the hiring and firing of teachers and, inpractice, has excluded municipalities from thewage bargaining process. Third, municipalschools have been shielded from competition.

The idea behind (implicit) vouchers is that theincome of a school should depend primarily onthe number of students, and not on historicalbudget assignment. However, city mayors man-age the voucher income of students enrolled inmunicipal schools, and for political reasonssome mayors have refused to adjust the expendi-tures of schools with fewer students. 35 A secondfactor is imposed exogenously on the system:the Teachers Statute of 1991, which made it

almost impossible to fire teachers independentlyof their performance, and which set a fixed payscale that depends on seniority and not onperformance. Even though the minimum wageestablished in the Estatuto applies also to theprivate subsidized schools, other conditions donot apply, and it is possible to fire teachers atthe end of the school year.

By setting a national pay scale, the TeachersStatute re-established a centralized bargainingprocess between the Teacher’s Union and theMinistry of Education, distorting the market.Although the Statute was made more flexiblein 1995, following a financial crisis in the muni-cipal sector, 36 the Teacher’s Union managed tostall the individual evaluation of municipalteachers’ performance for more than a decade,despite being part of the Statute. After years ofnegotiating, and while several municipalitieswere designing their own evaluation proce-dures, the Teacher’s Union finally accepted anevaluation program for municipal teachers thatwill take place every four years starting in year2005 onwards. Those teachers that fail the testare forced to take a remedial course, and arere-evaluated the following year. If they fail thissecond test, they are suspended from teachingwhile they concentrate on training for a year.If they fail the test a third time they are dis-missed with a severance payment. 37

Another supply-side incentive mechanism,the National Performance Evaluation System(SNED) that provides supplementary fundingto the top quartile of schools, was introducedin 1996 to reward schools with better per-formance. In order to reduce any perverse in-centives, the SNED compares schools withrespect to their past performance and to similarschools (with similar socioeconomic character-istics) and penalizes schools that apply admis-sion tests or expel students. The SNED alsointroduced incentives for schools located in rur-al areas that are insulated from competition.Authorities are considering the publication ofvalue added indicators together with the rawscores and the results of individual students.

5. CONCLUSIONS

The privatization of social services producedseveral benefits. First, it reduced the uncer-tainty caused by political manipulation of thepublic pension system, increasing the securityand predictability of pensions. 38 The bur-den of the old system—given these political


pressures—was increasing over time and therewere doubts about the financial sustainabilityof the system, even at a time when the agingof the population was not envisaged. The pri-vate health insurance system has provided analternative to the public system for a fifthof the population, while at the same time putt-ing pressure on the public system to improve. 39

Similarly, while the evidence in favor of thevoucher system in education is not conclusive,it has put pressure on the public system to im-prove. This does not mean that privatizationhas been trouble-free. As is well known in priv-atization processes, the details of reform areessential to its success. The Chilean experienceoffers lessons both through the improvementsmade to the original design as well as by theproblems that remain.

Most households lack the information and/or knowledge needed to make rational deci-sions. 40 The cost of acquiring the necessaryknowledge is high and circumventing the infor-mation asymmetry is a hard task for indivi-duals. As a result, many individuals do notgrasp the main aspects that are involved inchoosing a provider of social services. The lackof understanding or information on the part ofconsumers has led providers to focus their com-petitive efforts on marketing rather than on thevariables that are relevant from the point ofview of an enlightened consumer (extent ofcoverage of a health plan, long-run net rateof return on a pension fund, and quality ofeducation). For instance, Isapres offered a widemenu of health plans, but few of them providedgood coverage for catastrophic illness until theywere required to do so. 41

The role of the government as a provider ofobjective information—a public good—hasbeen insufficient, though it has improved inthe last few years. In order to improve userinformation, the government could make AFPsprovide risk measures of their portfolios. Simi-larly, Isapres should publish their rate structurein terms that are more comparable to those ofother Isapres than at present. In education, un-til recently, parents did not have objective mea-sures of school performance, but the Ministryof Education still does not supply parents withinformation on their children’s performance, aswell as information on the educational valueadded provided by the school.

Another criticism of individual choice in so-cial services is that it is costly. In addition tothe normal costs that arise from the effort to at-tract more customers (advertising and the like),

there is the cost of attracting the best custom-ers, because the price structures allowed bythe regulator do not reflect the cost of provid-ing the service to different agents. A potentialsolution to this problem is to set prices in sucha way that all agents are equivalent for theprovider. For example, school vouchers couldbe higher for underprivileged students. 42 Inhealth insurance, possible solutions includecompensating Isapres (or Fonasa) for acceptingbeneficiaries that are costlier than average.Otherwise, an Isapre that offers good cata-strophic insurance may end up (throughadverse selection) with a portfolio of veryexpensive beneficiaries. These measures wouldnot only increase the efficiency of the systemby reducing socially unproductive expendituresbut would also attenuate any possible effects ofthe reforms on social equity and segregation.

Excessive regulation has been another factorin limiting the benefits from competition: a moreefficient and diverse supply of services. In thecase of private pension funds, restrictions onportfolio investment and rules that penalizeadministrators whose funds perform poorly ex-post (apart from the market punishment due tothe defection of affiliates from such a fund) re-sulted in similar performances from all pensionfunds. This is another reason why competitionbetween administrators has focused on variablesthat are irrelevant from the social welfare pointof view. The Teachers’ Statute has reducedflexibility in the public system and municipalschools have been shielded from competition.Experimentation is restricted by the curriculumset by the Ministry of Education. Beneficiariesof the private health insurance system preferplans with little coverage for expensive but infre-quent diseases because they can always switch tothe public system. Hence, the full gains fromcompetition have yet to be achieved.

There are measures that could facilitate entryof new providers. AFPs should establish a cen-tralized facility in charge of keeping records ofindividual accounts and mailing periodic state-ments to affiliates. This organism would easeentry by having regulated essential facilityperforming tasks that have strong scale econo-mies (see Diamond & Valdes-Prieto, 1994). 43

Employers are already able to electronicallytransfer their employees’ pension contributionsto a centralized system set by the AFPs. Provid-ing new entrants with information on all work-ers might also facilitate entry.

Critics of privatization have suggested that ithas reduced social equity, an unwarranted claim,


as no solid evidence has been produced to dem-onstrate the point. Before the reforms, social ser-vices were regressive, with most of the benefitsaccruing to the middle and upper income classes.Health provision was segregated. Until 1979, theServicio Nacional de Salud (SNS) served blue-collar workers and the destitute through its net-work of hospitals and clinics, while the ServicioMedico Nacional de Empleados served white-collar workers. Only the latter system allowedbeneficiaries to choose between the SNS or pri-vate providers—via co-payments. In education,the main channel for a potential reduction inequity after privatization is an increase in segre-gation. Many private subsidized schools havefocused on the best students, leaving the rest(normally corresponding to economically disad-vantaged families) for the municipal system.Under the old system, however, the best publicschools were able to select their students (andattracted middle class families). Hence, it is notclear if segregation has increased in education.

Moreover, the government created a safety netfor the poor. There is a minimum pension guar-anteed to all workers that have contributed atleast 20 years to the AFP system. The publichealth system serves workers whose income doesnot allow them to buy into a good private plan,as well as the destitute. Finally, most municipalschools are open to all students. Thus, publicexpenditure can be focused on the poor, and thishas already occurred in the pension and thehealth systems and is being discussed for theschool system. Nevertheless, there are indica-tions that choice benefits higher incomehouseholds more than lower income households.The reason is that choice is costly and favors thebetter-informed workers. 44, 45 Current policymakers believe that the poor would benefit morefrom less choice, in exchange for lower costs

and/or more expanded services. This is at leastthe observed trend in regulation. The Auge planguarantees more services and limits costs, andIsapres have responded by reducing the choiceof providers. Similarly, the regulator of the pen-sion fund system is considering a plan where theadministration of large groups of workers is auc-tioned to AFPs, with the expectation that costswould be substantially reduced, due to competi-tion for these pre-packaged groups.

Moreover, it is plausible that privatizationdid not reduce social equity because the pre-1974 situation was very inequitable, and a safetynet, which was previously absent, was built. Webelieve this comparison is relevant for mostdeveloping countries, as their starting point islikely to be similar in terms of an inefficient stateproviding low quality social services that inmany cases do not reach the poor or at leastare very unequally distributed (see World Bank,2003). The fact that richer and more educatedhouseholds are more likely to derive benefitsfrom choice is relatively less important. This isa relatively smaller advantage as compared tothe possibility of capture of social services bythe middle classes that existed in the less trans-parent system prevailing before 1974.

In short, information asymmetries present adifficult problem for individuals that have ledsocial service providers to focus competitionon marketing variables rather than on reducingcosts or enhancing quality. In spite of thisshortcoming, privatization-cum-competition isincreasingly successful, as social services mar-kets continue to improve. This has requiredregulatory improvements to reduce informationasymmetries and eliminate regulations thatdampen competition. However, additional re-forms that would increase competition or catermore to the needs of the poor are still missing.

NOTES

1. This was part of an all-encompassing privatizationprocess, which, in turn, was a building block of thecountry’s shift toward a market economy in 1974.

2. In the past, there was no relation between thecontributions of workers and the benefits they received,so they were perceived as payroll taxes.

3. This point is stressed by Orszag and Stiglitz (2001)and Mesa-Lago (2002).

4. The notion that citizens use shortcuts to get theinformation they need to make appropriate choices iswell documented by political scientists (Iyengar, 1989;Lupia, 1992).

5. Any departure from the sanctioned curricula re-quires ministerial approval.

6. The effect of the ownership change is likely to bestronger when public bureaucrats had considerablescope to pursue their own agenda before privatization.


7. For a detailed account of the pension reform, seeDiamond and Valdes-Prieto (1994) and Mesa-Lago(2004).

8. Self-employed workers can also accumulate funds inthe system, and this provides same tax advantages.However, only 7% of them contribute to the pensionfund system and virtually all of them are high-incomeprofessionals (Arenas de Mesa, 2004).

9. Most of them choose to switch as their pensioncontributions were reduced from 20% to about 15%,raising their take home wage.

10. AFP managers may also charge a fixed fee,allowing for some degree of differentiation amongworkers. However, this fee represents a small fractionof AFP’s revenues. Some authors believe that thispricing strategy is the result of the authorities’ pressurein order to avoid the regressive effects of the fixed fee.

11. The recognition bond pays a 4% annual realinterest rate from the time of the switch to that ofretirement.

12. A shortcoming of the AFP system is its lowcoverage. According to Arenas de Mesa (2004) andArenas de Mesa and Gana (2004), the effective coveragerate of the pension system was 62% in 1975, when theliberalization process began. It fell to 48% in 1980, justbefore the new pension system was introduced, and roseagain to 61% in 2002. Thus, the low coverage rate seemsto be caused by changes in the structure of employment,and is not due to features of the AFP system.

13. Moreover, as benefits were calculated on the basisof the average contributions during the last three yearsbefore retirement as well as the number of years ofcontribution. Salaries were underreported in the initialyears and boosted in the last few years, increasing thegovernment burden.

14. According to Orszag and Stiglitz (2001), this is truefor all private pension fund systems.

15. This is similar to the 8% of contributions that werespent on administration in the 1960s and 1970s, exclud-ing the cost of capital (Diamond & Valdes-Prieto, 1994).

16. Potential entrants have to incur in the cost ofcapturing affiliates from existing AFPs. Moreover, theyhave no information on which are the most attractiveclients, and hence are likely to end with a less profitableportfolio of affiliates.

17. The effective rate of return of workers’ contributionis computed by subtracting the commission, net of theinsurance premium, from the return on the funds.

18. Mesa-Lago (2002) demystifies this claim by signal-ing that workers cannot select specific investmentinstruments or even the profile of their portfolios.

19. Moreover, the government does not report thestandard deviation of past returns.

20. See Webpage of the Superintendencia de AFPs(http://www.safp.cl).

21. A complete, though dated analysis of the Isapresystem and the role of asymmetric information (moralhazard and adverse selection) can be found in Fischerand Serra (1996). See also Bitran and Almarza (2000) foran analysis of equity in the health system and Titelman(2000) for a recent analysis and comparison of the publicand private sectors.

22. High-income workers did not use the public systemdue to its low quality or long waiting times.

23. While Isapres cannot charge differently depend-ing on the health related risks, they are not requiredto accept all applicants. In practice, this means thatan applicant that suffers from a chronic condition(i.e., a condition that increases her expected healthcosts) will not be expelled by his Isapre, but on theother hand becomes captive in his original Isapres andcannot switch, except to Fonasa, which accepts allcomers.

24. Costs in the public system rose faster, as totalexpenditure in Fonasa increased by 290% in 1990–99,while the index of health procedures increased by only22%. Thus, unit costs increased by 141% in the publicsector (institutional) in the same period.

25. The increase in health costs is a worldwidephenomenon. Hall and Jones (2004) show that asincome increases, the marginal utility of other spendingdecreases relative to the increasing value of life, and thisexplains the increasing share of health expenditures innational accounts.

26. Interestingly, in the free choice form of the publicsector, the caesarean rate was the same as in the privatesector (Fischer & Serra, 1996). This suggests that freechoice rather than the affiliation of the health systemleads to overprovision. Gruber and Owings (1996) testthis induced demand model for caesarian sections in theUSA.

27. The Isapres must also ensure against fraud, whichoccurs when affiliates lend their personal identificationcards to non-beneficiaries.

http://www.safp.cl


28. This is characteristic of private health insurance:Cutler (2002) shows that adverse selection in plans isempirically very important.

29. This is even more so for men, since they do nothave the risk and associated expense of pregnancy.

30. Recently, catastrophic risk insurance programshave become more popular, as clients of Isapres decideto cap their payments at the compulsory amount, whilepurchasing insurance programs that guarantee freechoice in the case of a catastrophic health event. Thisshows that consumers are becoming aware of the riskassociated to catastrophic illness.

31. Quoting a former Isapre Superintendent: ‘‘It isdifficult to regulate an industry which grew used (as theIsapres did in the period 1981–89) to being unregu-lated.’’

32. In 1981, only 15% of all primary and secondarystudents attended private subsidized schools (mainlynon-profit religious schools) and 7% non-subsidizedprivate schools. In 2003, the figures increased to 38(mainly for profit) and 9%, respectively.

33. Auguste and Valenzuela (2003) found that compe-tition has had a positive effect in the overall performanceof the Chilean system, improving achievement by 0.4standard deviations. Recent studies for the USA seem toprove that choice improves education (Hanushek &Rivkin, 2003). However, these experiments are for smallpopulations, and adverse effects of segregation onoverall system results do not operate. Woessman(2000) finds that competition has positive effects oneducation using the TIMSS 1999 sample of countries.

34. The Ministry of Education has recently startedpublishing results adjusted by family income and othercharacteristics.

35. Political surveys indicate that the quality ofmunicipal education is not decisive when electing may-ors.

36. The introduction of more flexibility in the Statuteallowed for some rationalization in municipal schools,especially rural schools. A once and for all specialincentive retirement program accompanied the change.

37. Additionally, a voluntary system of accreditation inteaching excellence in subsidized schools was imple-mented in 2003. This year 935 teachers applied forcertification and 409 of them obtained it. Writtenexaminations and observations of classroom perfor-mance are used in the assessment process.

38. Diamond and Valdes-Prieto (1994) asserts that thisis the main benefit of the Chilean private pension system,which otherwise should be thoroughly revamped.

39. A recent survey found that most affiliates areunhappy with their Isapres, but more than 80% declaredthat they do not want to switch to the public system(Adimark, 2003).

40. Of course, their decisions may be rational in aworld where agents have limited information. Moreover,rationality lies in the eyes of the beholder.

41. Though, as we show later, there are other possibleexplanations for this feature.

42. There are pilot programs focused on studentscoming from extremely poor households. A largerprogram is planned for 2006, covering a third ofstudents enrolled in pre-school and in the first fouryears of primary education.

43. Note, however, that in Chile several financialclearinghouses are used to forestall competitive entry.

44. Levin (1998) writes that ‘‘families that are better-offmay be more likely to take advantage of school choice...because of better access to information, greater ability toafford transportation, a higher penchant to exerciseeducational alternatives, and greater generic experiencewith choice and alternatives.’’

45. In the pension fund system, 54% of men and 64% ofwomen have always stayed in the same pension fund.However, younger and higher income affiliates switchfund managers more often. This fact is consistent withthe idea that high-income, better-informed workers aremore able to benefit from competition. Another possibleexplanation is that younger and higher income workersare more attractive for AFPs, and hence, more effortsare made to attract them.

REFERENCES

Adimark (2003). Imagen de Isapres, October, http://www.adimark.cl.

Arellano, J. P. (1985). Polıticas sociales y desarrollo.Chile 1924–1984. Santiago: CIEPLAN.


Arenas de Mesa, A. (2004). El sistema de pensiones enChile: Principales desafıos futuros. In El sistema depensiones de Chile en el contexto mundial y deAmerica Latina: evaluacion y desafıos, ponencias delseminario internacional. Santiago: Oficina Internac-ional del Trabajo.

Arenas de Mesa, A., & Gana, P. (2004). Proteccionsocial, pensiones y genero en Chile. In F.Bertranou, & A. Arenas de Mesa (Eds.), Pro-teccion social, pensiones y genero en Argentina,Brasil y Chile. Santiago: Oficina Internacional delTrabajo.

Auguste, S., & Valenzuela, J. P. (2003). Do studentsbenefit from school competition? Evidence fromChile. Unpublished manuscript, University ofMichigan.

Bitran, R., & Almarza, F. (2000). Equity in the financingof Social Security for Health in Chile. Health Policy,50, 171–196.

Castro, R. (2005). Seguro de invalidez y sobrevivencia:Que es y que le esta pasando. Santiago: Superinten-dencia de AFP.

Cutler, D. (2002). Health care and the public sector.NBER Working Paper 8802.

Diamond, P. (1996). Privatization of social security:Lessons from Chile. In P. Diamond, D. Lindeman, &H. Young (Eds.), Social security, What role for thefuture?. Washington: Brookings.

Diamond, P., & Koszegi, B. (2003). Quasi-hyperbolicdiscounting and retirement. Journal of Public Eco-nomics, 87, 1839–1872.

Diamond, P., & Valdes-Prieto, S. (1994). Social securityreforms. In R. Bosworth, R. Dornbush, & R. Laban(Eds.), The Chilean economy: Policy lessons andchallenge (pp. 257–328). Washington, DC: TheBrookings Institution.

Elacqua, G., & Fabrega, R. (2004). El consumidor de laeducacion: el actor olvidado de la libre eleccion decolegios en Chile. Unpublished manuscript, Univers-idad Adolfo Ibanez.

Fischer, R., & Serra, P. (1996). Analisis economico delsistema de seguros de salud en Chile. Revista deAnalisis Economico, 11, 187–217.

Gruber, J., & Owings, M. (1996). Physician financialincentives and caesarean section delivery. RANDJournal of Economics, 27(1), 99–123.

Hall, R. E., & Jones, C. I. (2004). The value of life andthe rise in health spending. NBER Working Paper10737.

Hanushek, E., & Rivkin, S. (2003). Does public schoolcompetition affect teacher quality? In C. Hoxby(Ed.), The economics of school choice. Chicago:University of Chicago Press.

Iyengar, S. (1989). How citizens think about nationalissues: A matter of responsibility. American Journalof Political Science, 33, 878–900.

Levin, H. (1998). Educational vouchers: Effectiveness,choice, and costs. Journal of Policy Analysis andManagement, 17(3), 373–391.

Lupia, A. (1992). Busy voters, agenda control and thepower of information. American Political ScienceReview, 86, 390–404.

Mesa-Lago, C. (2002). Myth and reality of pensionreform: The Latin America evidence. World Devel-opment, 30(8), 1309–1321.

Mesa-Lago, C. (2004). Evaluacion de un cuarto de siglode reformas estructurales de pensiones en AmericaLatina. Revista de la CEPAL, 84, 59–82.

Mizala, A., & Romaguera, P. (2001). Factores socio-economicos explicativos de los resultados escolaresen la educacion secundaria en Chile. El TrimestreEconomico., LXVIII(4), 515–549.

Mullainathan, S., & Thaler, R. (2001). Behavioraleconomics. In International encyclopedia of socialsciences (1st ed., pp. 1094–1100). Pergamon Press.

Orszag, P. T., & Stiglitz, J. (2001). Rethinking pensionreform: Ten myths about social security systems. InR. Holzmann, & J. Stiglitz (Eds.), New ideas aboutsocial security: Toward sustainable pension systems inthe 21st century (pp. 17–56). Washington, DC: TheWorld Bank.

Parry, T. (1996). Will pursuit of higher quality sacrificeequal opportunity in education? An analysis of theeducation voucher system in Santiago. Social ScienceQuarterly, 77(4), 821–841.

Parry, T. (1997). Achieving balance in education decen-tralization: A case study of education decentraliza-tion in Chile. World Development, 25(2), 211–225.

Rodrıguez, J., & Tokman, M. (2000). Resultados yrendimiento del sector publico de salud en Chile.Technical Report 106, CEPAL, Serie Financiamientodel Desarrollo.

Sapelli, C., & Vial, B. (2001). Evaluating the Chileaneducation voucher system. Unpublished manuscript,Pontificia Universidad Catolica de Chile, Santiago,Chile.

Schneider, M. (1999). Information and choice in educa-tional privatization. In Conference on setting theagenda, Teacher’s College, Columbia University,April.

Superintendencia de Isapres (2004). Resultados SistemaIsapre, April.

Titelman, D. (2000). Las Reformas al Sistema de Saluden Chile, Desafıos Pendientes. Financiamiento para elDesarrollo CEPAL Series, Number 104.

Woessman, L. (2000). Schooling resources, educationalinstitutions, and student performance: The interna-tional evidence. Kiel Working Paper No. 983, Kiel,Institute of World Economics, December.

World Bank (2003). Making services work for poorpeople. World Development Report 2004, September,Washington, DC.

control 3 - fisher gonzalez serra wd 2006_pdf-notes_flattened_201212172252

Documents

provision of social

ecientof social services

ecient supply of services

social servic1

pension plans

private pension fund

arrivets of privatization

privatization of ch