33392 World Bank Pension Reform Primer Coverage The scope of protection in retirement income systems T he coverage of old-age protection systems is a However defined, it does appear that economic central concern in developing countries. While development is the major determinant of coverage. most countries mandate that workers make Indeed, the level of income per capita is an contributions to a retirement-savings plan of some excellent predictor of coverage rates. In contrast, kind, fewer than ten per cent comply in South Asia there is no empirical evidence to suggest that the and Sub-Saharan Africa. Coverage is less then 30 nature of the pension scheme itself has more than per cent in most of East Asia, around 30-40 per marginal impact on the likelihood that individuals cent in mid- dle-income Latin Pension coverage around the world 1 American countries. It reaches an average of 60 per cent in the transition economies of Eastern Europe and the former Soviet Union. In contrast, higher-income Coverage, per cent OECD countries (number of countries) manage to cover 80 80 to 100 (15) per cent or more of 60 to 80 (21) their workforce. The 40 to 60 (15) map (chart 1) shows 20 to 40 (26) the global pattern. 0 to 20 (37) Note: `missing' countries indicate data unavailable. Source: ILO, World Bank The definition applied here does not distinguish will participate. This supports the hypothesis that between those workers that evade the scheme and broader forces are at work and that participation in those that are not legally required to participate. pension schemes must be viewed in the context of Also, the ratio of contributors to the labor force underlying factors that determine the size of the masks potential differences in terms of dormant informal sector. For example, many developing accounts (those with accrued pension rights but countries are characterized by large scale not currently contributing) and in the proportion of subsistence agriculture that is essentially outside of worker's income actually covered which can vary the government's tax net. What drives the informal significantly across schemes. sector is likely to drive pension coverage. This briefing is part of the World Bank's Pension Reform Primer: a comprehensive, up-to-date resource for people designing and implementing pension reforms around the world. For more information, please contact Social Protection, Human Development Network, World Bank, 1818 H Street NW, Washington, D.C. 20433; telephone +1 202 458 5267; fax +1 202 614 0471; e-mail socialprotection@worldbank.org. All Pension Reform Primer material is available on the internet at www.worldbank.org/pensions 2 Coverage Incentives toward informality Contributions or taxes? In many countries, the non-agricultural labor force Viewed through this lens, it becomes clear that is dominated by small firms and self-employed coverage rates in contributory pension schemes individuals who generally do not pay taxes nor may be driven by non-pension factors that lead to comply with other government regulations. informal sector activity. This would help explain Understanding the phenomenon of informal sector why the dearth of evidence linking coverage rates activity is important towards explaining observed to pension scheme design. coverage rates in pension schemes. Nevertheless, poor incentives in a pension scheme Loayza reviewed the empirical work measuring the are likely to exacerbate the situation, especially costs of operating in the formal sector. These when the deduction from wages earmarked to pay costs could be grouped into two categories ­ the for pensions is large (see Figure 2) and when this cost of entry and the recurrent costs required to deduction is viewed as more of a tax than a stay formal. Several studies found that the costs in contribution. terms of time, registration fees and even bribes could be extremely high. One study of Latin Chart 2 shows that social insurance taxes add more America found that the cost of entry was on than 30 percent on average to labor costs in the average ten percent of annual profits of the firm former socialist countries. As these countries and the average time required for completing the moved toward a market economy, smaller firms process was almost one year. proliferated while large state enterprises shed excess labor in the process of restructuring or Recurrent costs of maintaining formal sector status privatization. The combination of a tax collection arose from taxes, regulations and bureaucratic apparatus with little capacity to collect from small requirements (i.e., "red tape"). Hernando De firms and high rates of taxation probably explains Soto's pathbreaking study on the informal sector in much of the decline in coverage observed in the Peru, El Otro Sendero, found that in a sample of 50 region during the 1990s. Farmers have been an small firms, the cost of staying formal represent an especially difficult category following the demise of average of 348 per cent of after tax profits. Some collective agriculture in the region. After individual 22 per cent of this was due to taxes, 5 per cent to plots of land were distributed in the 1990s for higher public utility rates and 73 per cent due to example, only five percent of Albanian farmers regulations, especially labor standards. continued to contribute to the system. Regulations and taxes added an average of 20 per Social insurance/pension taxes cent to labor costs in the same sample of Latin 2 American countries. Another study in Thailand found that informal sector firms saved 13-22 per pensions cent in labor costs by avoiding regulations. Eastern Europe & total FSU (13) The costs of formality are so large that it pays small High-Income OECD (17) firms to devote significant resources to evasion and to forfeit access to legal protection and regulated North Africa & the Middle East (13) credit markets. Hiding from the authorities often also means sacrificing economies of scale which Latin America (19) might otherwise have significantly increased returns East Asia (16) to capital. In short, there are both private and social costs to informal sector activity, but S-Sahara Africa assuming these decisions are rational, it suggests (29) that the costs of moving into the formal sector are 0 5 10 15 20 25 30 35 even higher. %of labor costs Coverage 3 Coverage and pension-system types investment returns. In pay-as-you-go, public Not all countries rely exclusively or even primarily earnings-related plan, the link is less direct but may on contributory pension schemes. Non- be quite strong nevertheless. Contributions are contributory schemes, financed by general revenues typically levied as a proportion of earnings. rather than earmarked taxes are important in a Benefits depend on some measure of pay and years handful of countries including Australia, Canada, of membership of the scheme. However, the Mauritius South Africa and the Nordic Countries. relationship between contributions and benefits In New Zealand, a universal flat pension is the only varies both across countries and across individuals form of direct government intervention in this area. within a country. This is because of rules, such as minimum number of years of membership, normal Because these schemes do not link benefits to and early retirement ages, maximum pensions etc. contributions, coverage is effectively, universal. Some experts have argued that a shift from a Where they play an important role, these schemes defined-benefit to a defined-contribution system fall into two main categories: will, at the margin, encourage some workers to flat benefits to all above a certain age; and participate. This effect comes both from the closer broad social-assistance that ensure that all link between contributions and benefits and from people above a certain age reach a minimum the potentially higher return earned by defined- standard of living. contribution plans in the market than the implicit return that can be earned on pay-as-you-go A well designed non-contributory or `basic' defined-benefit schemes. pension scheme should be considered if the purpose is to reduce poverty among the old and But this effect may be overstated. Few pension resources allow. (See the Primer note on `basic schemes can offer returns comparable to pensions'.) alternative investments in a capital scarce environment (witness the high rates of return in But most of the world still relies mainly on micro-finance or to investment in education). And `contributory' schemes. Historically, this outcome lower income households have especially high was probably driven by a second, more paternalistic discount rates and liquidity needs. Finally, it would motivation for government intervention ­ myopia. be inconsistent to argue that the same myopic In addition to providing an absolute minimum level individuals that had to be forced to save for old of income during old age, in most cases age, would turn around and assign a high value to policymakers advocated replacement rate targets distant payoffs from their pension scheme, that prevented sharp relative declines in regardless of whether that payoff came from a consumption at the end of one's life. Individuals public DB scheme or a private DC scheme. were considered to be too myopic to save for their own old age or by other accounts, so farsighted We conclude that any pension system that relies that they might save less due to the presence of exclusively on contribution based schemes will basic pension schemes funded by general revenues. experience incomplete coverage. Moreover, the Forced savings at levels that would `replace' a levels of evasion will be strongly related to the size significant fraction of pre-retirement income was of the informal sector, itself a function of many considered necessary to smooth consumption over other factors that may have little to do with an individual's lifetime. Annuitization was required pension provision. This sobering assessment to make sure that one did not outlive these savings. suggests that coverage rates are likely to increase very slowly until underlying structural problems in As a result, most countries emphasize schemes that the labor market and tax administration are link benefits to previous contributions. In funded improved. Ambitious plans to extend coverage defined-contribution plans, the link is immediately should also take into account the perspective of evident because the value of the pension depends potential participants who may behave very directly on the accumulation of contributions and rationally in the face of onerous costs of formality. 4 Coverage Is expanding coverage always best? Economic aspects: expanded coverage might The International Labor Organization (ILO) aid labor mobility and flexibility by encouraging addresses the coverage issue in its series of people to move out of state-owned enterprises conventions on social protection for old age. The that are restructuring or into smaller employers 1967 convention (number 108) states that schemes when previously this would have entailed the should cover: loss of pension coverage. `all employees, including apprentices; or Financial resources: new contributors will im- `prescribed classes of the economically active prove the finances of pay-as-you-go pension population, constituting not less than 75 per schemes in the short term, especially if cover- cent of the whole economically active popula- age is extended to `dynamic sectors of the tion' (where `prescribed' means `determined economy, e.g., those established in the suburbs by...national laws or regulations'. of major cities, which primarily recruit young workers from the rural surplus labor force'. The convention permits the exclusion of casual workers (and others) and allows temporary limits to This last rationale in our view, is seriously flawed. coverage to: Expanding coverage to `young workers from the `prescribed classes of employees, constituting rural surplus labor force' is likely to result in a not less than 25 percent of all employees; or regressive transfer from lower to higher income `prescribed classes of employees in industrial individuals. The last people to join the scheme are undertakings, constituting not less than 50 per likely to get the worst deal when the Ponzi scheme cent of all employees in industrial finally comes to an end. undertakings'. But the key point is the second, namely the need to The basis for these targets is not clear and it is provide social protection to workers that would unlikely that country-specific circumstances can be otherwise be excluded and therefore would fully respected by such broad criteria. But experience a loss of welfare. It is a message generally, the message is that most countries should predicated on the view -- in Mr Gruat's words -- be trying to expand the coverage of their pension that `there is a prominent interest from non- scheme. covered categories for this type [sickness, accident, old age, disability] of protective measures'. This Mr J.-V. Gruat, head of social security planning, leaves `no other alternative...but to expand...the development and standards at the ILO, presents scope of coverage of existing schemes...failing four reasons for extending the scope of coverage which the social cohesion of the country as a whole of social protection: may well be put into jeopardy.' Legal requirements: `In most States where the rule of law prevails, ...basic legal But while there is evidence of latent demand for instruments very commonly encompass the group insurance ­ for example, the existence of following two principles... non-discrimination; thrift or mutual societies ­ there is little to suggest and relieve want, prevent destitution.' that vast numbers uncovered workers are anxious Together, it is argued, these justify the coverage to participate in government-run schemes. Rather, expansion. evasion appears to be rampant. Workers choose to Social considerations: `...when classes of the `exclude' themselves. In addition to the costs of workforce remain excluded from forms of so- joining the formal sector mentioned above, a cial protection they feel are essential to their number of factors specific to the pension system well-being' then `social cohesion becomes itself could contribute to this attitude. threatened' as people feel discriminated against. Other commentators see participation in One obvious reason to stay out is poor governance. pension schemes as an important part of the Contributors to the provident funds in Kenya, formalization of the labor force, which in turn Uganda, Sri Lanka and Zambia received dismal should improve working conditions. returns over the last few decades and would have Coverage 5 done better with their money in bank deposits. more effective use of limited resources than say, Low income workers, who usually lack access to favorable tax treatment for voluntary pension credit markets, could lose even more by contributions that cater to high income individuals. participating in these schemes as they forfeited the chance to invest in land or tools or a vehicle. Another strategy is to offer incentives to providers that expand coverage through their own efforts. In A more subtle reason for avoidance would suggest Bolivia, a five-year concession for two companies an even greater degree of rationality. Workers may to manage the new pension system included an be instinctively aware of the well- documented, incentive to expand coverage. Coverage has risen positive correlation between income and life over the period. But the incentives were geared expectancy which would greatly reduce the towards gaining new affiliates and some worry that perceived value of longevity insurance for many. accounts may lay dormant once they are opened. Others may realize that short vesting periods allow them to maximize the returns from the scheme Perhaps the greatest challenge lies in convincing after only a few years of contribution. the self-employed to participate. Here a useful distinction can be made: on the one hand, there are Finally, the subtle interaction between pension doctors and lawyers whose incomes are high and schemes and traditional family support systems who cannot afford to forego the benefits of being should not be ignored. (The extensive literature on in the formal sector. The vast majority of the self- this topic is surveyed in a separate Primer Note). employed however, are individuals offering services Theory suggests that much of the welfare gain such as vendors or casual labor. There are millions from an annuity can be obtained within a family of such small-scale enterprises in developing unit ­ kind of intrafamily pension scheme. In countries. The fixed costs of registration in the many societies, this family support system is more formal sector can be prohibitive and monitoring reliable than what the state currently has on offer. them next to impossible. Figure 3 below shows the relationship between firm size and participation in New approaches to coverage the pension scheme for Peru based on household The preceding discussion suggests that coverage survey data in 1997. Firms with more than 100 expansion should be viewed from the perspective employees are 4.5 times more likely than the of the participant. At the root of the problem is average firm to participate, while micro-firms are the excessive burden of joining the formal sector only 0.2 percent as likely. This typical pattern is and this can only be addressed in the context of repeated across the world. broader reforms. In addition, it is clear that forcing more workers to participate in a scheme that is Coverage by firm size in Peru 3 unsustainable or provides very poor returns on investments is not good policy. These important 5 caveats aside, there are some policies specific to 4.5 pension design that may help increase participation. 4 3.5 The most direct approach is to subsidize coverage 3 by providing a matching contribution to an 2.5 individual account. This is done in the new, egarevocegareva 2 defined contribution scheme in Mexico and in the of 1.5 voluntary pension scheme in the Czech Republic. 1 In Mexico, the government deposits a flat amount ipleltuM0.5 equivalent to about 2.2 percent of the average wage 0 into individual defined contribution accounts. For Up to 5 6-20 21-99 Over 100 low income workers, the match is quite significant. Such a policy could be expensive. But it may be a Number of workers 6 Coverage Even in the United States, where tax compliance is Industry-wide schemes have been introduced in high, more than half of the self-employed are Hong Kong under the new Mandatory Provident thought to evade contributions to Social Security. Fund (MPF) system. These cover the transient The figures look much worse in other countries: in workers in the construction and entertainment the first year that the self-employed and small firms industries. The fund managers and recordkeeping were covered by law in Korea, only about one third agents are selected by the regulator based on strict declared any income. Recent figures reveal that criteria and competitive bidding. only one in 20 self-employed contributed in Argentina and in Turkey, only one percent of the Better service, lower costs members of the special scheme for self-employed Group schemes may reduce transaction costs to a workers were making contributions. certain extent, but even small costs can constitute a major barrier for those with small amounts to save Many of the small firms or self-employed and accumulate. Providers may find lower income experience seasonal or transient labor market workers an expensive proposition. Branch offices participation. They may have employers, but these in remote areas may be costly and individuals with contracts are of short duration and are not easily limited financial sector experience may need extra subsumed within a legal and administrative attention. If fixed costs are high, it will simply not framework often designed for larger firms. pay to market to small investors or even to groups Construction workers tend to fall into this whose members have low incomes. category. Like the low income self-employed, it can be costly to keep track of their income at an One strategy for dealing with this problem has individual level. already been mentioned. Using the recordkeeping infrastructure of existing groups, like MFIs, could In some countries, the initial criteria recognized the reduce costs, but would have to be accommodated difficulty of incorporating smaller firms into the within the legal and administrative framework. system by setting a minimum firm size. In India, Another approach is to cross-subsidize smaller for example, the initial limit on the number of accounts. This can be done by prohibiting certain employees in firm that triggered legal coverage was types of charges, such as flat fees, in such a way 50. This was later reduced to 20 and recently there that providers cannot separate fixed and marginal have been proposals to reduce it to 10. Statutory costs when they charge customers. coverage expansion is unlikely to work by itself, however. Some countries have recognized the There is another type of cost however, that rarely futility of mandates on the self-employed and have enters into the ongoing debate on this topic (see simply made participation voluntary. This is true the Primer note entitled, `Administrative Charges'). for example in Chile, Colombia and Mexico. This is the time cost and even pecuniary costs involved in obtaining information and eventually Perhaps a more promising target for coverage benefits from institutions that provide pensions. expansion can be found in existing groups. Industry-wide schemes can reduce the transaction Even when explicit costs are low, service standards costs for individuals in the same sector. A study by may not be adequate. India's Employee Provident the Superintendent for Private Pensions in Peru Fund (EPF) for example, has not provided millions suggested working with non-governmental of its members with annual statements for several organizations (NGOs) to expand coverage in the years. Individuals seeking to remove their funds in pension scheme. A similar proposal has been made accordance with the law are frequently faced with a in India. Micro-finance institutions (MFIs) in sea of bureaucracy or worse. Accounts may be lost particular, might facilitate the recordkeeping and upon the death of a worker with his family never transfer of monies, thereby reducing unit costs of receiving the balances. A letter from one the transactions. disgruntled member illustrates the frustrations felt by members of a monopoly: Coverage 7 `Dear Sir (September 1984) Unfortunately, the answer remains obscured by limited experience, poor data and definitional Kindly let me know the balance lying in problems that make comparisons over time or my above account after transfer of my funds from across countries difficult. A/C No. ##. I very much regret to inform you that For example, Chilean pension experts Rodrigo since 1972 (when I left Assam i.e., my previous Acuña and Augusto Iglesias cite an increase in employer XXX), I am trying to get my account coverage in Chile between 1982 and 1997 from 53 position and statement of my funds without any to 70 percent of the labor force. Others note that effect from your side. pre-reform coverage (in the 1970s) was even higher Initially, there was a confusion and your and that many AFP affiliates do not contribute office had informed me that you had not received regularly. But another researcher, Claudio "K" form from your Calcutta office while the Chamorro, explains most of the discrepancy Regional Provident Fund Commissioner, West between contributors and affiliates as a result of the Bengal had informed that "K" Form was duly sent growing population of `affiliates' who are actually to you on 22.6.74 and 14.8.75 via their letter pensioners and the fact that certain important number TR/2099/WB/X/607 dated 19.11.75. groups such as the self employed are not required Under the circumstances, non-receipt of to contribute. Among those legally required to statement of account and non-confirmation of participate, compliance was close to 95 percent. transfer of my fund from West Bengal to Assam by your office has put me in great trouble to Coverage in Argentina has been falling for several withdraw my Provident Fund amount from your decades. In their 2002 study, Argentine researchers office. It is a pity that innumerable reminders Jorge Colina, Lucas Ronconi and Mariano from the side are evoking no response from you. Tommasi used household survey data to test for In the absence of any communication from your the presence of a structural change in coverage due side within 10 days on receipt of this letter shall to the introduction of the new, private pension constrain me to take legal action. scheme in 1994. The study provided econometric Would you kindly personally look into the evidence that the "costs associated with the labor matter and let me know why your office should code, collective bargaining and social security have take over 12 years to confirm such a simple a negative impact on coverage, especially for those matter.' workers with low productivity." However, they could not identify any statistically significant impact on coverage due to the introduction of the new This particular individual finally did receive his private pension scheme. The concluded that this provident account balance ­ 16 years later. The reform appears to have had no impact or at most EPF, like many other public pension institutions an impact much less important than other factors around the world, is currently undergoing unrelated to the design of the pension system. administrative reforms intended to improve service to members. However, in the absence of While there is general agreement that incentives to competitive pressures, past experience suggests that participate in publicly-mandated pension schemes poor service, among other factors, will continue to can be improved through better design, there is dissuade potential contributors. little empirical evidence to suggest that these are crucial determinants of formal sector participation. Systemic reform and coverage Many other factors ranging from rigidities in the In the past few years, more than twenty countries labor market code, minimum wage rules, have replaced part or all of their publicly-managed transaction costs associated with labor registration DB schemes with private, individual funded and even confidence in government institutions are accounts. Has this lead to higher coverage? likely to be just as important. Thus, applied in isolation, systemic pension reform is not likely to lead to a significant increase in coverage. 8 Coverage Further reading Conclusions and recommendations Acuna R. and Iglesias, A. (2001) `Chile's pension reform after 20 years', World Bank Pension Coverage rates track income levels reform primer working paper, forthcoming. closely and evasion is largely driven by Colina, J., L. Ronconi and M. Tommasi (2002) the high cost of joining the formal sector `Problemas para la expansión del grado de Pension scheme design can exacerbate cobertura en el sistema reformado de pensiones the evasion problem when "contributions" Argentino', Centro de Estudios para el are large and perceived as taxes Desarrollo Institutional (CEDI). A poorly designed and managed scheme Chamorro-Carrizo, Claudio (1992), "La cobertura should be reformed prior to attempts to del sistema de pensiones Chileno", Tesis de expand its coverage Grado No. 107, Pontifica Universidad Catolica Extending financial solvency of a pay-as- de Chile. you-go scheme is not a good rationale for Corsetti, G., and Schmidt-Hebbel, K. (1994), `An expanding coverage endogenous growth model of social security A safety net can help cover the inevitable and the size of the informal sector', Policy Re- gaps in a contributory scheme search Working Paper no. 1471, World Bank. Defined contribution schemes tend to Cuesta, J., Holzmann, R. and Packard, T. (2000), provide better incentives for coverage, `Extending coverage in multi-pillar pension but the positive impact of scheme design systems: constraints and hypotheses, prelimi- on coverage rates should not be nary evidence and future research agenda', overstated as there is little evidence to Social Protection Discussion Paper no. 0002, support these claims to date World Bank. creative approaches to expanding De Soto, H. (1989), `The Other Path', Harper and coverage include direct matching Row, New York. contributions for low income workers Gruat, J.-V. and Thompson, K. (1996), `Extending and finding ways to reduce transaction coverage under basic pension schemes: general costs by harnessing existing groups and and Chinese considerations', Issues in Social pooling individuals where possible Protection Discussion Paper no. 4, Interna- tional Labor Organization, Geneva. International Labor Organization (1967), Invalidity, Old-Age and Survivors' Benefits Convention (no. 108), Geneva. aymployer ent James, E. (1999), `Coverage under old-age security perient oe specifi odic pr em ed age programs and the uninsured: what are the is- n.1.nm sues?' Policy Research Working Paper no. etc 2163, World Bank pe´np. by siongoveron abov pers r-for´m Loayza, N. (1997), `The economics of the informal sector: a simple model and some empirical on retire madment o e es r to evidence from Latin America', Policy Research Working paper #1727, World Bank. Superintendencia de Administradoras Privadas de PENSPRIMERORM IORNEF Fondos de Pensiones (1999), "Ampliacion de la cobertura provisional en el SPP", Estudios equip pmerson with information pr er n. 1. elementary book to ofimp.e)rfections,faultsorerrors betterby removal orabandonment v.t. & i. 1. make (institution, procedure previsionales, No.1, Lima Peru. Van Ginneken, W. (1999), `Social security for the informal sector: a new challenge for the devel- oping countries', ISSA Review.