Institutions

Abstract

This chapter is concerned with the design, implementation, and outcomes of current social protection institutions in the region. It examines in turn the three core social protection institutions: occupational pensions, individual retirement savings plans, and social assistance. Occupational pensions in Latin America are restricted to selected groups of workers, mainly skilled workers in large firms. They are also highly fragmented and dependent on public subsidies, although the level of subsidies is increasingly contested. Recent trends in labour force participation in occupational pensions suggest stagnation during this century. Individual retirement savings plans are based on individual workers’ saving capacity. They were initially implemented in ten countries in the region, under different modalities. Individual retirement savings plans have not proved successful in the region, and it is fair to say their institutional structures are in transition. Social assistance greatly expanded in the new century. Most countries in the region have implemented large scale old age transfers and conditional income transfer programmes. By 2015, social assistance transfers reached around one third of the population in the region.

This chapter is concerned with the design, implementation, and outcomes of current social protection institutions in the region. As discussed in the Introduction, the book will focus on three core social protection institutions: occupational pensions, individual retirement savings plans, and social assistance. These three institutions are the core of social protection provision in the region, secondary or complementary social protection programmes are often offshoots of them and serve a sub-group of their participants. The three main institutions reflect diachronic dimensions, their origins and evolution, as well as synchronic dimensions in their deployment across countries. They also provide a fertile and effective entry point for the comparative analysis of social protection in the region and theory development.

The chapter is organised as follows: Section 4.1 reviews indicators and data, probing the strengths and weaknesses of survey and administrative data. Section 4.2 studies pension schemes covering in turn occupational pensions and individual retirement savings plans. Section 4.3 examines social assistance, including conditional income transfers and old age transfers. Section 4.4 offers a brief review of secondary provision covering unemployment insurance, family and children, and active labour market policies. A final section concludes.

4.1 Data and Indicators

It will be helpful to start with a brief discussion on social protection data sources and indicators. There are currently two main sources of data on social protection institutions: household surveys and administrative registers, each have strengths and limitations.

Administrative data on social protection schemes are collected by the relevant agencies with the objective of selecting participants and of monitoring implementation, entitlements, and costs. Administrative data on participation are affected by gaps and errors associated with lags in record updating (Rofman & Oliveri, 2012). Where a multiplicity of agencies are involved, gaps and data harmonisation can be an important issue (Arenas de Mesa, 2019). Consistent historical administrative data is scarce due to repeated institutional change affecting the relevant agencies. Cross-country administrative data requires a common approach to data collection and presentation by the countries involved.

Regular collection of household surveys by the countries in the region began in the 1990s and has become an important source of information for social protection researchers. Household surveys provide regular, consistent, and up to date information on participation in social protection programmes. However, they have several limitations. First, they are subject to limitations in their sampling, associated with groups harder to reach such as the very rich or the very poor, and with people living in insecure communities or remote rural areas. Sampling limitations can have a measurable impact on estimates. Second, in common with all household surveys they are subject to errors and omissions in reporting. Third, restrictions in the time available for respondent interviews imposes restrictions on the selection of social protection programmes included in the collection of information.

Comparison of administrative and survey data on social protection programme participation suggest their limitations can bias their information in different directions. Villatoro and Cecchini (2018) compared administrative and survey data capture of social assistance participants. They found that survey data underestimated participation in conditional income transfers (old age and disability transfers) by 13% (21.9%) on average. They attributed the disparities principally to household survey sampling deficiencies.

In the literature, participation in social protection schemes is usually measured through indicators of coverage. The latter are used as an indicator of the reach of social protection schemes, that is the population they benefit. The term coverage has its provenance in insurance approaches to social protection. A standard insurance scheme draws a premium from all members to compensate the fraction of them affected by the insured contingencies. Coverage, measured as a proportion of the labour force or population, indicates the proportion of that group that stands to be compensated by the scheme. The ILO’s measure of ‘legal’ coverage comes close to capturing this narrative (ILO, 2021). In social assistance programmes, reach is a more appropriate measure of participation than coverage.¹

In the Latin American context, ‘legal’ coverage is very limited as an indicator of the reach of social protection programmes. In part this is because in practice not all workers with a legal entitlement might be protected. In unemployment insurance schemes, for example, there are additional requisites workers might be expected to meet to access entitlements: to have continuous employment prior to the separation, a minimum set of contributions with a specified period. Behavioural responses might prevent access to entitlements. Even if explicitly included in the programme, workers are likely to stay away if they expect to find it difficult to meet these requirements. Occupational pensions often exclude workers who are not in a regular and permanent employment relationship. Own account workers, unwaged workers, workers in short-term or intermittent employment, domestic, and rural workers find it hard to access unemployment protection. In practice, ‘actual’ coverage of insurance programmes is likely to diverge substantially from ‘legal’ coverage. Examining Chile’s individual savings unemployment insurance, Sehnbruch et al. (2020) find that only 11.8% or workers with a permanent contract were actually entitled to make a claim and only 2.9% of workers with fixed term contracts were entitled to do so. Ceilings on the period time covered by unemployment insurance, say 3–6 months, represents a further restriction on ‘coverage’. It is more accurate to describe ‘legal’ coverage as ‘conditional’ coverage because it is conditional on meeting a battery of programme requisites. In pension schemes, vesting periods describe the length of time participants need to make contributions before accessing pension benefits. Assessing coverage on an annual basis does not tell us whether workers will reach the age of retirement with 20–30 years of contribution required to access entitlements.

A measure of the share of workers participating in pension programmes is described as ‘active’ coverage, whereas a measure of the share of persons over the retirement age who are in receipt of a pension benefit is described as ‘passive’ coverage. These measures are calculated or estimated on an annual basis, often at the end of the year. As will be discussed in more detail below, these measures require contextualisation in the context of Latin America. Due to large volatility in employment and vesting periods, ‘active’ coverage could significantly over-estimate the proportion of workers who will be able to access pension benefits at retirement. This will be discussed in some detail below. ‘Active’ pension scheme coverage does not provide a reliable measure of old age income security, for example. On the other hand, ‘passive’ coverage does provide a reasonable measure of old age income security, assuming minimum pension benefits are adequate. The generosity of social protection programmes is usually measured by the share of earnings they are likely to replace, or replacement rates.

Improvements in the availability of cross-country administrative and survey data on social protection participation in recent times has facilitated comparative research. Issues of data comparability across survey and administrative data and lack of precision in some of the indicators in common use recommend care should be exercised in their interpretation.

4.2 Pension Schemes

This Section provides a description and analysis of pension schemes, including occupational pension schemes and individual retirement savings plans.² Old age income security has been the dominant concern of social protection policy in Latin America since the emergence of occupational pension funds. Pensions are the largest component of social protection expenditure, enjoy a high degree of institutionalisation, and monopolise the attention of policy makers (national and international) and researchers. They signal a strong age bias in the development and current structure of social protection provision in Latin America.

Table 4.1 summarises the institutional set up of pension schemes in the region. The distinction between schemes based on defined benefits and those based on defined contributions serves as the main organising concept. Defined benefit schemes establish a formula to calculate retirement benefits, normally based on years of service and final salary parameters. Defined benefit schemes in the region are unfunded, or pay-as-you-go, with current pension contributions and public subsidies financing current benefits. They are managed by public or semi-public agencies and are stratified for different occupations. Defined contribution pension schemes, on the other hand, set the contribution rate as a share of labour earnings. Savings are collected in individual retirement accounts managed by private or public pension funds. At retirement, workers use their accumulated balances to make pension arrangements. The options include purchasing an annuity from an insurance provider guaranteeing a fixed monthly income during retirement or agreeing a schedule of withdrawals from the pension funds.

Table 4.1 Pension schemes in Latin America

As can be seen from the Table, countries in the region show a variety of arrangements. Nine countries have defined benefit schemes only—Argentina, Brazil, Cuba, Ecuador, Guatemala, Honduras, Nicaragua, Paraguay, and Venezuela. Five countries have defined contribution schemes only or are transitioning to it—Bolivia, Chile, El Salvador, Dominican Republic, and Mexico and Panama.³ Five countries have both defined benefit and defined contribution pension schemes deployed in different combinations. Colombia and Peru have competing defined benefit and defined contribution schemes, meaning that workers can be in one or the other but not both. Costa Rica and Uruguay have defined benefit schemes as a first pillar with defined contribution schemes as a second pillar. Some countries that initially embraced individual retirement accounts have subsequently reformed them, a process referred to as ‘re-reform’ (Mesa-Lago, 2020). In 2008 Argentina transitioned back from individual retirement plans to a defined benefit scheme. In 2010 Bolivia opted to transition to a public defined contribution scheme, not yet fully implemented.

The Table shows the share of the active labour force contributing to pension schemes, with the numbers in brackets indicating the share in defined contribution schemes for those countries that have them. Uruguay stands out with just over 80% of the active labour force contributing to pension schemes. In Chile, Costa Rica, and Panama, around two thirds of the active labour force contribute to a pension scheme, with Argentina and Brazil just behind them. In the rest of the countries, between a fifth and a quarter of the active labour force are engaged in pension schemes. Overall, participation in pension schemes is higher in the more economically developed countries of the region. For the region taken as a whole, less than one in two workers contribute to an occupational scheme or individual retirement account.

The numbers in brackets show the share of the active labour force saving in individual retirement plans, where the two types of pension schemes coexist. Aside from Chile and Costa Rica, individual retirement plans reach a small proportion of the active labour force. In Peru and Panama, the share of the active labour force contributing to individual retirement plans is very small.

Figure 4.1 draws on administrative data to provide a picture of pension programmes participation rates for Latin American countries for the period 2000–2018. There is a group of countries showing sustained higher rates of participation of the labour force. They are Argentina, Brazil, Chile, Costa Rica, Panama, and Uruguay. They have participation rates around the 60% mark for a significant portion of the period. Unsurprisingly, aside from Costa Rica these countries experienced deeper industrialisation and early implementation of occupational insurance funds. Another group of countries have participation rates at around 40% for significant portions of the period. They include Ecuador, Mexico, and Venezuela. A third group of countries have very low participation rates, around the 20% mark. They are Central American and Andean region countries.

Fig. 4.1

Pension scheme contributors as fraction of the economically active population. Data sources: All Contributors from Arenas de Mesa (2019); Individual retirement savings plan contributors (Asociación Internacional de Organismos de Supervisión de Fondos de Pensiones, 2022)

The Figure shows contributors to pension schemes, and participants in individual retirement savings plans, as a fraction of the economically active population.

Overall, participation rates paint a frozen picture of pension programme coverage in the region. Uruguay and Brazil are perhaps the exception. Their participation rates rising during the first decade of the new century but registering a decline in the second decade. Chile shows a sustained decline over the period. There are few changes in group membership for the period, with only a weak rising trend for some countries in the middle of the period that stabilises around 2015.

The Figure shows changes in participation rates in individual retirement savings plans as a separate trend. This series can be expected to rise over time, as the newly introduced individual retirement savings plans attract new entrants to the labour market and, depending on the rules, workers switching to the new programmes. However, the Figure does not show a sustained rising trend in individual retirement savings plans participation. In countries where individual retirement savings plans are intended to substitute existing defined benefit programmes, Bolivia, Chile, Dominican Republic, El Salvador, Mexico and Panama there is no observable net increase in participation. In Costa Rica, and Uruguay, where individual retirement savings plans are complementary to existing defined benefit programmes, participation rates have increased as expected. In countries like Colombia and Peru, where defined contribution pension programmes are in direct competition with defined benefit programmes, there is very little change over time.

Figure 4.2. draws on household survey data to measure participation. Participation rates are captured by the share of workers who report having pension rights. The participation rates are based on household survey data harmonised by CEDLAS (2012). The survey data enable measuring participation rates for all workers and separately for salaried workers. Participation is expected to be higher for salaried workers as non-salaried workers face restricted access to pension schemes.

Fig. 4.2

Share of workers with pension rights. Data source: CEDLAS (CEDLAS & World Bank, 2012)

The Figure shows the estimated share of workers who report having pension rights in household surveys. It shows pension scheme participation rates as estimated in survey data.

The participation rates estimated from survey data confirm the general trends observed with administrative data. Country groupings also remain the same. Participation rates are higher when only salaried workers are considered, underlining the exclusion of workers in informal employment from pension schemes. Widening or shrinking gaps between the two series provide information on changes in the share of workers in informal employment in the countries of the region or changes in access to pension schemes.⁴ In the interest of comparability, it makes sense to focus on participation rates for the labour force. The longer span of the survey data helps to contextualise some of the trends observed from administrative data shown in Fig. 4.1. In Argentina, for example, coverage appears to have a ‘U’ shape, with participation higher at the beginning of the period than at the end. The relaxation of access to entitlements following the 2001 financial crisis might be responsible for the upward trend in participation in the latter period.

Pension participation is higher for males, workers with higher educational qualifications and higher labour earnings, and workers in larger firms (CAF, 2020; Eslava et al., 2021; OECD/IDB/The World Bank, 2014). This applies to all pension programmes in the region (Rofman & Oliveri, 2012). These correlates of pension participation are common to countries in the region and are relatively constant in time. In sum, participation in pension schemes in Latin America shows significant stability over time, even in countries that have implemented far-reaching pension reforms.

The rest of the section discusses features and issues that are specific to each pension programme design and then examines issues they have in common.

4.2.1 Occupational Pensions

4.2.1.1 Bismarck vs. Beveridge

In the more industrialised countries in the region, occupational pension schemes originated in worker led self-protection institutions. State-led developmental policies, and the first wave of worker incorporation associated with it, explain to government attempts to influence, and control over the independent insurance funds. This involved changes in the financing of occupational pension funds to a pay-as you-go-basis and efforts to consolidate and streamline the governance of the schemes. According to Mesa-Lago (1989), the occupational pension funds emerging in intermediate countries—Andean, Costa Rica and Panama—from the 1940s were controlled by governments from the outset. They attempted to follow more closely Beveridgean guidelines, advocated by the ILO, than the Bismarckian logic in the pioneer countries. However, surveying occupational pension funds today it is hard not to come to the conclusion that Bismarck won the contest (Bertranou et al., 2018).

Government led reforms, motivated by financial considerations, sought to streamline the governance of occupational pension funds, but their fragmentation has proved resilient. What is sometimes presented in the literature as ‘social insurance systems’ are nothing of the sort. ‘Systems’ are in fact common governance structures enabling large differentials in access and entitlements for different groups of workers, military and police, civil service, teachers, etc. Large exclusions, implicit and explicit, remain in the face of repeated attempts at improving inclusivity in the 2000s (see Arenas, 2019). Even where common governance structures apply to the unified pension programmes, some workers have access to special voluntary complementary programmes. Mesa-Lago concludes that “the process of unification of multiple contributory pension programs and schemes in the region has been difficult, prolonged and uneven” (Mesa-Lago, 2007, p. 83). Mesa-Lago places Argentina, Colombia, El Salvador, Mexico, Uruguay, Brazil, Honduras, Paraguay, and Venezuela in the group of countries with highly segmented pension programme provision (See Table 4.1). This is at the core of large inequalities in pension benefits despite common governance administrative structures.

4.2.1.2 Membership Entitlements

To the extent that occupational insurance funds are insurance instruments, they emphasise ex ante resource pooling to address idiosyncratic risk and ex post redistribution from the fortunate to the less fortunate. They reflect strong incentives to form common groups with similar risk profiles. As such they reinforce status differentials in the labour market. Their underlying logic of entitlements based on group membership follows directly from their stratified risk pooling (Baldwin, 1990). It explains the resilience of pension programme segmentation in the face of sustained governmental attempts at consolidation and streamlining.

In practice occupational insurance funds are an instrument of employment stratification. Occupational insurance funds reinforce intra-class differentials within wage earners. This is just another way of making the same point, but with the advantage that it throws light on the employment effects of social protection. Their stratification effects work in two ways. First by excluding groups of workers with less advantageous risk profiles: agricultural workers, domestic workers, urban informal workers, workers in small firms, women, and new entrants (Garay, 2021). Latin American governments have over time made repeated efforts to reduce legal and regulation barriers to the participation of these workers in pension programmes, with mixed success.

Second, pension scheme design features often act as a deterrent to the participation of outsiders despite common governance. For example, long vesting periods, the density of contributions required to access pension benefits, preclude workers in intermittent formal employment accessing their entitlements. Long vesting periods redistribute pension funds from workers with shorter than vesting period contributions at retirement to workers with permanent and secure employment who will find it easier to meet the vesting requisite.⁵ This applies especially to women who are likely to have large contribution gaps.

There has been a recurrent argument in policy debates as to why Latin American countries failed to extend the reach of social protection schemes along the lines of European welfare states. Leaving aside for the moment the issue whether all, or indeed most, European countries really achieved this standard (Marx & Nelson, 2012), it is not a productive way to approach this issue. It is the same as asking why Latin American countries failed to construct welfare states. Acute differentiation of wage earners since the 1980s has removed the possibility that homogeneous risks pools among wage earners could have created the conditions for the emergence of inclusive social protection institutions.

4.2.1.3 A Less Advantaged Worker Perspective

It is informative to consider participation in pension schemes from the perspective of less advantaged workers. By less advantaged, I mean workers who are less likely to benefit from stable, continuous, and reasonably well-paid employment. For advantaged workers’ pension schemes participation decisions take the conventional form of paying contributions to secure a pension benefit related to their final earnings in accordance with the scheme entitlement formula (say 1/80th of final salary per year of service). For less advantaged workers, on the other hand, it makes sense to look at pension schemes participation decisions as a bet - an option in financial market terms—on securing a guaranteed minimum pension benefit. This is an important distinction because for the less advantaged worker pension participation decision is a binary choice, securing a guaranteed minimum pension or not. It is not surprising that for workers in Latin America who fall in the disadvantaged group, the regulations on the minimum pension are all that matters. In fact, as we will find in the next chapter, in Latin American countries most pension recipients access the minimum pension guarantee.

Altamirano et al. (2018) study replacement rates from participation in pension schemes in Latin America and the Caribbean. They simulate replacement rates as predicted to apply to workers with average employment and pay in each of the countries.⁶ They find that defined benefit pension schemes in Latin America and the Caribbean have a replacement rate of 68.7% on average while defined contribution schemes have a replacement rate of 38.9% on average. They calculate the net subsidy to pension recipients as the difference between (i) the benefit they are likely to receive according to pension scheme entitlement regulations; and (ii) the pension generated by savings equivalent to the contribution rate accumulated at a fixed rate of interest. The net subsidy to pension recipients, in percentage points of the replacement rate, turns out to be 28.3 pp for defined benefit programmes and 12.4 pp for defined contribution programmes. Given the design parameters of the simulation, the net subsidy necessarily comes from two sources: provisions for guaranteed minimum pension leading to pensioners getting a higher benefit than that merited by their contributions; and redistribution from participants who at retirement fail to meet the requisites for securing a benefit in defined benefit programmes. In defined contribution pension schemes only the first source of a net subsidy applies, that is minimum pension guarantees. In defined benefit pension programmes, most of the net subsidy is in fact a net tax on workers who fail to meet access conditions. The findings from this study underline the adverse redistribution inbuilt into defined benefit programmes faced by less advantaged workers.⁷

In sum, occupational insurance funds are designed to redistribute resources away from workers with low density of contributions and towards workers with high density of contributions. In the Latin American context, this involves a redistribution of pension benefits from the mass of disadvantaged workers to a minority of advantaged workers.

There are sustained pressures on occupational insurance funds in the region. Secular demographic change and changes in family structures threaten the male breadwinner model under which Bismarckian pension schemes were designed. Rising contribution rates, the ending of service pensions, and an extension of the retirement age, could contribute to counteract the effects of ageing populations, but they generate significant resistance. Changes in employment are also a significant threat. A decline in stable longer-term employment relationships brought in by harsher business competition in increasingly open economies will undermine the conditions in which occupational insurance funds thrive. Governments are everywhere under pressure to lower subsidies to pensions programmes that benefit advantaged workers (Arenas de Mesa, 2020), especially as consumption and corporate taxes provide the bulk of fiscal revenues in Latin America.

4.2.2 Individual Retirement Savings Plans

Individual retirement savings plans have been available from financial institutions for high net worth customers in Latin America and elsewhere for some time. The policy change we are interested in here is the adoption of mandated individual retirement savings plans. Beginning with Chile in 1981, ten governments in the region legislated to compel workers to save a proportion of their earnings in dedicated accounts with newly established ad hoc pension fund managers (Barrientos, 1998a). The purpose of these mandates was to create a pension fund market. Pension fund managers collect the savings from workers and invest them in a range of government regulated assets. On reaching retirement age, workers can access the balances in their accounts and make appropriate retirement income arrangements.

Advocates of pension reform claimed a wide range of potential advantages: improving the working of the labour market and productivity, reducing old age poverty and vulnerability, lifting governments’ fiscal deficits, and strengthening financial markets (Edwards, 1996; Orszag & Stiglitz, 1999; World Bank, 1994). An assessment of then existing occupational insurance funds in the region emphasised the strong incentives for rent-seeking offered by the stratification in provision. Powerful labour organisations successfully lobbied governments for pension privileges and funding, in the process setting barriers to the mobility of workers to their most productive employment. A gap opened between contributions and entitlements with deleterious effects in incentives for human capital accumulation and effort. Individual retirement savings plans, it was argued, would restore the direct connection between contributions and entitlements with favourable outcomes for human capital accumulation, labour market competition, and productivity. Current or projected occupational insurance funds’ deficits underlined the need for growing fiscal support (Mesa-Lago, 2020). Mandating individual retirement savings plans would lift the threat of impending deficits and restore government fiscal balances. Shallow financial asset markets in the region were held responsible for low investment rates, short-termism, firm concentration and uncompetitiveness. They contributed to recurrent financial crises and low productivity and competitiveness. The emerging pension fund market was expected to deepen financial markets by redirecting workers’ retirement savings through the financial sector, while pension fund managers could provide the kind of institutional investors that would push through reforms and practices to strengthen the allocative role of financial markets (Barrientos, 1999). Individual retirement savings plans would have far-reaching effects on the performance of the region’s economies. The outcomes of pension reform failed to fulfil the transformational claims made on its behalf (Gill et al., 2004). Forty years after the Chilean reform, it continues to be contested.

Table 4.2 outlines the key features of individual retirement savings plans in Latin America. The first column reminds readers of the relationship existing between occupational insurance funds and individual retirement savings plans in countries’ pension design, whether occupational insurance funds are transitioning to individual retirement plans, individual retirement savings plans are in competition with occupational insurance funds or are integrated with them. The next four columns focus on the parameters of pension entitlements for the countries involved. In individual retirement savings plans, there is more scope for participants to make retirement decisions. The absence of redistributive features means that the consequences emanating from these decisions will not affect other participants. Column 2 notes the legal retirement age in the country, as a reference point, however in most countries it is possible for participants to access their individual funds balances earlier. There are, of course, implications for the government as the protector of last resort, hence some countries set a minimum vesting period before participants can claim any minimum pension guarantees.

Table 4.2 Main features of Individual retirement savings plans in Latin America

In the same prudential vein, there are restrictions on the form of entitlements. Under a life annuity, participants are required to use their accumulated balances to contract a life annuity with an insurance company. This has the effect of protecting the government from individual’s retirement funds running out. Unfortunately, insurance financial sectors are relatively thinly developed in Latin American countries with the implication that annuity rates are rather expensive. Two issues are important. First, the fact that financial markets trade very few long-term instruments makes it difficult for insurance companies to manage long-term liabilities. Second, the choice of mortality tables used in the calculation of annuities can be distorted by the selective nature of individual retirement savings plans participants.⁸ Population-wide mortality tables would significantly underestimate life expectancy after retirement for more advantaged workers. Alternatively, individual retirement savings plans can replace the purchase of an annuity with arrangements for scheduled withdrawals, effectively an annual pension which generosity is dependent on the fund balances at retirement. In Uruguay and Costa Rica individual retirement pension plans provide an annual pension supplement to the defined benefit pillar.

From the perspective of less advantaged workers, likely a plurality of mandated participants, minimum pension guarantees are paramount. Individual retirement savings plans have two main modalities of guaranteed minimum provision. Some offer a top up where fund balances could only finance an annuity below a pre-defined threshold (Valdés-Prieto, 2009). This threshold is shown in brackets in the Table as a proportion of mean labour earnings. It varies in generosity across countries, but the mode is around a quarter of mean salary (IMF, 2018). In other countries, participating in individual retirement savings plans guarantees a minimum pension level which replaces the annuity otherwise generated by the fund. These guarantees are financed from additional solidarity contributions or, in their absence, from public funds.

A deeper question surrounding individual retirement savings plans is whether they are ‘pensions’. A defining feature of pension schemes is that they protect the population from longevity risk, the risk that they may outlive their resources. This is a necessary and arguably sufficient condition for retirement arrangements to be described as pensions. By themselves, individual retirement savings plans are at best a conditional pension, that is conditional on participants accumulating sufficient funds - taking account of annuity rates - and conditional on meeting relevant vesting periods. Individual retirement savings plans could only qualify as pension schemes where governments can act as guarantors of contracted annuity income streams or where governments provide minimum pension guarantees.

As discussed above, individual retirement savings plans advocates made a great deal of their potentially favourable effects on financial markets. The Table shows that over time IRAs have accumulated large funds as a proportion of GDP. The Table shows accumulated funds at the end of 2019 prior to the COVID19 pandemic. The latter resulted in large withdrawals that may or may not be recovered in the future (Kay & Borzutzky, 2022). Chile’s accumulated fund is the largest, reflecting early reform and a sizeable share of the labour force as participants. Debates on the composition of fund asset portfolios are recurrent in all the countries involved, for example as regards investment in foreign assets. Government regulations must strike a balance between the need to maximise rates of return for participants, ensure that funds deepen domestic financial markets, and ensure they help finance public debt. Real rates of return of individual retirement savings plans are variable, but as the Table shows recent rates of return are healthy. Pension fund markets are highly concentrated. Popular levels of dissatisfaction are apparent in the interaction between pension fund managers and participants.

The fact that individual retirement savings plans now control a large share of financial assets has not led to a transformational change in the nature of financial markets or indeed the orientation of the economy (Gill et al., 2004). It has not led to a change in the distribution of assets in the economy. Discussing the distribution of income form labour and income from capital in different countries, Ranaldi and Milanovic (2020) draw a distinction between ‘class’ economies, where the vast majority of labour income is captured by workers and most income from capital is captured by capitalists, and ‘diversified’ economies where the income from capital and labour is more widely distributed across workers and capitalists. Empirically, they place Latin American countries as ‘class’ economies. However, classifying pension wealth as a private asset can nuance this assessment. As “…the share of older people is on the rise in many advanced countries and the use of private pensions becomes more popular, one can envisage a somewhat novel form of “classless” society where relatively equal shares of capital and labor across distribution are achieved through savings over active life and capital returns once in retirement.” (Ranaldi & Milanovic, 2020, p. 22). individual retirement savings plans fund in Chile are equivalent to 80% of annual GDP, but they have not provided wage earners with any influence over economic policy, or over the preferred structure of the economy, or over the distribution of assets. individual retirement savings plans participants have less influence on social policy than the managers of pension funds. If anything, individual retirement savings plans have delivered huge political and financial power to fund managers who have deployed it to preclude structural pension reform in Chile (Castiglioni, 2005; Dorlach, 2020; Fairfield & Garay, 2017).

Pension reforms implementing individual retirement savings plans in Latin America have not been entirely successful, certainly not in terms of the claims made by advocates. Attempts at reforming the reforms confirm this to be the case (Arza, 2017; Carrera & Angelaki, 2021). They map a direction of travel towards growing government involvement in supporting and regulating private pension plans. Argentina went back to a pay-as-you-go defined benefit pension programme in 2008. Bolivia legislated to retain individual retirement savings plans but under a single public fund manager, although the reforms have not been fully implemented to date. In successive reforms, Chile has strengthened the budget-financed first pillar basic pension and offered publicly funded graduated incentives for individual savings plans for lower income groups. The 2007 financial crisis and the COVID19 pandemic have underlined individual retirement savings plans exposure to financial volatility.

Individual retirement savings plans have a role in the institutional architecture securing income security in old age, mainly as voluntary instruments open to workers with the capacity and desire to save for retirement. Promoting retirement savings is a sensible policy in the context of population ageing. Uruguay and Costa Rica offer examples of pension reforms making individual retirement savings plans mandatory for advantaged workers as a means to supplement defined benefit pension schemes that are mandatory for all (de Melo et al., 2019).

4.2.3 Pressure Points for Pension Provision

Measures of pension scheme participation in the labour force are informative of compliance with of labour market regulation, but they are less reliable as a measure of prospective old age income security. Employment volatility and the temporary nature of the employment relationship characteristic of countries in the region means that only a fraction of those currently participating in pension schemes will eventually manage to secure adequate entitlements. Research indicates that due to spells of unemployment and under-employment, job transitions, gaps in compliance and fraud, pension contribution density in the region is around 50% (Altamirano Montoya et al., 2018). Applying this figure linearly into the future suggests only one half of the workers currently participating in pension schemes will secure an adequate pension at retirement.

Pension participation decisions for disadvantaged workers is equivalent to an option value, the value of retaining options for the future. Design pension scheme features, especially those relating to the bottom line like vesting periods and minimum pension guarantees, become highly salient. In occupational pension schemes, minimum pension benefits are often easier to access and more generous than in individual retirement savings plans. In individual retirement plans vesting periods tend to be high, around twenty years of contributions. The implication is that for a significant proportion of those currently in the labour force participation in a pension scheme is best described as an option on the minimum pension benefit guarantee. The expansion of old age transfers, to be examined below, is also relevant to the bottom line. Perhaps the takeaway point is that, in the context of Latin America, reforms affecting the bottom line are hugely important.

Pension reform, and ‘re-reform’, have generated a debate around the privateness-publicness of pension schemes (Mesa-Lago, 2020). Some researchers predicts a return of the state in pension provision (Carnes & Mares, 2015). The distinction between defined benefit and defined contribution pension schemes can overlap with the privateness-publicness spectrum (Mesa-Lago, 2020). But this can be exaggerated. As noted above, the ‘first incorporation’ in the 1950s and 1960s resulted in growing government involvement in occupational pension schemes.⁹ Latin American governments have had since a significant stake in the design, implementation, and financing of occupational pension schemes, especially as no other institution could effectively manage and sustain a pay-as-you-go financing model. Governments also have a large stake in the design, implementation, and financing of individual retirement plans. Governments mandate contributions; set the financial, tax, and regulatory frameworks; and ensure minimum pension guarantees. Individual retirement plans are not feasible without government involvement. Government also provide or regulate minimum pension guarantees without which individual retirement plans would most likely fail in the context of volatile employment. The privateness-publicness of pension schemes in Latin America is not its defining dimension. Governments are firmly positioned in the pension business (Blinder, 1988).

4.3 Social Assistance

Large scale social assistance provision emerged in the region following the decline of the neoliberal economic model and the return to democracy in the 1990s. It is associated with a ‘second incorporation’. Two social protection instruments dominate the growth of social assistance in the region: old age and disability transfers and conditional income transfers to families.

Figure 4.3. shows the expansion of social assistance in the new century. It tracks the reach of social assistance and old age transfers over time. Reach measures as the share of direct and indirect beneficiaries of social assistance programmes. Direct beneficiaries are transfer recipients while indirect beneficiaries include other household members. Conditional income transfers explicitly support households. Old age transfers are paid to qualifying older people but are shared within households. Reach is a more accurate measure of the population groups who benefit from social assistance transfers.

Fig. 4.3

Reach of social assistance. Notes: Reach is a measure of direct and indirect beneficiaries (co-residents). Data from CEPAL (2020) and Barrientos (2018). Estimate of the reach of Old age transfers constructed by applying average household size from UNDESA (2017) to direct recipient data

Table 4.3 provides an overview of current social assistance institutions by country, with information on expenditure levels, and age of entitlement and adequacy for old age transfers.

Table 4.3 Social assistance in Latin America—Programmes and budgets

A discussion of key features of old age and conditional income transfers follows.

4.3.1 Old Age Transfers

Old age and disability transfers¹⁰ were introduced early in the pioneer social protection countries. They were initially designed to complement or facilitate occupational insurance funds. Uruguay is credited with the first budget financed transfer programme for poor older. Brazil’s FUNRURAL, introduced in the early 1970s to cater for informal workers in agriculture, mining and fishing in rural areas, was intended to serve as a subsidised bridge to full membership of occupational insurance funds (Lewis & Lloyd-Sherlock, 2009).

In the new century, most countries in the region have established budget financed old age transfer programmes as self-standing institutions addressing old age and disability poverty and vulnerability:¹¹ Bolivia (1996), Argentina (2003), Colombia (2003 and 2013), Ecuador (2003 and 2009), Guatemala (2005), Mexico (2007, 2012 and 2013), Chile (2008), El Salvador, Panama (2009), and Peru (2011). Old age transfers are fast becoming a first pillar of old age income security provision (Bosch et al., 2013).

Rofman et al. (2013) distinguish three main strategies in the expansion of old age transfers in the region: universal, complementary and targeted. Bolivia’s Renta Dignidad provides an example of universal provision. It emerged in the mid 1990s out of policy proposals aimed at lowering public opposition to the privatisation of the energy sector, promising to redistribute dividends from partial public shareholding in the privatised energy corporations. After several permutations, it became a budget financed transfer paid to all citizens over 60 years of age. In countries with large scale pension schemes - e.g., Argentina, Uruguay, Brazil, Chile -, old age transfers have a complementary role. They reach the older population without access to pension scheme benefits. Consideration of work and saving incentives usually restrict the generosity of complementary old age transfers. In countries with limited pension scheme provision – e.g., Colombia, Ecuador, El Salvador, Peru -, old age and disability transfers have a poverty reduction role and are targeted to vulnerable group. Access to transfers is dependent on socio-economic status re-assessed at specified intervals. The complementary and targeted strategies can be embedded in different programmes. In Brazil, Previdência Social Rural, a follow up to FUNRURAL, represents the complementary strategy while Beneficio de Prestação Continuada represents the poverty reduction strategy (Barrientos, 2013). Today people with disabilities account for a plurality of Beneficio de Prestação Continuada.¹²

The rapid expansion of old age and disability transfers, especially among women, in the region has greatly improved old age income security in the region (Barrientos, 2021).

4.3.2 Conditional Income Transfers

Conditional income transfers are the most significant innovation in social protection systems in the region. Brazil’s Bolsa Escola and Mexico’s PROGRESA (Programme for the Advancement of Health and Education) are the pioneer programmes (Levy, 2006; Rocha, 2013). Introduced in mid-1990s they became large scale social assistance programmes providing regular and reliable transfers to families in poverty or at risk of falling into poverty. Transfers are conditional on children’s schooling and on household’s primary health care utilisation. All three core features of conditional income transfers, they are budget financed family subsidies, with a social investment orientation, are significant innovations in the context of Latin America (Barrientos, 2019). Conditional income transfers also reflect a fourth, crucial, innovation: Latin American governments’ newfound commitment to poverty reduction.

Between 1997 and 2006 conditional income transfers spread to all countries in the region. As expected, conditional income transfer programmes show diversity in design, scope and implementation (Cecchini & Madariaga, 2011a; Fiszbein & Schady, 2009; Ibarrarán et al., 2017). They are rules-based in the sense that entry and exit from the programme, and entitlements, follow the application of pre-established rules. Entry conditions are based on household demographics, the presence of children of school age for example, and socio-economic conditions. The latter is assessed via income or proxy means tests. Participant households’ compliance with programme conditions ensure the continuation of the transfers. Administration of programme conditions necessitates coordination between the programme agency and education and primary health care providers. Some countries like Mexico enforced conditions with greater regularity and force, whereas in countries like Brazil non-compliance with conditions served as an indicator that further support was needed. Transfers are mostly fixed in value and are paid through financial providers. Transfer values are a fraction of the poverty line, with the implication that conditional transfers are intended as a contribution towards improving household finances. They are not intended to lift recipients out of poverty by themselves.¹³ There are no labour supply restrictions associated with the receipt of the transfer.¹⁴

Programme designers paid more attention to entry conditions than to exit conditions. In fact, initially most programmes simply applied entry conditions to assess conditions for exit. This has not worked well (Barrientos & Villa, 2017). Changes in employment due to labour market volatility are perhaps the most significant factor pushing people into poverty. Consequently, the socio-economic conditions of a large proportion of low income families oscillate around the levels required by entry conditions. Families exiting the programme because they fail to meet entry conditions may soon find themselves meeting them. It makes sense to implement exit conditions to take account of labour market volatility. Brazil’s Bolsa Família guarantees transfers for two years after entry. Colombia’s Más Familias en Acción offers a reduced level of transfer support for families unable to meet entry conditions but remaining vulnerable. This support helps them smooth out exit from the programme (Barrientos & Villa, 2017).

The process of assessing entry conditions involves collecting data from current and potential participants. It led to the development of beneficiary databases available to programme agencies. There are two types of database. Participant registers collect information on current participants. Social registers collect information on current and potential participants. Governments in the region normally offer several public programmes supporting disadvantaged households including, for example, interventions in education, health, housing, etc. They used to involve separate registers and entry conditions. It made sense to consolidate these registers into a single social register, covering the population at risk of poverty. Colombia’s SISBEN collects information on all households applying for any public programme. As the social registers grew, they provided governments with essential information on the target population for a variety of social policies. Social registers have been crucial to government policy addressing COVID19 (Berner & van Hemelryck, 2020).

Mexico’s PROGRESA was introduced in the context of a financial crisis and against the opposition of pre-existing poverty programme (Levy, 2006). The designers of the programme included well-defined evaluation protocols to protect its initial development. By collecting a baseline, and follow up, household surveys and by exploiting a phased implementation, it became possible to compare outcomes for early and late entrants. Demonstrating the effectiveness of anti-poverty transfer programmes helped mitigate political opposition to the programme (Skoufias, 2005). Quasi-experimental impact evaluation studies have become the norm for conditional income transfers, generating valuable information on their effectiveness. They have contributed to the experimental turn in economic development (Banerjee & Duflo, 2011). Monitoring and evaluation protocols stand out as a very significant innovation in social assistance.

The evolution of conditional income transfer programmes over time in Latin America suggests three interesting trends.

First, conditional income transfers link programme agencies with the, often more powerful, ministries of education and health. Linking transfers in cash with access to basic services is central to the social investment orientation of these programmes (Barrientos, 2022). Inter-agency coordination can lead to improved access to a wider range of services and intermediation by conditional income transfer beneficiaries. Income transfers prove effective for most participant households, but some participants may require more specialised and direct forms of support. Chile Solidario was designed around dedicated household support (Barrientos, 2010). It has promoted the integrating income transfers and service provision to address inclusion barriers for poorest households.

Second, over time conditional income transfers have developed from fixed-term programmes into permanent government institutions. Institutionalisation involves the establishment of social development agencies or ministries with parliament sanctioned budgets and legal oversight and operating frameworks (Székely, 2015). The evolution from interventions to institutions is not yet complete in less developed countries in the region (Cecchini & Madariaga, 2011b).

Third, conditional income transfers pose longer term policy dilemmas in need of future resolution. Some of these policy dilemmas have been covered in the discussion above: the definition of programme exits; the appropriate balance between protection and investment; their impact of growing social assistance institutionalisation on other parts of government; their linkages to occupational insurance. Conditional income transfers have unfolded against a sustained improvement in living standards and poverty levels in the region. The more developed countries in the region have evolved conditional income transfers into permanent support for less advantaged families with children. Argentina’s Asignación Unica por Hijo is perhaps a case in point as it concluded a policy shift from emergency public works to conditional income transfers to child benefit (Gasparini & Cruces, 2010). On becoming President of Mexico, Lopez Obrador rescinded the long standing conditional income transfer (Progresa, Oportunidades, Prospera) are replaced it with scholarship programmes (Huesca Reynoso et al., 2020).¹⁵ This pathway leans towards a stronger social investment orientation for social assistance.

4.4 Secondary Programmes

The discussion in this chapter has focused on the main programmes providing pension benefits and social assistance transfers. In practice, some countries have implemented a range of complementary programmes. A brief discussion of unemployment protection, family benefits and active labour market programmes follows.

With few exceptions, social protection in Latin American countries have not implemented large scale programmes against unemployment. Latin American countries have sought to protect jobs, not workers. Severance pay regulations raise the costs of worker separation for employers by mandating compensation to affected workers. This is normally a lump sum reflecting years of service and other parameters. They are widely considered to be inadequate in terms of providing insurance for workers (Holzmann & Vodopivec, 2012). Because only a minority of workers are employed in firms that comply with employment regulations, severance pay is restricted in its application. Severance pay regulations “encourage stratified labor markets and impose barriers in employee’s mobility and the firm’s adjustment to changing labor market conditions” (Ferrer & Riddell, 2011, p. 1). They are consistent with import substitutions policies restricting competition in selected sectors. Severance pay regulations were always expected to apply selectively.

Unemployment insurance schemes are hard to implement successfully in labour market conditions such as those prevailing in Latin America. In conditions where a significant share of the labour force is self-employed or have weak and volatile employment relationships, programmes providing income protection for workers affected by unemployment are undermined by potential moral hazard. There are few examples of unemployment insurance programmes implemented early in the region (Velásquez Pinto, 2010). Argentina set up a scheme for construction workers in 1975. Panama introduced a scheme in 1972 for selected groups of workers. Brazil introduced the Fondo de Garantia por tiempo de servicio in 1967 and a second one, the Fondo de Amparo al Trabajador, in 1986. Other countries implemented partial unemployment insurance schemes in the 1990S. They include (start date): Ecuador (mixed 2001); Peru (1991); and Venezuela (1997). They have very limited coverage, as they exclude workers with weak job attachment, rural and urban informal workers, and public sector workers. In practice, they cover the workers least likely to experience unemployment. They are financed by a mix of employee and employer payroll contributions and attract government financial support.

Interest in unemployment insurance rose in the 2000s as individual retirement savings plans offered a ready-made infrastructure to collect workers’ dedicated unemployment savings. Colombia (1990), Chile (2002) and Ecuador (2002) established unemployment savings plans solely on worker savings. Sehnbruch et al. (2020) offer an analysis of Chile’s programme, using a sample of administrative records. They conclude that the programme has limited coverage among workers most likely to need it. Only 11.8% of workers with open ended contracts were entitled to make a claim and only 6.7% made a claim. Among fixed term contract workers, the relevant figures are 2.3% and 0.9% respectively. The low share of successful claims is due to the high level of job rotation (Gualavisi & Oliveri, 2016). An attempt to introduce an element of redistribution across participants did not fare well, “…the solidarity fund embedded in the system, which was originally supposed to redistribute funds towards lower income workers, does exactly the opposite” (Sehnbruch et al., 2020, p. 27).

As Bertranou and Maurizio (2011) conclude “…contributory-based unemployment insurance has not been fully developed in Latin America…This is mainly due to the high levels of informality (since these programmes have been designed to protect wage earners that hold a formal job, generally under a permanent contract) … and, to a lesser extent, to the occupational instability of workers in the formal market. Although these systems receive contributions from workers, they tend to depend on fiscal revenues, be it through regular contributions or indirectly when the State has to cover its operational deficits” (Bertranou & Maurizio, 2011, p. 17). Unemployment insurance schemes have not been successful in the region. They show the same limitations as the pension schemes their design is borrowed from, whether occupational pensions or individual retirement savings plans.

Active labour market policies are a component of social protection growing in significance in the region, albeit from a low base (Escudero et al., 2019). Active labour market policies consist of distinct main policies addressing unemployment: training programmes offering an upgrade in skills; public works and employment subsidies providing short term work to the unemployed; microenterprise development; and labour market intermediation. Their main objective is to raise employment. In the context of Latin America, active labour market policies are in the main oriented to low paid workers who are more likely to be affected by unemployment. While there are countries with large scale training infrastructure like Brazil financed from payroll taxes and with centralised governance, recent activation policies are more likely to complement participation in social assistance programmes (ECLAC/ILO, 2014) with which they share a social investment orientation (Barrientos, 2022).

Escudero et al. (2019) offer a meta-analysis of impact evaluations of active labour market policies in the region. They find that three quarters of the programmes studies focus on training. They are followed by initiatives to strengthen private sector incentives and labour market intermediation. Programmes have relatively short duration, less than 9 months. The impact of the programmes on employment is stronger for women and young workers, and for low-income workers. They find that activation programmes generally have stronger effects in the short-term.

Rossel (2013) offers a comprehensive review of family policies in Latin America. All countries in the region have regulations on maternity license and subsidies, but there is considerable variation across countries in the reach and generosity of their provision. The relevant legislation excludes several categories of workers so that in practice entitlements are only available to workers with permanent open ended employment contracts. Conditional income transfers reach low income and informal population groups, but their support for families is often limited to children and mothers.

Overall, this brief discussion of complementary social protection programmes confirms the main findings extracted from the examination of the main programmes as regards the structure and orientation of institutions in the region.

4.5 Conclusions

This chapter has examined the three main social protection institutions in Latin America countries: occupational pensions, individual retirement savings plans, and social assistance. It discussed the core features of each of these institutions, offering a perspective on their manifestation in each of the countries in the region as well as a more analytical assessment of their institutional scope and logic. This is necessary grounding for the study of their stratification effects to come in the next chapters.

In essence, social protection institutions can be arranged into two. Insurance-based institutions collect contributions from employees or citizens and compensate them in the event they are afflicted by pre-specified contingencies. Assistance based institutions funded from government revenues and providing income transfers to citizens affected by poverty and vulnerability. These two sets of institutions are involved respectively in horizontal redistribution, transferring income for the same people in different circumstances or stages of the line course, or alternatively vertical redistribution, transferring income from better off to worse off citizens. The bases of entitlements are respectively group membership and citizenship.

Current Latin America institutions show some adjustment to the two sets of institutions, and an addition.

Insurance-based institutions in Latin America are restricted to selected groups of workers, mainly skilled workers in large firms. They are also highly fragmented, despite governments attempts to streamline their governance. They are strongly dependent on public subsidies, although the level of subsidies is increasingly contested. If in their original designs, occupational insurance funds covered a wide range of contingencies, current occupational insurance is primarily focused on old age. This is in part due to the demographic transition, but it is also due to the politics of social protection, the fact that occupational insurance in Latin America is a stratification mechanism.¹⁶ Recent trends in labour force participation in occupational insurance largely suggest stagnation during this century. Keeping in mind that participation rates significantly underestimate the share of participants who will secure adequate pension benefits at retirement, occupational insurance institutions in the region fail to guarantee income security in later age except for a selective group of workers. Occupational insurance is stronger in early industrialising countries and weak in the rest.

Social assistance greatly expanded in the new century. All Latin American countries implemented conditional income transfers, some countries did so to scale. Most countries have implemented large scale old age transfers or have strengthen pre-existing provision. By 2015, social assistance transfers reached around one third of the population in the region. The expansion of social assistance has involved a welcomed upgrade in institutional capacity. Conditional income transfers diffused rules-based programming, social registries, monitoring and evaluation, and evidence-based policy throughout the region. There is closer institutional harmonisation in social assistance than in occupational insurance in the region, bar outliers like Nicaragua and Venezuela. Argentina and Brazil provide contrasting examples of social assistance institutional transition.

Social protection institutions in Latin America contribute an additional set of institutions: individual retirement savings plans. They are neither grounded in insurance nor in assistance, but instead in individual saving capacity. They have not been implemented throughout the region. They show three distinct institutional settings. In Colombia and Peru, individual retirement savings plans compete with occupational insurance. In Costa Rica and Uruguay, they complement occupational insurance for advantaged workers. In the rest of the countries that have adopted them, a transition process will make them the sole mandated retirement income institution. Individual retirement savings plans have not proved successful in the region, and it is fair to say their institutional structures are in transition.