Eligibility for retirement and replacement rates in the Uruguayan multi-pillar pension system

Document

Citado por

Autor	Gioia de Melo, Nicolás Castiñeiras, Analía Ardente, Oriana Montti, Braulio Zelko, Federico Araya
Páginas	105-144

105

DESARRO. SOC. 71, PRIMER SEMESTRE DE 2013, PP. X-XX, ISSN 0120-3584

Revista

Desarrollo y Sociedad

Segundo semestre 2019

PP. 105-144, ISSN 0120-3584

E-ISSN 1900-7760

Eligibility for retirement and replacement rates

in the Uruguayan multi-pillar pension system

Elegibilidad para el retiro y tasas de reemplazo

en el sistema previsional multi-pilar en Uruguay

Gioia de Melo1

Nicolás Castiñeiras 2

Analía Ardente3

Oriana Montti4

Braulio Zelko5

Federico Araya6

DOI: 10.13043/DYS.83.3

Abstract

We project the levels of eligibility and gross replacement rates of the pay-as-

you-go and individual capitalization pillars in Uruguay. Based on a random

sample of worker administrative records, we estimate years of contributions,

formal income, and the evolution of the individual savings fund. Our results

suggest that while 51% would be eligible for retirement at age 60, 28% would

not be able to retire from the contributory system even at age seventy. We

expect that 34% of those retiring at age 60 will receive a minimum pension

1 Organización para la Cooperación y el Desarrollo Económico, Correo electrónico: Gioia.DEMELO@ocde.org

2 Uruguay XXI, Correo electrónico: ncastineirascassanello@gmail.com

3 Ministerio de Economía y Finanzas, Correo electrónico: aardente@tgn.gub.uy

4 Ministerio de Economía y Finanzas, Correo electrónico: oriana.montti@mef.gub.uy

5 Ministerio de Economía y Finanzas, Correo electrónico: braulio.zelko@mef.gub.uy

6 Ministerio de Trabajo y Seguridad Social, Correo electrónico: faraya@mtss.gub.uy

Este artículo fue recibido el 11 de septiembre del 2018, revisado el 8 de abril del 2019 y finalmente

aceptado el 14 de junio del 2019.

Eligibility for retirement and replacement rates in the Uruguayan

106

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

while the replacement rate is estimated to be 52% relative to the previous

year’s wage. We conclude that Uruguay still faces challenges regarding indivi-

duals’ density of contributions and amounts declared as both reduce eligibility

levels and impose financial pressure on the pay-as-you-go pillar.

Keywords by author: Informality, pay-as-you-go, individual savings, minimum

pension, Uruguay.

JEL Classification: H55, J26, J46.

Resumen

Este artículo proyecta los niveles de elegibilidad y las tasas de reemplazo brutas

de los pilares de reparto y capitalización individual en Uruguay. A partir de una

muestra aleatoria de registros administrativos de trabajadores se proyectan los

años de servicio, los ingresos formales y la evolución del fondo de capitalización

individual. Se estima que 51% de los trabajadores generaría causal de reparto a los

setenta años, mientras que 28% no lo lograría ni siquiera a los setenta años.

Los resultados sugieren que 34% de los que se retiren a los setenta años percibirán

la jubilación mínima en tanto que la tasa de reemplazo promedio en relación con

el ingreso del último año sería de 52%. Se concluye que Uruguay aún enfrenta

desafíos en materia de densidad de contribuciones y de montos declarados que

impactan tanto en los niveles de elegibilidad como en la presión financiera

que registra el pilar de reparto.

Palabras clave del autor: informalidad, reparto, capitalización, jubilación

mínima, Uruguay.

Clasificación JEL: H55, J26, J46.

The purpose of this study is to project eligibility and replacement rates for the

Uruguayan social security system in its two pillars: mandatory individual sav-

ing accounts (defined-contribution) and the pay-as-you-go system (defined-

benefit), which expresses them as a ratio of the last salary and the average

working-life earnings. We aim to analyze both the coverage of the system and

its ability to smooth consumption over its life cycle. Bucheli, Ferreira-Coimbra,

Forteza & Rossi (2006) and Bucheli, Forteza, & Rossi (2010) estimate eligibility

prior to the 2008 law (18395) that loosened access requirements to contribu-

tory pensions. This study updates projections for the percentage of individuals

Gioia de Melo

et al.

107

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

expected to become eligible for contributive pensions and most importantly, for

the first time, estimates replacement rates for Uruguay using individual data

from social security records. Studies that project pension replacement rates

using administrative data are extremely scarce in the international literature.

Using individual data is relevant because it allows us to analyze distributive

implications and is of particular importance in countries with non-negligible

levels of informality as most countries in Latin America.

Our data consists of a random sample of 14,207 contributors to the Banco

de Previsión Social (Uruguay’s major social security agency) who in January

2017 were between 40 and 60 years old. Based on this sample, we estimate

the probability of them contributing each month until they reach retire-

ment age. We then model individual earnings based on personal characteris-

tics and macroeconomic variables. We also project the accumulated fund in

the mandatory individual savings account (FAP for its Spanish acronym). The

high quality of the data used enables us to analyze heterogeneity among

the cohort under study.

We compute gross replacement rates that result from considering both the

defined benefit and the individual savings pillar as a whole (which we name

joint replacement rate) and also analyze the replacement rate yielded by each

pillar separately and simulate how different macroeconomic scenarios affect

our results. We analyze the correlation of the resulting rates in terms of gen-

der, age cohort, average density of contributions, income quintiles, and sector

of activity. Furthermore, we examine how the rate of return of the pensions

funds’ investments and management fees affect the final amount accumu-

lated in the individual savings account and, subsequently, the annuity. Finally,

for those individuals who have a high contribution density, we analyze the

impact of postponing the decision to retire until age 65.

The paper is organized as follows: the next section describes the Uruguayan

pension system. Section II reviews the related literature. Sections III and IV

describe the sources of information and methodology used. Section V presents

the results while section VI concludes.

I. The Uruguayan pension system

In 1996, the Uruguayan pension system administered by Banco de Prevision

Social (BPS) was transformed into a “Mixed System” (Law 16713) with two

Eligibility for retirement and replacement rates in the Uruguayan

108

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

pillars: a pay-as-you-go (PAYG) or defined benefit scheme and an individual

savings accounts or defined contribution scheme. The system became com-

pulsory for those under 40 as of April 1, 1996. A monthly earnings threshold

separates the two pillars. Individuals with monthly wages lower than $5,000

(Uruguayan pesos in May 1995, approximately USD 1,700 in 2017) are served

by the PAYG unless they explicitly choose to participate in the individual savings

scheme. They can choose to assign 50% of their monthly contributions to the

individual savings account (Article 8 Law 16,713). If they do, BPS grants a bonus

and increases the individuals’ contributions to a defined benefit scheme by

50%. Earnings between $5,000 and $15,000 (Uruguayan pesos in May 1995)

must contribute to the individual account; it is optional to do so for earnings

above this amount.

Within the PAYG pillar, a minimum of 30 years’ contributions is required in

order to be eligible (the minimum was set at 35 in 1995 and then reduced by

Law 18,395 in 2008). Required years of contribution decrease if individuals

retire at older ages (minimum of 15 years after 70). Women benefit from one

year’s contribution per child (maximum of five years). The benefit not only

improves eligibility conditions but also increases the replacement rate of the

defined benefit scheme.

In the PAYG pillar, the earnings considered to calculate the pension are the

average wage earned in the last 10 years. Depending on what is more beneficial

for the worker, the average wage earned in the best twenty years (increased

by 5%) may also be used. The rate applied is 45% and there is an additional

1% for every extra year’s contribution between 30 and 35. For thirty-six years

of contributions onwards, every additional year worked adds 0.5% to the rate

(there is a maximum of 2.5%). Moreover, for each year that the person post-

pones retirement after the age of 60 (whether s/he continues working or not)

the replacement rate increases by two pecentage points. Therefore if an indi-

vidual reaches age 60 with 30 years’ contributions and decides to continue

working until 61, 48% will be used to calculate their pension.

As for the individual savings pillar, Law 16,713 created AFAPs that administer

the savings fund (Fondo de Ahorro Previsional) that consists of the mandatory

and voluntary workers´ contributions and its returns. AFAPs charge an admin-

istration fee and deduct an insurance premium (“Prima de Seguro Colectivo”)

that covers the risks of disability and mortality during working life. The annuity

Gioia de Melo

et al.

109

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

depends on the final accumulated fund in the individual account, sex, and age

at the time of retirement (which reflect the individuals´ life expectancy) and

the probability of having a member of the family who can claim the pension

as well as a technical interest rate. Individuals can cash their annuity when

they retire from the PAYG system or at a minimum of 65.

II. Related literature

To the best of our knowledge, only one study from Chile projects pension

replacement rates using administrative data. Berstein, Larrain and Pino, (2006)

estimate the probability of future contributions using a probit model with

random effects and an earnings model that includes individual characteris-

tics and variables related to the macroeconomic context. Replacement rates

are estimated at 121% of the average salary during the entire working life

and 44% of the last salary. Paredes and Díaz Fuchs (2013), in turn, calculate

replacement rates for Chile using the administrative data of individuals who

had already retired in 2012. They estimate an average replacement rate for the

old age pension (with at least 10 years’ contributions) of 76%. They additionally

highlight the substantial differences due to gender and contribution densities.

OECD (2013) simulates replacement rates assuming individuals contribute

to social security for 40 years earning the average wage in their respec-

tive country. For OECD member countries, the average gross replacement

rate is estimated at 54%. Durán Valverde and Pena (2011) project replace-

ment rates from the individual savings account systems of different Latin

American countries using information from household surveys. In the case of

Uruguay, they estimate replacement rates of 43% and 44% for men and wo-

men, respectively, relative to the last wage and 39% and 38% for men and

women relative to the average income of the last 20 years they worked. Forteza

and Ourens (2012) analyze the characteristics of several Latin American pen-

sion programs evaluating their impact on income inequality, insurance, and

incentives to work. They simulate lifetime contributions and benefits and con-

clude that most programs are progressive while implicit rates of return exhibit

marked discontinuities in the length of service, which they attribute to vest-

ing period conditions. A recent study by Altamirano, Berstein, Bosch, García-

Huitrón and Oliveri, (2018) analyzes pension systems across Latin America and

the Caribbean and consider hypothetical individuals with a perfect contribution

Eligibility for retirement and replacement rates in the Uruguayan

110

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

record in the baseline scenario. For Uruguay, the estimate for the average

replacement rate is 72% in the baseline scenario.

In contrast to the lack of studies that project replacement rates and eligibility

while using individual administrative records, there is a greater abundance of

literature that forecasts pension eligibility in Uruguay. Bucheli et al., (2010)

estimate the proportion of workers who would gain access to a pension at

the normal retirement age by using a random sample of work history records

for individuals between 18 and 70 who contributed for at least one month

between April 1996 and December 2004. They use survival analysis to model

transitions between contributing and not contributing and then apply the esti-

mated transition rates to project worker trajectories using Monte Carlo simu-

lations. Their main findings suggest that, as workers spend more time in any

of the two states (contributing or not contributing), their chances of remain-

ing in the same state increases. They also find that young workers are more

dynamic and have a higher probability of leaving either of both states. They

project that only 41.3% of workers will have accumulated 30 years’ contribu-

tions at age 60. Also, they find very significant differences between income

quintiles. It should be noted this study was carried out prior to the 2008 reform

that relaxed pension eligibility requirements.

Forteza and Sanroman (2015) estimate a structural life-cycle model for retire-

ment behavior using work history records. They find that individuals, espe-

cially women, prefer not to postpone retirement. They also find that retirement

decisions are not particularly responsive to changes in retirement modifica-

tion rules. More recently, Lavalleja, Rossi and Tenenbaum (2018) have studied

the impact of the latest legal modification in pension requirements that was

passed in 2008 using administrative records. Following Bucheli et al. (2006),

they divide the sample into groups that consider sex, income level, and con-

tribution sector and estimate the probability of contributing by group (con-

ditional on the previous month’s contribution status). Subsequently, they

simulate each group’s contribution history by using Markov chains. They con-

clude that for the minimum age of retirement (60), 38% of workers would be

eligible for retirement under the current legislation compared to 11% under

the 1996 regime (Law 16,713): a scenario in which the minimum years’ con-

tributions had not been reduced from 35 to 30 and there was no bonus year

per child for women.

Gioia de Melo

et al.

111

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Regarding other Regarding other estimates of pensions adequacy in Uruguay,

Forteza and Rossi (2016) simulate theoretical pensions and replacement rates

for those in their 50s in 2016 and compare them with those who retired from

the PAYG system under the conditions prior to the 1996 reform. They conclude

that some individuals in their 50s (mainly high-income) would have received

a higher pension under the system prior to the reform. Alvarez, Silva, Forteza

and Rossi (2010) focus on retirement incentives by comparing the PAYG sys-

tem prior to the 1996 reform with the multi-pillar system. Using simulations

for hypothetical individuals, they conclude that the 1996 reform improved the

incentives to postpone retirement, but the incentives are still low compared

to developed countries.

III. Data

This study uses a random sample of 14,207 workers´ administrative records for

individuals aged between 40 - 60 years old as of January 2017 who contrib-

uted at least once as a dependent worker between 1996 and 2015. Using the

Longitudinal Survey of Social Protection (2015), we estimate that the sam-

ple analyzed is representative of 73% of the age bracket (with the remaining

20% never contributing to social security and 7% contributing to other pen-

sion systems). The rest of the cohort either never made contributions to social

security or made contributions to other schemes such as military, police, or

professional ones. We concentrate on this age cohort because, given the policy

debate that took place in 2016 – 2017, we were interested in comparing the

fate of those in their 50s, which was the first cohort about to retire from the

multi-pillar system, with those 10 years younger. It should also be noted that

no administrative records exist prior to 1996. The following table presents the

descriptive statistics of the variables included in the database.

Figure 1 shows the distribution of the average contribution density for the

employment history sample over the observed period (1996-2015). We observe

a significant disparity in individuals’ contribution density. In the lower distri-

bution tail, 31.6% of men and 26.9% of women contributed up to 30% in the

1996-2015 period, which is equivalent to a maximum of six years’ contribu-

tions of the 20 years considered. Moreover, approximately 25% persistently

contributed (100%) during the period. It is also important to note that slightly

less than 10% of the sample shows a contribution density of less than or equal

Eligibility for retirement and replacement rates in the Uruguayan

112

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Table 1. Descriptive statistics

Variables Mean

Gender

Men 0.52

Women 0.48

Monthly income a ($ Jan-17) 26,891

Standard deviation (35,777)

Age (Jan-17) 49.2

Contribution density 0.57

Opted for Art. 8 Lay 16,713 0,60

Contribution sector

Industry and Commerce 0.63

Public servants 0.13

Rural 0.12

Construction 0.06

Housekeeping Service 0.06

Cohort

40 to 44 Cohort 0.28

45 to 49 Cohort 0.25

50 to 54 Cohort 0.23

55 to 60 Cohort 0.24

Sector

Agriculture 0.13

Manufacturing Ind. 0.13

Electricity, gas, and water 0.02

Construction 0.07

Commerce, restaurants and hotels 0.19

Transport and Communications 0.07

Financial intermediation, insurance. 0.05

Administrative Act. 0.03

Public Administration 0.11

Education 0.03

Health 0.06

Arts and others 0.04

Home Serv. 0.07

Extraterritorial Org. 0

Quintiles a

Quintile 1 4,023

Quintile 2 9,408

Quintile 3 13,699

Quintile 4 21,823

Quintile 5 55,265

Source: Authors’ own calculations based on social security records. a) Income for 1996 - 2015 at January 2017

prices in Uruguayan pesos.

Gioia de Melo

et al.

113

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

to 5%, which means that they contributed to social security for at most 12

months between 1996 - 2015. This polarization in terms of contribution density

highlights the importance of analyzing eligibility and replacement rates for the

different segments of the distribution. The percentage of workers who did not

contribute to the savings accounts pillar is estimated to be 32%. Indeed, even

if approximately 75% of the workers are situated in the first income bracket,

where contributions to the savings account are not compulsory, 60% of the

sample opted for Article 8 of Law 16713, which enables voluntary contribu-

tions to the individual savings pillar.

Figure 1. Distribution of the observed contribution density by gender (1996-2015)

Male Female

30%

25%

20%

15%

10%

0.05

0.95

0.90

0.85

0.80

0.75

0.70

0.65

0.60

0.55

0.50

0.45

0.40

0.35

0.30

0.25

0.20

0.15

0.10

We define contribution density as the relationship between the number of periods contributed by the

employee to the social security system and potential contributions during his or her working life.

Source: Authors´ own computations derived from social security data.

IV. Methodology

A. Contribution density and years of contributions

To project the contribution density, we model the probability of making monthly

contributions to the social security system from 1996 onwards using a pro-

bit model with random effects including both individual characteristics and

macroeconomic indicators.

Eligibility for retirement and replacement rates in the Uruguayan

114

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Let

yit

∗

be the unobserved propensity of individual i contributing to social secu-

rity in month t, while

yit

is the observed contribution status of individual i in

month t.

yit iiti



 =+ +++

−

∝β βββ

12 13 4

Female AccumDensity Cohort Sectoor Age

InformalityFemalePar

GDP





ββ

8 9 tticipationtit

 

+ε(1)

yit

=1

if yit

*.>06 and 0 otherwise.

In line with Bernstein et al. (2006) and Castañeda, Castro, Céspedes and Fajn-

zylber, (2013) we fit a random effects model, where

it it i

vu=+

and

vit

and

are independent random variables and [

|

()

]. Hence, the individual-

specific effects

vi

are assumed to be uncorrelated with the regressors and

constant through time.

The explanatory variables of the model are: female, the individual’s cumulative

average contribution density, generation cohort dummies, main contribution

sector dummies, age, age squared; and the following are the macroeconomic

variables: evolution of GDP, informality rate, female participation rate, and

the interaction between female and female participation rate.

The cumulative average contribution density, for each period t and for each

individual i is the ratio between the number of months that the person actu-

ally contributed to social security as a dependent worker in the period from

April 1996 to t-1, over the total number of months in the same period. We

define the binary cohort variables by age groups as of January 2017 from 40

to 44 years old (omitted category), 45 to 49 years old, 50 to 54 years old, and

55 to 60 years old. The contribution sector dummy variables follow the social

security agency´s categories: Industry and Commerce (omitted category), Pub-

lic, Construction, Rural, and Housekeeping Services.

Various specifications of the model were estimated. The final specification was

the one that best predicted the evolution observed in the 2011 - 2015 period

and yielded a reasonable evolution from 2016 onwards.7 Under our preferred

specification of a variable that reflects past contributions, that is, including

7 Indicators such as the last month or average contribution density of the last twelve months lead to

slightly higher percentages of correctly predicted observations than the cumulative average contribu-

tions density (83, 78, and 77 %, respectively). However, when extended to 2016 onwards, they tend to

forecast a steep increase in the average contribution density that is unlikely to take place given that in

the 2004 – 2014 period many policies that encouraged formalization were implemented and GDP grew

at rates above 3%. These are two features that we do not expect to prevail in our baseline scenario.

Gioia de Melo

et al.

115

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

the individual’s cumulative average contribution density, the model correctly

predicts 77% of the observations in the 2011 – 2015 period.

We should mention that not including a variable that has past information on

the individual’s contributive behavior only predicts 51% of observations cor-

rectly, which reinforces the idea that past contribution is a key regressor to

accurately forecast future. We subsequently applied a threshold to translate

the predicted probabilities. We selected a threshold of 0.6 by also testing which

was the best predictor in the sample period (2011 - 2015). The projection of

the density of contributions results from multiplying the explanatory variables

by the coefficients estimated in the probit model in table A.1 in the Appendix.

It is worth noting that given that the modelling objective is to achieve the

best possible projection and not to explain the role of a certain variable of

interest on the probability of contributing to social security, the endogene-

ity considerations of the explanatory variables are irrelevant. Besides, given

that several explanatory variables are included, it is possible that the sign of

some of these variables may seem counter-intuitive, but it should be taken

into consideration that we are comparing them with respect to very particu-

lar cases. That is to say, the group that meets all the omitted characteristics:

men under 45 whose main contribution sector is Industry and Commerce as

well as other aspects.

From 2016 onwards, we project GDP growth, female activity, and informality

rates for 2046 for three macroeconomic scenarios (Table A.3 in the Appendix).

We construct the base scenario under the assumption of a 3% GDP growth

[in line with Güenaga, Mourelle and Vicente (2013) and Dassatti and Mariño

(2014)]: a 2% increase in wages and a 3.5% return rate on constant prices

of the individual savings fund (FAP). Likewise, we assume a gradual reduction

in the labor supply gap between men and women of approximately 20% by

the end of the period with a corresponding informality rate of 21%. The ran-

dom effect is incorporated to the projections by adding a term [

rnormal0,



()

]

to the projection equation that generates a standard random variable with a

0 mean, and  is the panel level standard deviation of the regression.8

Individuals’ behavior in the labor market significantly changes by the age of

sixty when retirement becomes an option (at least for those who have met the

contributions requirements). Therefore, it is not sensible to use the projection

8 This term is averaged in over 100 values returned by the rnormal function in order to replicate what

stata does when computing individual random effects.

Eligibility for retirement and replacement rates in the Uruguayan

116

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

model described in the previous section for ages over 60. For these ages, we

chose to replicate the evolution of the contribution density observed by gender

and contribution sector for a random sample of 40,000 individuals who were

between 60 and 70 years old in 2016 and apply this evolution to the projected

trajectory for the sample of individuals currently aged 40 to 60 in our sample.

Even if the behavior of individuals from our cohort of interest might change

relative to what has been the behavior of older cohorts from age sixty onwards,

we believe this approach yields a more reasonable evolution than the results

from extending the probit model to older ages.

Administrative records are only available from April 1996 onwards. In order to

compute monthly contributions prior to this date, the social security agency

requires witnesses who can testify that the individual actually worked in a

certain firm in a given time period. To determine the years of contributions

prior to April 1996, we envisage two options: First, we use the information

from the administrative records regarding the start date of the job the indi-

vidual had by April 1996. We assume that since that date the person contrib-

uted without any interruption, and that s/he did not have another formal job

prior to that date. This may underestimate the years of contributions the social

agency will recognize especially for those who were close to forty in 1996 and

can provide witnesses. For this reason, we also consider another option

and choose the maximum years of contributions yielded by the two. The second

option consists of considering the observed contribution density during the

1996-2002 period for each individual and applying that value since the per-

son was 21 years old (age at which the majority of the population entered the

labor market in the nineties according to Household Surveys) until April 1996.

Below we present the contribution projection for the different cohorts (spot-

ted line) up to 70 years old for the base macroeconomic scenario.

B. Income model

In order to estimate individual’s i annual income at time t, we propose a linear

regression model with random effects that includes the following explana-

tory variables: female dummy, five-year age cohorts’ dummies, economic sec-

tor dummies according to ISIC classification, experience, experience squared,

annual income in t -1, and GDP. The estimated income equation is as follows:

YFemaleCohortSectorExperienc e

it ii

=+ 

∝β βββ++

1234

Experience GDPIncomeβββε+

671it titit

−



(2)

Gioia de Melo

et al.

117

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Figure 2. Projected percentage of individuals contributing to the social security

system by age cohort. Base scenario

0,10

0,30

0,50

0,70

0,90

20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70

Cohort 40 to 44 Co hort 45 to 49 Cohort 50 to 54

Cohort 55 to 60 Projection

Ages

Source: Authors´ own computations derived from social security data.

where 

it it i

vu=+

, and v

is an individual random effect which is constant

through time.

Following Berstein et al. (2006), we define the variable experience as the num-

ber of years contributed at time t. We estimate five separate models, one for

each income quintile.9 Quintiles are calculated considering income between

2013 and 2015 at constant prices. Income is projected solely for those peri-

ods in which it is projected that workers will contribute to the social security

system. Similarly to the contribution projections, for 2016 onwards, the ran-

dom effect is incorporated by adding a term [

rnorma

l0,

()

] to the projection

equation that generates a standard random variable with a 0 mean and when

 is the panel level standard deviation of the regression.

9 We initially estimated a single income model including, among other variables, the contributor’s income

quintile for the period 2013-2015. Although this specification showed an appropriate evolution, when

observing the evolution by quintile, the average income of each quintile converged to the average, and

there was a considerable increase in the income of quintile 1 and a decrease in quintile 5.

Eligibility for retirement and replacement rates in the Uruguayan

118

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

The figure below shows the evolution (and projection in dotted lines) of income by

gender. It should be noted that the graph considers the evolution of income

only for individuals who continue to contribute to the social security system.

In this sense, the further ahead in time, the fewer individuals continue con-

tributing, and those who do so belong to the younger cohorts and have the

best contribution and income profiles. We present the coefficients of the esti-

mated model for each income quintile in Table A.2 in the Appendix.

Figure 3. Monthly wage projections in the base scenario (2017 average in US dollars)

Year

Total Male Female Projection

500

1.000

1.500

2.000

2.500

3.000

3.500

1996

1998

2000

2002

2004

2006

2008

2010

2012

2014

2016

2018

2020

2022

2024

2026

2028

2030

2032

2034

Note: The graph considers the evolution of income only for individuals who continue to contribute to

Social Security. The further ahead in time, the fewer individuals continue to contribute and the better

their income profile.

Source: Authors´ own computations derived from social security data.

C. Reconstruction and projection of funds in the

individual savings accounts and annuities

Firstly, for each individual in the sample we reconstruct the history of the

contributions made to the individual account, which depends on the personal

contributions earmarked for this pillar. They are related to each person’s income

level and the choice of voluntary affiliation to the AFAP provided by Article 8

of Law 16,713. As we mentioned before, there is an administration fee charged

by the AFAP, and a collective disability and death insurance premium.

Gioia de Melo

et al.

119

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

The sample contains information about the date of affiliation with the AFAP

for each individual and whether they chose to affiliate under the option pro-

vided by Article 8 of Law 16,713 but does not inform about the administrator

chosen or assigned to them. Using information published by the Central Bank,

we construct an average AFAP of the system (AFAP-System), calculated from

the relative monthly participation of each administrator in the FAP. In this way,

we calculated the average monthly quota of the AFAP-System between 1996-

2005, which reflects the historic average rate of return of the system as well

as the administrative fees paid and disability and death insurance premium.

Based on the information from the BPS administrative records, we compute

the individual’s average gross contribution to the savings pillar and subtract

commissions paid, thus obtaining the net contribution to the FAP. From the

quotient between the net monthly contributions and the average quota value

of the AFAP System, the number of FAP shares (“cuotas”) that are acquired

monthly by each worker are estimated. Thus, when valuing the number of

shares acquired by the worker, at any given time, we obtain the total amount

of savings in the individual account. For the projection, we assume an aver-

age yield of 3.5% measured in indexed price units (UI), which are equivalent

to 1.5% over the evolution of the average wage. We assume that the expected

performance of the accumulation sub-fund is identical to that of the retire-

ment sub-fund.

Figure 4 shows the projected FAP at age 60 for ages in January 2017 (horizon-

tal axis). We can observe the “cohort effect” in relation to the specific situa-

tion within each age subgroup in the sample.

The age bracket in our sample contains individuals who were of different

ages at the time when the multi-pillar system was introduced. This results in

them having a different total number of years’ contribution to their individual

savings accounts. Those close to 60 by January 2017 will have contributed a

maximum of 21 years to their individual savings accounts: on average 80%

less than those in their forties.

Based on the ratio between the FAP estimated at 60 years old and the sum

of gross contributions to the individual, savings accounts at constant values

(updated by the Consumer Price Index), we obtain a measure of the return of

the contributions to this pillar (before paying for administration fees). Accord-

ing to our estimates, the average ratio at 60 years old for the entire sample is

1.33, which implies that, on average at this age, the FAP is 33% higher than

the sum of gross contributions to this pillar.

Eligibility for retirement and replacement rates in the Uruguayan

120

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Figure 4. Projected FAP at age 60 (2017 average in US dollars)

Ages

10.000

15.000

20.000

25.000

30.000

35.000

40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

60 years old Average 60 years old

Source: Authors´ own computations derived from social security data.

Figure 5 shows that this ratio is increasing with the person’s age. This evolution

indicates that older cohorts benefited to a larger extent from the high rates

of return on Unidades Reajustables (unit value that is periodically adjusted

according to the evolution of wages) registered until 2004, which we do not

expect to recur in the future. This partly compensates for the shorter con-

tribution of older individuals and thereby helps explain the low variation in

replacement rates among cohorts.

According to our estimates, 29% of the cohort under study will never contrib-

ute to the individual savings pillar (AFAP). In turn, those that are expected to

contribute for more than 30 years represent only 13%.

Regarding the calculation of life annuities (RVP for its Spanish acronym), we

use the mortality tables published by the Superintendence of Financial Services,

part of the Central Bank of Uruguay (SSF-BCU), at the beginning of 2017. This

happened at a time during the public consultation on modifications to the indi-

vidual savings scheme promoted by this institution.10 The tables published by

the BCU include the update and dynamization of mortality tables, an increase

10 We chose to use these tables given that they improve the calculation of RVPs compared to those in

use, which were calculated using the 1996 Census.

Gioia de Melo

et al.

121

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

in the reference population, and a better adjustment of the surcharge for the

risk of survival. The update and correction of the probability of generating a

pension for a family member in the event of the death of the RVP holder also

represents an improvement relative to the previous tables. Likewise, for the

calculation of the RVPs, we decided to use the current technical interest rate,

which was set at 1.5% in UR by the SSF-BCU.11

Figure 5. Ratio between FAP at age 60 and gross contributions to individual savings

accounts by age (%)

Ages

40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

60 years old

1,00

1,05

1,10

1,15

1,20

1,25

1,30

1,35

1,40

1,45

Lineal (60 years old)

Source: Authors´ own computations derived from social security data.

V. Results

A. Years of contributions in the baseline scenario

Table 2 presents the distribution of retirement ages when individuals retire as

soon as they become eligible (immediate retirement), which is our baseline

11 Decree No. 221/017, published on 15 August 2017, eliminated the differentiation by sex in the mortality

tables, and this only allowed differentiation of premiums by age. It is expected that this change will improve

annuities for women as they have a higher life expectancy than men. This effect is likely to exceed the

opposite sign adjustments associated with the probability of generating a pension, the partial (although

limited) effect of which is expected to favor men. However, as stated above, the Mortality Tables that

come from adopting this measure are not yet available, and therefore are not part of this study.

Eligibility for retirement and replacement rates in the Uruguayan

122

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

scenario. To present the results in a clearer way, we choose to analyze the

situation of each individual in the sample with respect to their possibilities

of becoming eligible for retirement at three different ages: 60, 65, and 70. An

individual can retire at age 60 (“the normal age”) provided s/he has contrib-

uted for thirty years. If s/he has not done so at 60 years old, but does before

the age of 65, for simplicity we assume s/he will retire at 65. In addition, if

the person has accumulated between 25 and 29 years of contributions at age

65, we assume they retire as part of the old age regime. Furthermore, under

this regime, a person can retire at age 70 if s/he has contributed for between

15 and 24 years. Estimates for years of contributions take into account special

computations and the bonus contribution years per child for women.

Table 2. Distribution of retirement ages in the baseline scenario (%)

Total Men Women Cohort

40 - 44

Cohort

45 - 49

Cohort

50 - 54

Cohort

55 - 60

Quintile

Eligible for retirement at age 60

Normal

age 51.1 52.4 49.8 54.0 50.7 49.1 50.3 19.7 25.3 47.4 73.9 89.4

Eligible for retirement at age 65

Normal

age 5.4 5.5 5.3 6.3 6.7 5.0 3.4 3.0 7.3 9.5 5.7 1.6

Old age 6.3 6.4 6.1 6.1 7.8 6.3 4.8 5.0 9.4 10.1 5.1 1.8

Eligible for retirement at age 70

Old age 8.4 7.7 9.3 5.2 6.7 10.5 12.2 9.4 15.0 11.0 5.2 1.6

Do not become eligible for retirement in the defined benefit scheme

28.7 28.0 29.5 28.5 28.0 29.1 29.3 62.9 42.9 22.0 10.1 5.5

Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0

Note: Calculations in Table 2 refer to our base macroeconomic scenario. Both the optimistic and moderate

macroeconomic scenarios also project that, on average, 51% of the sample will become eligible for retire-

ment at sixty. The main difference by scenario is observed for those who become eligible at 65 years old:

7.2% in the optimistic and 3.9% in the moderate. This is because the projected macroeconomic growth for

2017 onwards would have a greater impact on contribution densities projected at ages above 60 given the

age structure of the sample. Therefore, the percentage of individuals who are not likely to retire from

the PAYG scheme varies from 24.6% in the optimistic scenario to 30.5% in the moderate.

Source: Authors’ own computations based on social security records.

We estimate that 51% of the workers would be able to retire at age 60. This

estimation is higher than that projected by Bucheli et al. (2006, 2010) using

administrative records for the period 1996 – 2004: 38 and 41.3%, respectively.

Gioia de Melo

et al.

123

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

This could be explained by the sustained economic growth, lower unemploy-

ment rate, and decreasing informality rates observed during more recent years,

as well as the bonus for the contribution per child introduced by Law 18,395

in 2008. Indeed, the percentage of workers not contributing to social secu-

rity decreased significantly from 40.7% in 2004 to 24.7% in 2017 (the aver-

age informality rate in the 1996-2004 period considered by previous studies

was 41.1%).

Men exhibit a slightly more favorable prospect in terms of their retirement

although the gender gap seems to have significantly reduced in the last decade

due to both bigger increases in the percentage of women contributing to social

security as well as the 2008 reform that granted women an additional year

of contributions per child (up to five). Also, there is a slightly better situation

for the youngest cohort. On the other hand, we also observe that 28.7% of

our sample would not be eligible for retirement in the PAYG scheme even at

age seventy. About 60% of individuals who are not expected to retire by the

pay-as-you-go scheme will accumulate five years’ contributions or less. These

individuals are likely to be covered by the non-contributory pension although

it currently targets individuals living in households with extremely low income.

B. Replacement rates in the baseline scenario

A pension replacement rate is the defined as the pension entitlement divided

by labor income . It can be defined in several ways, especially depending on

the labor income considered. The most popular measure uses the wage earn-

ings of the year prior to retirement. In this paper, we analyze the ratio consid-

ering two earning periods: average earnings from the last year the individual

worked and their average working life earnings.

Using earnings from the last year worked is more suitable to assess the pension

system’s ability to smooth consumption between working and retirement peri-

ods, which is what a worker would evaluate when deciding whether to retire.

Moreover, this approach enables us to make international comparisons as it

is the most widely used ratio in the literature. In turn, the analysis of replace-

ment rates expressed relative to average earnings throughout the emplo-

yment history provides a better picture of the value of the pension in terms of

the contributions the worker made during his active stage. It should be noted

that this average only comprises periods with positive earnings.

Eligibility for retirement and replacement rates in the Uruguayan

124

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

We define the joint replacement rate as the ratio of the sum of the pension

of the PAYG scheme and the annuity received from the individual savings

scheme relative to total labor income for which the worker contributed. The

PAYG rate considers the pension in that pillar in terms of the wage it covers,

and the individual savings’ replacement rate is defined accordingly (annuities

in relation to income devoted to individual saving).

In this section, we show our results in terms of replacement rates in the

immediate retirement scenario discussed in Table 2, section 5.1. In addition to

showing average replacement rates, in Table 3 we include median estimates

to show to what extent average rates are sensitive to extreme values.12 We

include the average contribution density and most frequent earnings quintile

to emphasize that replacement rates at different ages are not comparable as

they involve individuals with very different contribution and income profiles.

Furthermore, it is worth mentioning that the effective replacement rates shown

in this study should not be compared to technical rates used for the calcula-

tion of pensions. This is, firstly, because our replacement rate estimates con-

sider the existence of minimum and maximum pension amounts. Secondly,

it is because when the pension is calculated by the BPS, technical rates are

applied to the average wage earned in the last 10 years, and the maximum of

the average wage earned in the best twenty years is increased by five percent.

Instead, as we already mentioned, to compute effective replacement rates we

consider the ratio between the estimated pension to average earnings from the

last year worked in one case and average working life earnings in the other.

Average replacement rates are significantly affected by extreme values due to

the high incidence of minimum pensions that we estimate 45% of the sample

will receive. At the same time, 23% of those who will receive minimum pen-

sion declare particularly low wages (below the national minimum wage for

44 hours per week). It should be noted that we could not obtain good quality

information on hours worked by the individuals in the sample. The low levels

of declared income have a substantial impact on the averages calculated for

the sample as a whole, which can be seen by comparing the means and medi-

ans of individuals who become eligible at different ages.

12 All tables exclude 1.4% of individuals in the sample that had replacement rates relative to the last

year’s earnings that were above 300%.

Gioia de Melo

et al.

125

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Table 3. Replacement rates in the baseline scenario (%)

Relative to last

year’s earnings Relative to working life average earnings

Average

contribution

density a

Most

frequent

quintile

Mean Median Mean Median

JointbJoint Joint PAYG

pillar

Individual

savings Joint PAYG

pillar

Individual

savings

Eligible for retirement at age 60

Normal age 51.9 41.3 79 108.4 25.6 61.2 82.8 25.3 0.87 Q 5

Eligible for retirement at age 65

Normal age 61.7 50.3 92.7 131.2 28.3 74.8 107.8 29.3 0.56 Q 3

Old age 68.1 56.2 97.9 133.4 25.5 76.7 106.7 25.4 0.47 Q 3

Eligible for retirement at age 70

Old age 88.7 77.6 116.7 155.6 25.3 95 121.2 24.3 0.39 Q 3

Total 58.6 44.8 86.2 118 25.8 66.1 90.0 25.5 0.75

a Refers to the observed period in the sample: 1996-2015

b Comprises both the pay-as-you-go and the individual savings schemes

Source: Authors’ own computations based on social security records.

On average, those eligible for retirement at age 60 are expected to obtain a

replacement rate of approximately 52% with respect to the average earning

of their last year’s contribution. On the other hand, the replacement rate in

relation to the average salary of the individual’s employment history is 79%.

This is consistent with the fact that wages generally increase throughout

working life, therefore pensions expressed as a ratio of the last year’s earnings

are lower than when expressed as a ratio of working life earnings. Individuals

eligible for retirement at age 60 on average present a high contribution den-

sity (87%) and mainly belong to the fifth income quintile.

The remainder groups that would be eligible for retirement at age 65 and 70

(with lower contribution density and income than those eligible at age sixty)

have higher replacement rates due to both the higher incidence of minimum

pension recipients and the higher technical rates that are applied when indi-

viduals are older at the moment of retirement.

Considering all those individuals who will at some stage be eligible for retire-

ment, and relative to their last year of earnings, average replacement is

Eligibility for retirement and replacement rates in the Uruguayan

126

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

estimated at 58.6%, and the median is approximately 45%. Likewise, replace-

ment rates calculated using the average income of working life have an aver-

age of 86% and a median of 66%.

Focusing on the relationship between pensions and average working life earn-

ings, we observe that the replacement rates of the PAYG scheme are on aver-

age above 100% for different ages and regimes (normal age and old age). This

result is strongly influenced by the presence of the minimum pension benefi-

ciaries, which we analyze in a following section. Therefore, those who man-

age to accumulate at least 15 years’ contributions would receive, on average,

a pension above their working life earnings. This may suggest sustainability

challenges that should be analyzed in future research.

Individual savings’ replacement rate estimates are on average 25%. They are

expressed with respect to the wage covered by that pillar over the course of

working life. We do not calculate the replacement rate relative to the last year

of earnings for the individual savings pillar because it may be the case that

workers do not contribute to their individual account during the year prior to

their retirement. When comparing replacement rates for the two pillars, it is

important to bear in mind that while the PAYG pillar also receives contribu-

tions from employers, the funds for individual savings only come from work-

ers. Furthermore, the BPS provides other coverage, such as unemployment

insurance, which does not have a specific financing. Finally, the replacement

rates for the PAYG pillar do not consider the administration costs of the pen-

sion system, which are covered by the fee that AFAPs charge workers in the

case of the individual savings scheme.

Figure 6 is a histogram showing the joint replacement rate (comprises both

PAYG and individuals’ savings pillars) computed in terms of the last year’s

earnings.

Although it is difficult to compare studies across and within countries due to

the different assumptions involved in each exercise, we conclude that Uru-

guayan replacement rates do not seem to differ much from those estimated

for the average of OECD countries and other specific studies available for Chile.

Gioia de Melo

et al.

127

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Figure 6. Histogram showing joint replacement rates calculated relative to the average

wage in the last twelve months of contributions

0100 200 300

Joint replacement rate

1,00

1,20

1,30

1,45

Density

Source: Authors´ own computations based on social security records.

OECD (2013) estimates a replacement rate of approximately 54% for middle-

income workers of 34 member countries, which to some extent is comparable

to our replacement rate of 51.9% estimated for those eligible for retirement

at age 60 and relative to the last year’s earnings. However, it should be taken

into account that OECD (2013) assumes individuals exhibit continuous con-

tributions for forty years (i.e. no informality or unemployment spans).

According to our estimates, Uruguayan replacement rates that calculated

the last year’s earnings seem to be slightly higher than those computed by

Bernstein et al. (2006) for the average of Chile’s affiliates (44% in relation to

last month’s and the last three years’ wages). As for replacement rates com-

puted using the average working life earnings, according to Berstein et al.

(2006), Chilean replacement rates are estimated at 121%. Chile requires men to

retire at age 65, and the study also considers zero income during periods with-

out contributions when computing average income. Our estimated replacement

rate of 79% at age 60 did not include zero income periods when computing

Eligibility for retirement and replacement rates in the Uruguayan

128

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

average working life earnings, but if we did the same, this estimate would

rise to 131%. It should also be taken into account that we only have infor-

mation on wages from 1996 onwards, that is, we lack information on the

older cohorts’ first years of work, and this can result in overestimating aver-

age working life earnings.

Another study on Chile carried out by Paredes and Díaz Fuchs (2013), esti-

mates replacement rates based on average income and pensions rather than

on individual results. In this case, the results for retirement at the legal age

result in an average gross replacement rate of 81%, with respect to working

life earnings for both men and women. This is in line with our estimates for

Uruguay (79% at age 60).

C. Impact of the minimum pensions benefit

According to the social security agency, 28% of the individuals who retired

in 2015 received the minimum pension. It is reasonable to expect this figure

to increase from 2016 onwards as those who contribute to the multi-pillar

system begin to retire. This is due to the high percentage of workers with

income lower than $48.953 (pesos as of January 2017) that voluntarily chose

to assign 50% of their contributions to the individual savings account (Arti-

cle 8, Law 16,713). As such, the amount of contributions to the PAYG pillar

significantly decreased and that makes it more likely for workers to receive a

minimum pension.

By 2017, the minimum contributive pension was $9,930 (approx. US 350

dollars).13 We assume this minimum evolves in line with the variation in real

wages: an increase of 2% per year. According to our projections, minimum

pension recipients would reach 45% of those eligible for retirement (70% of

those expected to receive a minimum pension chose to assign half of their

contributions to the individual savings pillar (Article 8, Law 16713). As men-

tioned before, this has a significant impact on average replacement rates.

Our findings suggest that there is a significant percentage of individuals

who have high contribution densities and a particularly low monthly income

(possibly explained by declaring only a few working days per month), and

13 Established by Decree No. 252/2016: in force since January 1, 2017.

Gioia de Melo

et al.

129

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

who would therefore qualify for a minimum pension. We estimate that 34%

of those who would become eligible for retirement at age 60 would obtain

the minimum amount: this percentage is higher for women (39.7%) than it

is for men (29.3%). This proportion is even higher for those who become eli-

gible for retirement at 65 years old and reaches 84% for those who would

only be able to retire at 70.

According to our estimates, beneficiaries of the minimum pension on average

duplicate the pension they would have received if no minimum existed. None-

theless, the current system of access to minimum pension does not provide

sufficient information to evaluate whether this benefit is targeting the poorest.

Table 4. Minimum pension recipients (%)

Total Men Women

Eligible for retirement at age 60

Normal age 34.1 29.3 39.7

Eligible for retirement at age 65

Normal age 62.1 54.1 71.2

Old age 73.0 66.9 80.1

Eligible for retirement at age 70

Old age 84.1 78.6 89.0

Total 45.0 40.0 52.0

Source: Authors’ own computations based on social security records.

Following Decree 233/017, if we add the annuity derived from the individual

saving accounts to the resulting pension from the PAYG to determine whether

the individual would qualify for a minimum pension, the percentage of indi-

viduals who would receive a minimum pension would drop from 45% to 38%.

However, the average replacement rate remains barely unchanged (58.4%

relative to last year’s earnings and 85.6% relative to average lifetime earn-

ings: compared to 58.6 and 86.2 reported in Table 3). This corresponds to the

fact that the projected amount of annuities for the individuals who would no

longer receive a minimum pension are expected to be very low.

Eligibility for retirement and replacement rates in the Uruguayan

130

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

D. Analysis based on socioeconomic characteristics

Table 5 shows the average replacement rates by gender, income quintile, and

contribution sector for individuals who are expected to be eligible for retire-

ment at age 60. Women exhibit lower participation rates and lower wages than

men. These factors have a negative influence on eligibility for retirement and

the expected pension amounts. However, because of the introduction of the

one-year contribution per child scheme and the existence of minimum pen-

sions, the projected replacement rates suggest that women will on average

achieve higher replacement rates than men at the PAYG pillar and, therefore,

also higher joint replacement rates. The computation of pensions in the indi-

vidual savings scheme considers different mortality tables by gender, which

reflects women’s higher life expectancy. This means that, for the same total

savings, women have to finance a longer period of pensions than men, which

results in lower amounts.14

Pensions play a significant redistributive role; with substantially higher replace-

ment rates at lower wage levels. Average rates among quintiles range from

124.8% in Q1 to 33.6% in Q5 relative to the last year’s earnings. These pat-

terns correspond with the significant incidence of minimum pensions in Q1, Q2,

and to a lesser extent Q3. Overall these results show that the PAYG pillar plays

an important redistributive role for those who become eligible for retirement.

According to our estimates, workers in rural and housekeeping sectors have

the highest replacement rates: around 100% relative to the last year’s earn-

ings. Again, this fits with the large percentage of workers from those sec-

tors that will receive a minimum pension when they retire (95% and 76%, in

housekeeping and the rural sector, respectively). We argue that this is both

due to low wages in these sectors and a high percentage of undeclared income

(note that only individuals expected to be eligible for retirement at age 60

are considered in Table 5). Indeed, the Household Survey provides informa-

tion on the percentage of undeclared income: the primary sector and house-

keeping services in 2017 were the sectors that had the highest percentage of

14 The tables used in this study are those published as a proposal by the Central Bank in 2017. However,

in 2018 the Central Bank substituted these tables with a unique table for men and women, removing

the difference in pension amounts due to life expectancy.

Gioia de Melo

et al.

131

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

workers who stated that they did not declare social security for a percentage

of their income.

Overall, replacement rates in the individual savings scheme do not show

substantial differences. Most variation in joint replacement rates for gender,

quintiles, or sectors is explained by differences in the percentage of mini-

mum pension recipients.

Table 5. Replacement rates for those eligible for retirement at age 60 by characteristic

Relative to

last year

earnings

Relative to average lifetime earnings Minimum

pension

recipients (%)

Joint Joint Pay-as-you-go Individual

savings

Total 51.9 79.0 108.4 25.6 34.1

Gender Male 47.5 72.4 99.8 25.8 29.3

Female 57.1 86.6 118.3 25.3 39.7

Income

quintile

Q1 124.8 226.6 288.1 26.4 88.6

Q2 80.4 112.3 156.3 26.5 83.2

Q3 55.3 79.3 112.5 25.8 72.3

Q4 41.6 62.3 86.8 25.4 17.6

Q5 33.6 49.4 69.1 25.3 0.2

Contribution

sector

Industry and

retail 45.8 68.1 95.7 25.1 33.0

Public sector 37.9 54.9 74.0 27.5 5.2

Rural 95.6 161.4 210.0 25.5 75.7

Construction 42.8 61.2 90.6 24.6 28.1

Housekeeping

services 113.3 170.6 229.5 22.2 95.4

Source: Authors’ own computations based on social security records.

Comparing replacement rates by age cohorts is particularly relevant given that

in 2016 the first workers covered by the 1996 reform began to retire. A group

of these workers, who called themselves the “cincuentones” (those in their

fifties), argued that they would receive lower pensions compared to slightly

older workers (not covered by this regime) and to those younger than them

and under the same social security system. This triggered a partial reform in

2017 that allowed the cincuentones to abandon the multi pillar system and

Eligibility for retirement and replacement rates in the Uruguayan

132

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

have their pension calculated as 90% of the pension they would have obtained

under the PAYG system, which was in place prior to 1996.

In order to compare replacement rates between different age cohorts, we

believe it is not desirable to express them relative to the average lifetime

working life income because administrative data including income is only

available from 1996 onwards. Therefore, for the older cohorts (ie. those aged

55 – 60 in 2017), we do not consider the income they earned during earlier

stages of their careers. In turn, for those aged between 40 and 44 in 2017 we

are likely to have data that covers almost their whole work history. For this

reason, we consider that the relevant replacement rates that should be com-

pared between different age groups (given the information available) are those

expressed relative to the last year and relative to the average of the wages

earned between age 40 and 60.

Table 6 breaks down average replacement rates by cohorts and quintiles using

the two types of replacement rates. The results show that prospects for those

in their fifties in terms of replacement rates are not that different to those of

younger cohorts with similar income levels (joint replacement rates for those

in Q5 are similar among age cohorts). The reason why we do not find signifi-

cant differences is because even if individual savings replacement rates are

lower for those in their fifties (this cohort had less time to grow their funds

in the individual savings scheme), the PAYG pillar weighs significantly more in

the joint replacement rate and there is no clear pattern between cohorts

in this pillar. Indeed, when comparing replacement rates relative to the last

year’s earnings, older cohorts in Q5 exhibit slightly higher joint replacement

rates due to the PAYG pillar while when replacement rates are computed rela-

tive to wages earned between age 40 and 60, these are slightly lower. Over-

all, we conclude that the individual savings pillar plays a minor role in terms

of its contribution to the joint replacement rates for all the age cohorts that

were analyzed.

E. Sensitivity analysis

Replacement rates under different macroeconomic scenarios show two types of

behavior. On the one hand, when considering the last year’s earnings, replace-

ment rates decrease as macroeconomic conditions improve. This is explained

Gioia de Melo

et al.

133

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Table 6. Replacement rates at age 60 by cohort and quintile (%)

Age Cohort 40-44 45-49 50-54 55-60

Relative to the last year’s earnings

Quintile 1 91.8 132.7 158.5 124.7

Quintile 2 65.9 75.5 86.5 102.4

Quintile 3 50.7 50.5 59.3 66

Quintile 4 40 40.8 42.2 44.7

Quintile 5 31.5 32.1 34.5 36.2

Total 47.3 50.5 56.4 55.4

Relative to average income 40-60 years old

Joint

Quintile 1 133.2 216.9 267.9 204.5

Quintile 2 91.5 99.4 109.3 115.3

Quintile 3 69.1 68.6 73.3 73.2

Quintile 4 54.9 55.2 54.9 53.1

Quintile 5 43.0 42.8 44.3 42.3

Total 80.6 82.7 84.0 67.7

Pay-as-you-go

Quintile 1 186.4 289.5 335.8 250.4

Quintile 2 137.9 145.9 152.0 148.3

Quintile 3 104.5 100.4 101.5 102.9

Quintile 4 80.3 78.6 77.8 71.7

Quintile 5 62.0 60.6 61.9 61.8

Total 114.5 114.5 111.3 90.4

Individual savings

Quintile 1 24.3 26.2 27.6 18.7

Quintile 2 24.9 24.1 24.1 19.2

Quintile 3 24.3 26.3 23.1 19.2

Quintile 4 29.5 29.7 20.0 15.7

Quintile 5 24.5 22.8 22.1 17.6

Total 30.8 28.0 23.5 17.5

Source: Authors’ own computations based on social security records.

Eligibility for retirement and replacement rates in the Uruguayan

134

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

by the fact that wage growth projected towards the end of individuals’ work-

ing lives is higher in the most favourable scenarios. Average rates decline by

about eight percentage points, from an average of 64% in the moderate sce-

nario to 58.6% and 56.6% in the base and optimistic scenarios, respectively.

In terms of replacement rates in relation to average working life earnings, we

observe the opposite behaviour. That is, they grow under the most favourable

scenarios, implying a greater correlation between contributions over the entire

working life and the resulting retirement. Average rates in the extreme sce-

narios increase 10 percentage points from 82.9% to 92.2% and from 64.6%

to 69.5%. In these scenarios, we also estimated the number of people receiv-

ing minimum retirement benefits, assuming a minimum retirement growth of

3% and 1%, respectively for the optimistic scenario and moderate scenarios.

In the moderate scenario, we estimated that 41% of the individuals eligible

for retirement would receive the minimum pension, while 49% would do so

in the optimistic scenario. Overall the main conclusions remain unchanged.

As a robustness check, we calculate replacement rates under the assumption

that individuals’ income from 2016 onwards grows at an annual rate of 2%

in real terms (provided they continue contributing to the social security sys-

tem); thus, we ignore the income model described in Section 4.2. Replace-

ment rates do not differ significantly from those reported in Table 3 (they are

slightly higher relative to the last year’s earnings but slightly lower relative to

working life earnings). Under this scenario, 47% of those eligible for retire-

ment would receive a minimum pension.

We calculated the effect of a one percentage point decrease in the variable

administrative fee charged by the AFAPs on the individual saving accounts

replacement rate and an increase of one percentage point in the rate of return

of the Fondo de Ahorro Previsional (FAP). The base macroeconomic scenario

was applied in both cases.

We estimate that a reduction of one percentage point in the administration

fee charged by the AFAPs will have a positive effect of 0.9 percentage points

on the rate obtained by this pillar, while an increase in the rate of return

of the same magnitude will increase the replacement rate by 2.4 percent-

age points. The limited effects estimated can be explained by the age profile

of the cohort analyzed, for whom the potential effects of these parametric

changes have a short-term impact: i.e. the last years of their work history. As

Gioia de Melo

et al.

135

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

expected, when analyzing the results after splitting them by smaller cohorts,

we can see that the younger workers have been impacted the most by these

exercises. A decrease in one percentage point in the administration fee the

younger members of the sample were charged would imply an average increase

in the replacement rate of 1.5 percentage points at age 60. Furthermore,

an increase of equal magnitude in the projected rate of return would result in an

increase in the replacement rate of 4 percentage points.

F. Replacement rates in the deferral scenario

In this section we estimate the effect the replacement rates have on decid-

ing to postpone retirement for five years under the baseline macroeconomic

scenario. We did this for a subgroup (43.6% of the sample) that is expected

to have accumulated at least 30 years’ contributions at age 60 and for whom

the contribution model projected they would continue to contribute at least

until age 65. The results are presented in the table below.

Table 6. Comparison of replacement rates between immediate retirement at age

sixty and deferred retirement until age 65

Relative to last year earnings

Mean Median

Total Men Women Total Men Women

Chooses to retire at 60

Joint 50.5 46.7 55.0 40.6 39.9 41.6

Chooses to retire at 65

Joint 57.6 54.1 61.7 51.1 50.7 51.4

Relative to working life average earnings

Chooses to retire at 60

PAYG pillar 106.5 98.8 115.6 81.1 79.2 84.5

Individual savings 26.0 26.2 25.8 25.7 26.4 25.0

Joint 77.8 72.2 84.5 60.6 59.0 63.2

Chooses to retire at 65

PAYG pillar 125.8 119.3 133.5 105.5 103.7 108.1

Individual savings 38.5 38.8 38.1 38.1 39.1 36.8

Joint 93.2 88.3 99.0 80.8 79.2 82.8

Source: Authors’ own computations based in social security records.

Eligibility for retirement and replacement rates in the Uruguayan

136

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

We estimate that the joint replacement rate measured with respect to the

income of the last 12 months would increase on average by approximately 7

percentage points if retirement was postponed for five years (increasing from

50.5% to 57.6%).

The decision to continue working results in an average increase in pensions

of 41% in real terms. However, the subgroups’ labor income is also projected

to increase in real terms during those years (18% in real terms). This attenu-

ates the increase of the replacement rate calculated using the last year’s earn-

ings. Within this subgroup of individuals with a high contribution density, it

is expected that 29.8% would receive the minimum pension if they retired at

age 60, whereas if they postponed their retirement until age 65, this percent-

age would be reduced to 18%.

With regard to the replacement rate measured in terms of the average income

of the individuals’ employment history, the joint rate would increase by 15

percentage points from 77.8% to 93.2%. That is, if individuals who accumu-

late at least 30 years’ contributions at age 60 continued to work until age

65, they would have a replacement rate of almost the average income earned

during their working life.

Individual savings´ replacement rates under a scenario where individuals post-

pone their retirement until age 65 are estimated at 38.5%. This is in line with

rates computed by Durán and Pena (2011) for Uruguay using household surveys.

Overall, our results suggest that the PAYG pillar has only mild incentives to

defer retirement. This is not only related to technical rates but also to the

incidence of minimum pensions. Alvarez et al. (2010) suggest incentives for

postponing retirement are low compared to advanced economies. Also, Cama-

cho (2016) argues that the premia added to technical replacement rates for

each additional year of service relative to the minimum required for a given

age (0.5%) is insufficient from a financial-actuarial perspective. It should be

mentioned that while in Uruguay technical replacement rates increase by 3

percentage points for every additional year of deferred retirement (and addi-

tional year of contribution) in the PAYG pillar, pensions increase by 10.4 per-

centage points in the United Kingdom, 8.4 percentage points in Japan, and 8

percentage points in the United States (OECD, 2013).

Gioia de Melo

et al.

137

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

VI. Conclusions

In this paper, we estimate eligibility for retirement and replacement rates of

the Multi-Pillar security system in Uruguay for a cohort of salaried workers

using a random sample of individual social security records.

The main results in terms of access to the contributive system show that there

is a significant degree of polarization: while approximately half of the cohort

would become eligible for retirement in the PAYG pillar at 60 years old, almost

30% would not be able to retire even at age 70. Most of the latter are expected

to accumulate less than five years’ contributions in their working life.

With regard to the adequacy of benefits, although we acknowledge the dif-

ficulty of making international comparisons, we believe that our estimated

joint replacement rate for the multi-pillar system at age 60 with respect to

last year’s earnings (52%) is comparable to the OECD’s average (54%), accord-

ing to OECD (2013), and higher than the minimum suggested by ILO (40%). It

is important to consider that pensions in Uruguay are indexed exclusively to

wage inflation, while according to OECD (2013) most countries do so relatively

to consumer price inflation. Thereby, two countries with similar replacement

rates at the moment of retirement may exhibit divergent trajectories if com-

pared years after retirement.

Although our results indicate significant differences by income quintiles in

terms of access to the contributive system, once individuals are eligible for

retirement the system has a marked distributive profile. The existence of a

minimum pension and the very recognition of years of contributions per child

generate an important redistributive effect on the transition from active to

passive life, which is shown by high replacement rates in the PAYG pillar for

the lower quintiles. In future research, it would be interesting to assess the

overall impact of the pensions system on inequality.

Our finding that 45% of those eligible for retirement would receive a mini-

mum pension also points to the relevance of addressing challenges regard-

ing informality both in terms of the percentage of workers who contribute to

social security and the amounts they declare. It should be highlighted that this

finding could only be obtained thanks to the use of individual administrative

records. The fact that for a relatively important part of workers the declared

Eligibility for retirement and replacement rates in the Uruguayan

138

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

income is smaller than the minimum wage suggests the need to revisit the

contribution incentives according to actual labor income. Also, minimum pen-

sions imply implicit subsidies to the PAYG pillar that should be quantified and

their ability to target poverty assessed. Eventually, the role of the contributive

and non-contributive pillars should be revised.

Overall, our results suggest that the PAYG pillar’s incentives to defer retire-

ment are mild. In the same vein, Alvarez et al. (2010) and Camacho (2016) have

pointed out challenges regarding incentives to postpone retirement because

of the current technical replacement rates. As for the individual savings pil-

lar, we conclude that this pillar plays a minor role in terms of its contribution

to the joint replacement rates. This is not only the case for older workers who

contributed to the multi-pillar system for a few years but also for younger

workers whose entire work history is covered by it.

When comparing results by age cohorts, we conclude that there are no sig-

nificant differences between the average joint replacement rates estimated

for individuals currently aged 49 and those aged 50-60, even when we break

down the results by income quintiles. This discussion is particularly relevant

for policy issues since, in 2017, a law was passed to allow the “cincuen-

tones” to be subject to conditions similar to those prevailing prior to the

1996 reform. In practice, the implemented changes delay the maturation of

the multi-pillar system and just perpetuate the previous regime the solvency

of which was questioned; thus, higher financial pressure was imposed on

the system.

This study analyzes coverage and retirement-income adequacy of the Uru-

guayan social security system. Some of the estimations in this paper rein-

force the ever-present need to assess the sustainability of the PAYG pillar in

the medium term. In this sense, our finding that average replacement rates

in this pillar relative to average working life earnings exceed 100% is a mat-

ter of concern. Analyzing the sustainability of the pension system should be

addressed in future work as Uruguay has reached almost full coverage if both

contributory and non-contributory schemes are considered and demographic

tensions are expected to become even more challenging in the future. Indeed,

the analysis of the sustainability of pension systems is likely to become part

of a global research agenda as many countries are beginning to face similar

demographic challenges.

Gioia de Melo

et al.

139

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Acknowledgments

This study was coordinated by Centro de Estudios Fiscales (CEF) and counted

on the participation of professionals from the Ministerio de Economía y Finan-

zas and the Ministerio de Trabajo y Seguridad Social. It did not require addi-

tional funding.

We are particularly thankful to the two anonymous reviewers for their com-

ments and suggestions as well as for the technical support received by the

Inter-American Development Bank (IDB). We are also grateful for advice

received from Javier Alejo, Marcelo Bérgolo, Mariano Bosch, Rodrigo Ceni,

Álvaro Forteza, Matías Giacobasso, Andrés Masoller as well as from seminar

participants at the Facultad de Ciencias Económicas y de Administración, the

Inter-American Development Bank, and the Banco Central del Uruguay. Finally,

we are very thankful to the Banco de Previsión Social for providing access to

the data used in this study. All errors are our own.

References

1. Álvarez, I., Silva, N., Forteza, A., & Rossi, I. (2010). ¿Qué incentivos al

retiro genera la seguridad social? El caso uruguayo. Latin American

Journal of Economics, 47(136), 217-247.

2. Altamirano, A., Berstein, S., Bosch, M., García-Huitrón, M., & Oliveri, M.

(2018). Presente y futuro de las pensiones en América Latina y el Caribe.

Monografía del BID, 649. Retrieved from: https://publications.iadb.org/

es/presente-y-futuro-de-las-pensiones-en-america-latina-y-el-caribe

3. Bernstein, S., Larrain, G., & Pino, F. (2006). Chilean pension reform:

Coverage facts and policy alternatives. Economía, 6(2), 227-279.

4. Bucheli, M., Ferreira-Coimbra, N., Forteza, A., & Rossi, I. (2006). El acceso

a la jubilación o pensión en Uruguay: ¿Cuántos y quiénes lo lograrían?

Comisión Económica para América Latina y el Caribe (CEPAL). Serie

Estudios y Perspectivas. Oficina de CEPAL, Montevideo. Retrieved from:

https://www.cepal.org/es/publicaciones/5157-acceso-la-jubilacion-o-

pensiones-uruguay-cuantos-quienes-lo-lograrian

Eligibility for retirement and replacement rates in the Uruguayan

140

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

5. Bucheli, M., Forteza, A., & Rossi, I. (2010). Work histories and the Access

to contributory pensions: The case of Uruguay. Journal of Pension

Economics and Finance, 9(3), 369-391.

6. Camacho, L. (2016). Análisis global sobre posibles cambios paramétricos

del régimen de reparto administrado por el Banco de Previsión Social.

Banco de Previsión Social. Retrieved from: https://www.bps.gub.uy/bps/

file/11120/1/52.-analisis-global-sobre-posibles-cambios-parametricos-

del-regimen-de-reparto.-bps.-lcamacho.pdf

7. Castañeda, P., Castro, R., Céspedes, L., & Fajnzylber, E. (2013). Estudio

sobre la Sustentabilidad del Fondo de Reserva de Pensiones. Dirección

de Presupuesto, Gobierno de Chile. Santiago de Chile.

8. Dassatti, C., & Mariño, N. (2014). Maduración del sistema previsional:

proyecciones y agenda (Documento de Trabajo Nº 003-2014, ISSN

1668-7565). Banco Central del Uruguay. Retrieved from: https://www.

bcu.gub.uy/Comunicaciones/Jornadas%20de%20Economa/t_dassatti_

cecilia_2014.pdf

9. Durán Valverde, F., & Pena, H. (2011). Determinantes de las tasas de

reemplazo de pensiones de capitalización individual: escenarios latino-

americanos comparados. Comisión Económica para América Latina y el

Caribe (CEPAL). División de Desarrollo Social. Serie Nº 64. Santiago de

Chile. Retrieved from: https://www.cepal.org/es/publicaciones/7020-de-

terminantes-tasas-reemplazo-pensiones-capitalizacion-individual-es-

cenarios

10. Forteza, A., & Ourens, G. (2012). Redistribution, insurance and incen-

tives to work in Latin American pension programs. Journal of Pension

Economics and Finance, 11(03), 337-364.

11. Forteza, A., & Sanroman, G. (2015). Social security and retirement in

Uruguay. Journal of Development Studies, 51(4), 386-406.

Gioia de Melo

et al.

141

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

12. Forteza, A., & Rossi, I. (2016). La jubilación de los “cincuentones” (Docu-

mento de Trabajo). Departamento de Economía de la Facultad de Ciencias

Sociales (DECON). Universidad de la República. Retrieved from: http://

cienciassociales.edu.uy/departamentodeeconomia/wp-content/uploads/

sites/2/2016/07/0316.pdf

13. Güenaga, M., Mourelle, J., & Vicente, L. (2013). Estimaciones alterna-

tivas de producto potencial y brecha de producto en Uruguay. La función

de producción versus filtros univariados. Banco Central del Uruguay.

Retrieved from: http://www.bcu.gub.uy/Comunicaciones/Jornadas%20

de%20Economa/t_mourelle_jos%C3%A9_2013_1527.pdf

14. Lavalleja, M., Rossi, I., & Tenenbaum, V. (2018). La flexibilización de

las condiciones de acceso a las jubilaciones en el Uruguay. Comisión

Económica para América Latina y el Caribe (CEPAL). Serie Estudios

y Perspectivas. Oficina de la Cepal en Montevideo. Retrieved from:

https://www.cepal.org/es/publicaciones/43250-la-flexibilizacion-condi-

ciones-acceso-jubilaciones-uruguay

15. OECD. (2013). Pensions at a Glance 2013, OECD and G20 Indicators. OECD

Publishing. doi: 10.1787/19991363.

16. Paredes, R., & Díaz Fuchs, D. (2013). Pensiones y tasas de reemplazo

generadas por el sistema de AFP en Chile. Retrieved from: https://www.

aafp.cl/wp-content/uploads/2016/05/Tasas-de-Reemplazo-Pensiones-

de-Vejez_2012-Informe-Final.pdf

Eligibility for retirement and replacement rates in the Uruguayan

142

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Appendix

Table A.1 Contribution model (probability to make monthly contributions). Model

coefficients for the observed period 1996-2015

Female -0.601***

(0.034)

Accumulated density 3.832***

(0.006)

45 to 49 Cohort 0.049**

(0.021)

50 to 54 Cohort 0.093***

(0.022)

55 to 60 Cohort 0.189***

(0.023)

Public 0.789***

(0.026)

Rural 0.116***

(0.024)

Construction -0.160***

(0.032)

Housekeeping Services. 0.074**

(0.032)

Age 0.017***

(0.001)

Age^2 0.000***

(0.000)

GDP 0.010***

(0.000)

Informality rate 0.003***

(1.080)

Female participation rate 0.013***

(0.001)

Female participation rate*Female 0.013***

(0.001)

Constant -3.876***

(0.053)

Observations 3,352,852

Numbers of individuals 14,207

Standard errors in parenthesis *** p < 0.01, ** p < 0.05, * p < 0.1*

Gioia de Melo

et al.

143

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Table A.2. Income model. Estimated coefficients for each quintile

Quintile1 Quintile2 Quintile3 Quintile4 Quintile5

Female 1,238** -6.255 47.16 -1,596 -5,903**

(575.1) (723.0) (809.4) (990.7) (2,456)

Last year earnings 0.630*** 0.643*** 0.694*** 0.759*** 0.934***

(0.00569) (0.00503) (0.00409) (0.00329) (0.00205)

45 to 49 Cohort -1,605** -1,688* 284.9 -2,344** 18,683***

(713.0) (875.2) (952.0) (1,186) (3,556)

50 to 54 Cohort -1,348* -1,156 -627.6 -3,764*** 37,055***

(726.2) (935.3) (1,062) (1,345) (4,031)

55 to 60 Cohort -1,015 -202.9 805.0 377.3 61,945***

(757.8) (1,003) (1,153) (1,571) (4,826)

Agro, Forest, Fishing, and Mines -7,448*** -9,446*** -7,742*** -14,248*** -2,655

(1,469) (1,654) (1,785) (1,867) (7,187)

Manufacturing Ind. 909.1 -5,616*** -6,093*** -9,697*** -3,380

(1,718) (1,775) (1,815) (1,509) (3,771)

Electricity, gas and water -1,649 -5,743** -8,717** -13,642* 16,734***

(2,082) (2,419) (3,540) (7,645) (5,643)

Construction -1,687 -11,239*** -8,459*** -12,495*** -32,009***

(1,651) (1,913) (1,989) (1,821) (7,078)

Commerce, Restaurants, and

Hotels -326.5 -5,806*** -7,212*** -12,107*** -462.6

(1,541) (1,623) (1,695) (1,419) (4,135)

Transport and Communication. 1,889 -2,089 -3,517* -10,196*** -4,146

(2,056) (2,111) (2,067) (1,740) (3,945)

Financial intermediation,

insurance. -5,468*** -3,888* -6,124*** -14,306*** -6,265

(1,909) (2,127) (2,272) (2,194) (5,602)

Administrative Act. 1,227 -2,448 -6,633*** -10,585*** 15,875

(2,081) (2,245) (2,484) (3,277) (11,123)

Education -135.7 -1,106 -3,187 -7,683*** 4,027

(2,594) (2,693) (2,664) (2,367) (5,143)

(Continued)

Eligibility for retirement and replacement rates in the Uruguayan

144

DESARRO. SOC. 83, BOGOTÁ, SEGUNDO SEMESTRE DE 2019, PP. 105-144, ISSN 0120-3584, E-ISSN 1900-7760, DOI: 10.13043/DYS.83.3

Table A.2. Income model. Estimated coefficients for each quintile

Quintile1 Quintile2 Quintile3 Quintile4 Quintile5

Health 1,217 1,707 994.2 1,024 15,370***

(2,171) (2,493) (2,438) (2,043) (3,738)

Arts and others -175.1 -8,846*** -5,003** -7,724*** -15,680**

(1,906) (2,156) (2,343) (2,420) (7,727)

Home Serv. -3,068** -6,935*** -9,797*** -21,762*** -

(1,492) (1,760) (2,420) (5,470)

Extraterritorial Org. - - - -59,879*** -70,489

(22,143) (58,349)

GDP 278.5*** 461.9*** 677.2*** 993.0*** 1,554***

(9.245) (12.34) (14.46) (19.56) (52.79)

Experience 577.4*** 755.3*** 832.6*** 926.5*** -2,642***

(94.54) (136.9) (152.3) (191.4) (471.4)

Experience^2 -23.99*** -25.62*** -30.81*** -41.10*** -46.36***

(3.070) (4.759) (4.879) (5.594) (12.11)

Constant -20,640*** -25,418*** -41,298*** -55,582*** -91,735***

(1,946) (2,312) (2,583) (2,872) (7,783)

The cohort of 40 to 44 and Public Administration are omitted. Age is calculated as of January 2017.

Standard errors in parenthesis *** p < 0.01, ** p < 0.05, * p < 0.1

Source: Authors’ own calculations based on social security records.

Table A.3. Macroeconomic Scenarios (%)

Macroeconomic Scenarios

Variables Moderate Base Optimistic

GDP growth 1. 5 3.0 4.5

Real Wage growth 1.0 2.0 3.0

Female participation rate in 2046a56 59 64

Informality rate in 2046a30 21 15

FAP return in Unidades Indexadasb3.5 3.5 3.5

a According to calculations from the Household Survey, the participation rate of women in January 2016

was 57% and the percentage of individuals who did not contribute to social security was 23.8% in the

same period.

b Unit of value that is adjusted with the evolution of consumer prices.

Source: Authors’ own calculations based on social security records.

Los suscriptores pueden acceder a la versión informada de este caso.

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Accede a más de 120 millones de documentos de más de 100 países, incluida la mayor colección de legislación, jurisprudencia, formularios y libros y revistas legales.
Miles de fuentes de datos

Actualizado cada día, vLex reúne contenido de más de 750 proveedores dando acceso a más de 2500 fuentes legales y de noticias de los proveedores líderes del sector.
Encuentra rápidamente lo que necesitas

Gracias a una avanzada Inteligencia Artificial desarrollada por vLex, enriquecemos editorialmente la información legal para hacerla accesible, incluyendo traducción instantánea a 14 idiomas para garantizar el acceso a la información y la capacidad de efectuar búsquedas comparativas.
Más de 2 millones de usuarios registrados

Fundado hace más de 20 años, vLex proporciona contenido de alta calidad y un servicio muy intuitivo para abogados, despachos, instituciones gubernamentales y universidades de derecho alrededor del mundo

Los suscriptores pueden ver una lista de toda la legislación y jurisprudencia citada de un documento.

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Accede a más de 120 millones de documentos de más de 100 países, incluida la mayor colección de legislación, jurisprudencia, formularios y libros y revistas legales.
Miles de fuentes de datos

Actualizado cada día, vLex reúne contenido de más de 750 proveedores dando acceso a más de 2500 fuentes legales y de noticias de los proveedores líderes del sector.
Encuentra rápidamente lo que necesitas

Gracias a una avanzada Inteligencia Artificial desarrollada por vLex, enriquecemos editorialmente la información legal para hacerla accesible, incluyendo traducción instantánea a 14 idiomas para garantizar el acceso a la información y la capacidad de efectuar búsquedas comparativas.
Más de 2 millones de usuarios registrados

Fundado hace más de 20 años, vLex proporciona contenido de alta calidad y un servicio muy intuitivo para abogados, despachos, instituciones gubernamentales y universidades de derecho alrededor del mundo

Los suscriptores pueden ver una lista de todos los documentos que citan el caso

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Accede a más de 120 millones de documentos de más de 100 países, incluida la mayor colección de legislación, jurisprudencia, formularios y libros y revistas legales.
Miles de fuentes de datos

Actualizado cada día, vLex reúne contenido de más de 750 proveedores dando acceso a más de 2500 fuentes legales y de noticias de los proveedores líderes del sector.
Encuentra rápidamente lo que necesitas

Gracias a una avanzada Inteligencia Artificial desarrollada por vLex, enriquecemos editorialmente la información legal para hacerla accesible, incluyendo traducción instantánea a 14 idiomas para garantizar el acceso a la información y la capacidad de efectuar búsquedas comparativas.
Más de 2 millones de usuarios registrados

Fundado hace más de 20 años, vLex proporciona contenido de alta calidad y un servicio muy intuitivo para abogados, despachos, instituciones gubernamentales y universidades de derecho alrededor del mundo

Los suscriptores pueden ver una lista de todas los versiones de la ley con las distintas afectaciones

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Accede a más de 120 millones de documentos de más de 100 países, incluida la mayor colección de legislación, jurisprudencia, formularios y libros y revistas legales.
Miles de fuentes de datos

Actualizado cada día, vLex reúne contenido de más de 750 proveedores dando acceso a más de 2500 fuentes legales y de noticias de los proveedores líderes del sector.
Encuentra rápidamente lo que necesitas

Gracias a una avanzada Inteligencia Artificial desarrollada por vLex, enriquecemos editorialmente la información legal para hacerla accesible, incluyendo traducción instantánea a 14 idiomas para garantizar el acceso a la información y la capacidad de efectuar búsquedas comparativas.
Más de 2 millones de usuarios registrados

Fundado hace más de 20 años, vLex proporciona contenido de alta calidad y un servicio muy intuitivo para abogados, despachos, instituciones gubernamentales y universidades de derecho alrededor del mundo

Los suscriptores pueden ver todas las afectaciones de un caso

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Accede a más de 120 millones de documentos de más de 100 países, incluida la mayor colección de legislación, jurisprudencia, formularios y libros y revistas legales.
Miles de fuentes de datos

Actualizado cada día, vLex reúne contenido de más de 750 proveedores dando acceso a más de 2500 fuentes legales y de noticias de los proveedores líderes del sector.
Encuentra rápidamente lo que necesitas

Gracias a una avanzada Inteligencia Artificial desarrollada por vLex, enriquecemos editorialmente la información legal para hacerla accesible, incluyendo traducción instantánea a 14 idiomas para garantizar el acceso a la información y la capacidad de efectuar búsquedas comparativas.
Más de 2 millones de usuarios registrados

Fundado hace más de 20 años, vLex proporciona contenido de alta calidad y un servicio muy intuitivo para abogados, despachos, instituciones gubernamentales y universidades de derecho alrededor del mundo

Los suscriptores pueden acceder a una representación visual de un caso y sus relaciones con otros casos. Como alternativa a las listas de casos, el Mapa de Precedentes facilita la tarea de encontrar que caso tienes más relevancia en tu búsqueda. Al mismo tiempo también tendrás una referéncia del grado de aceptación del caso.

Request your trial

¿Por qué debería darme de alta en vLex?

Más de 100 países

Accede a más de 120 millones de documentos de más de 100 países, incluida la mayor colección de legislación, jurisprudencia, formularios y libros y revistas legales.
Miles de fuentes de datos

Actualizado cada día, vLex reúne contenido de más de 750 proveedores dando acceso a más de 2500 fuentes legales y de noticias de los proveedores líderes del sector.
Encuentra rápidamente lo que necesitas

Gracias a una avanzada Inteligencia Artificial desarrollada por vLex, enriquecemos editorialmente la información legal para hacerla accesible, incluyendo traducción instantánea a 14 idiomas para garantizar el acceso a la información y la capacidad de efectuar búsquedas comparativas.
Más de 2 millones de usuarios registrados

Fundado hace más de 20 años, vLex proporciona contenido de alta calidad y un servicio muy intuitivo para abogados, despachos, instituciones gubernamentales y universidades de derecho alrededor del mundo

Los suscriptores pueden ver una lista de resultados conectados a su documentos vía tópicos y citas encontradas por Vincent.

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Accede a más de 120 millones de documentos de más de 100 países, incluida la mayor colección de legislación, jurisprudencia, formularios y libros y revistas legales.
Miles de fuentes de datos

Actualizado cada día, vLex reúne contenido de más de 750 proveedores dando acceso a más de 2500 fuentes legales y de noticias de los proveedores líderes del sector.
Encuentra rápidamente lo que necesitas

Gracias a una avanzada Inteligencia Artificial desarrollada por vLex, enriquecemos editorialmente la información legal para hacerla accesible, incluyendo traducción instantánea a 14 idiomas para garantizar el acceso a la información y la capacidad de efectuar búsquedas comparativas.
Más de 2 millones de usuarios registrados

Fundado hace más de 20 años, vLex proporciona contenido de alta calidad y un servicio muy intuitivo para abogados, despachos, instituciones gubernamentales y universidades de derecho alrededor del mundo

Eligibility for retirement and replacement rates in the Uruguayan multi-pillar pension system

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Miles de fuentes de datos

Encuentra rápidamente lo que necesitas

Más de 2 millones de usuarios registrados

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Miles de fuentes de datos

Encuentra rápidamente lo que necesitas

Más de 2 millones de usuarios registrados

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Miles de fuentes de datos

Encuentra rápidamente lo que necesitas

Más de 2 millones de usuarios registrados

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Miles de fuentes de datos

Encuentra rápidamente lo que necesitas

Más de 2 millones de usuarios registrados

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Miles de fuentes de datos

Encuentra rápidamente lo que necesitas

Más de 2 millones de usuarios registrados

¿Por qué debería darme de alta en vLex?

Más de 100 países

Miles de fuentes de datos

Encuentra rápidamente lo que necesitas

Más de 2 millones de usuarios registrados

Regístrate para una prueba y aprovechar al máximo nuestro servicio, incluidos estos beneficios.

¿Por qué debería darme de alta en vLex?

Más de 100 países

Miles de fuentes de datos

Encuentra rápidamente lo que necesitas

Más de 2 millones de usuarios registrados