Social Protection in the Global Development Agenda:

Lessons from Latin America

Asian Development Bank,
Manila, Philippines
August, 2016.

Santiago Levy, IDB*

* Authors opinions do not necessarily coincide with those of the institution he is affiliated with.
DISCLAIMER: This presentation does not necessarily reflect the views of ADB or the Government concerned, and ADB and the Government cannot
be held liable for its contents.

Contents
1. Social protection
2. Social insurance
3. Social assistance
4. Concluding remarks

1. Social protection

Objectives

Instruments
Social insurance programs

Protection against risks

Illness
Longevity, death, disability, workaccidents
Output and employment shocks

Note: All households are subject to risks

Usually associated with workers status in the

labor market.
Usually financed with wage-based contributions
However, lots of institutional variation and
recent innovation
Social assistance or poverty programs (+ .)

Redistribution/poverty reduction

Equity and poverty reduction as

good in their own right
Equity and poverty reduction
because with imperfect credit
markets some forms of redistribution
increase efficiency

Usually targeted transfers based on income or

asset indicators
Transfers in-kind or in-cash; sometimes
conditioned on behavior
Almost always financed from general revenues

Note: only for poor/low income households

2. Social Insurance

Motivation
In most countries in Latin America, the provision of social insurance is not the same for
all workers, but depends on their status in the labor market.
This creates substantive issues for:
The population covered
The risks against which households are protected (illness, longevity, disability, unemployment,
and so on)
The fiscal sustainability of social insurance provision, and
The behavior of firms and workers in the labor market, with spillover effects on productivity
and growth.

Although there are relevant institutional variations across countries, there are also
some issues that, mutatis mutandis, are common to all.

Distinction between salaried and non-salaried labor

salaried

Workers

self-employed
non-salaried
other

Salaried: have a boss (firm) and are paid a salary; there is a relation of subordination;
workers receive orders from bosses (effort observed, coordination of tasks, and so on).
Unions can be formed and minimum wages may apply; regulations on dismissal.

Non-salaried: work on their own, or are associated with a firm but are not subordinated to
it; contracts to share risk or elicit effort; payments take the form of commissions, profit
sharing, and so on. No minimum wages (no wages!) and no unions with free dismissal (no
boss!).
7

Bismarcks inheritance: asymmetry in social insurance,

CSI and NCSI
Salaried workers: bundled benefits, usually health, work-risk, death and disability
insurance, retirement pensions, and protections against loss of employment

benefits paid from earmarked wage taxes, hence the (mis)label of contributory social
insurance, CSI
benefits may also include labor training (Colombia), housing (Mexico), child allowances
(Argentina)
regulations on firing bundled as part of social insurance, but instead of unemployment
insurance, mostly one-time severance payments at dismissal time.

Non-salaried workers: until recently, uncovered by social insurance. But since

1990s unbundled pension, health and related programs

benefits paid from general revenues, hence the (mis)label of non-contributory social
insurance, NCSI
benefits targeted to workers not covered by CSI (often regardless of whether they are
salaried or not)
no costs of firing or minimum wages.

Who is formal? Large transits across labor status

Probability of change of status in one year

In the context of the CSI-NCSI dichotomy, transits across labor status reduce the efficacy
of insurance, as the same worker is protected against some risks only during some periods.
Transits may also have large implications for retirement pensions.

Structure of labor contracts under the formal-informal dichotomy

Contrat

Legal salaried
Illegal salaried

Non-salaried

Expected cost to
firm

Expected benefit
to worker

Implicit tax or
subsidy

w f T CSI

w f CSI T CSI

(1 CSI )T CSI

wi (.) F

wi NCSI T NCSI

wi

wi NCSI T NCSI

Formal
employment, Lf

(.) F NCSI T NCSI

NCSI T NCSI

Informal
employment

TCSI = monetary cost of the bundle of contributory social insurance benefits (including contingent costs)
TNCSI = monetary value of all non-contributory social insurance programs
CSI T CSI = workers valuation of TCSI ;

(.)F

NCSI T NCSI = workers valuation of TNCSI.

= expected penalty for violating a salaried contract (usually increasing in firm size).

Note: informal employment has a legal and an illegal segment. But under perfect enforcement, (.) 1 ,
its illegal segment would disappear as long as F > TCSI.

Impact of social insurance in the labor market

Key point:
costs of CSI (including severance pay and transactions costs of compliance) must be
internalized in the contract between worker and firm.
costs of NCSI programs are paid from resources external to the firm-worker contract.

CSI

L f / T CSI 0

tax on formality

(Relatively well known.

Because of bundling, TCSI is a scalar)

Social Insurance

T NCSI

L f / T1NCSI ,......, L f / Tn NCSI 0

subsidy to
informality

(Less well known.

Because of unbundling, TNCSI is a vector)

The combined impact of CSI and NCSI programs on the formal/informal composition of
employment (and other dimensions like participation rates) varies from program to program
and from country to country, since many parameter values determine the outcome.

Evidence from CSI programs (i.e., L / T )

Reducing the tax on formality: 2012 tax reform in Colombia
CSI

Health Contributions
50

49.0

12.5

Share of workers contributing to pensions

Reduction in pay-roll tax

47.5

49

46.7

48
47

44.4

44.5

46

42.8
42.0

41.9

43

41.4

41.9

44

42.1

43.2

43.7

45

42

The tax reform in 2012 reduced the health contribution for formal employees from 12.5 to 4 percentage points
There was a significant increase in formal employment, indicating that Tf was taxing formal employment

Sep - nov 15

May -jul 15

Sep - nov 14

Ene - Mar 15

May - jul 14

Ene - Mar 14

May-Jul 13

Sep - Nov 13

Ene - Mar 13

Sep-Nov 12

May-Jul 12

Ene-Mar 12

Sep - Nov 11

May - Jul 11

Ene - Mar 11

Sep - Nov 10

May - Jul 10

Ene - Mar 10

Sep - Nov 09

May - Jul 09

Sep - Nov 08

May - Jul 08

Ene-Mar 08

Sep - Nov 07

May - Jul 07

Ene - Mar 07

40

Ene - Mar

41

NCSI

L
/

T
Evidence from NCSI programs (i.e., f
j

Effect of Seguro Popular on the affiliation of

employers of firms up to 50 employees to Mexican Social Security

.3

Growth rate Employers in IMSS

This difference
translates into
a fall in formal
employment of
about 6%.

.2

.1

It also provides
indirect evidence
of higher evasion
by firms.

Beginning of implementation of
Seguro Popular in first group of
eligible municipalities.

Implemented in 2002-2003

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

.0

Implemented in 2006-2007

Source: Bosch and Campos-Vzquez (2010).

Reviewing the various papers, Bosch and Pages (2012) find that from 2002 to 2010,
Seguro Popular reduced formal employment by between 160,000 to 400,000 jobs,
or between 8 and 20% of all formal jobs created during that period.

Evidence from subnational NCSI programs

Effect of the Federal District (Mexico City) Health Program for Women

The probability of formal

employment fell by 8%.

Guadalajara and Monterrey

Federal District

Source: Jurez (2008).

14

More evidence from NCSI programs

Argentina, effects on labor informality of extending
child allowances (AUH) to informal workers
(% of informal workers that become formal)

The extension of the Asignacion Universal por Hijo (child allowances) to informal workers
reduced substantially the rates at which these workers entered formal employment.
Source: Garganta & Gasparini (2015)

More evidence from NCSI programs

Uruguay, effects on labor informality of family
allowances (AFAM, Plan de Equidad)

Workers who
do not qualify

Workers who
qualify

In Uruguay, the AFAM reduced registered (formal) employment by

In Uruguay, the AFAM reduced registered (formal) employment by 7.5
percentage points, i.e., from 48 to 41.5 percent, over the period 20082012.
Source: Garganta & Gasparini (2015)

Other evidence
Econometric studies of pure NCSI programs:

Camacho (2012 ), Colombia, the Regimen Subsidiado (health) reduces formal employment by 4%.

Emerging literature on the effects of NCSI pensions:

Galiani and Gertler (2009), Mexico, the Adultos Mayores de 70 program reduces labor supply of those near
retirement.

Attanasio et al. (2011), Chile, the 2008 pension reform will reduce the proportion of men and women working
formally by 1.7 and 4.3%, respectively, with larger reductions for those aged 50 and older. Todd and Joulbert
(2011) estimate even larger reductions. (These two papers are simulation models.)

Methodological observation
Countries operate many NCSI programs in parallel (health, pensions, day care, child
allowances, and so on). Further, NCSI programs co-exist with CSI programs.

What matters is the combined effect of all CSI and all NCSI programs, i.e., the
combined effect of the tax on formal labor and the subsidy to informal labor.
Few papers look at the full impact of [CSI + NCSI], as it is almost impossible to do so
with standard econometric techniques that allow proper identification; that is,
we want to know:

but most papers (like the examples above) measure

individual program j:

dL f
d (T

CSI

NCSI

L f / T j CSI

or

L f / T j NCSI

vectors

Anton, Hernandez and Levy (2012) use a simulation model to capture the joint effects
of [CSI + NCSI] programs in Mexico, and find that they reduce formal employment by
26% and increase informal employment by 45% relative to the equilibrium with TCSI =
TNCSI = 0.

The formal-informal dichotomy and firm/worker behavior

Firms and workers respond to the asymmetry in the regulation of labor implicit in the CSINCSI dichotomy. These responses depend on:
for firms, differences in expected costs of a salaried vs. non-salaried contract, possibilities of substitution
between them, possibilities of evading regulations on salaried labor (enforcement and fines)
for workers, the expected utility of each contract (earnings, social benefits, other dimensions of the job).

Given all other taxes, credit regulations, and so on, these responses impact:

the type and size distribution of firms: w/wo salaried contracts, legal/illegal, family firm, one-person firm
the efficacy of social insurance (who is covered against what risks, and how often)
the fiscal costs of social insurance
investments in labor training, adoption of technology and innovation, rotation of workers, tenure, etc.

The informal sector (informality) consists of economic activity where regulations on CSI
are not observed, regardless of whether this is a legal or illegal act.

Two distortions in the formal-informal dichotomy: firm type and firm size
Costs of Labor Contracts in Mexico, 2008

In fact, the
gap is larger
than 24%.

w f [1 (1 )

1.1

salaried labor: CSI +NCSI

10%
(tax on formality)

1.05

Average cost of labor contract

Average cost of labor contract

Note: these
estimates
exclude the
contingent
costs
of firing.

CaCost of salaried
labor to firmsCSI

A
1
labor: USI
Cost ofsalaried
laborandtononallsalaried
without
the
formal informal dichotomy

24%

0.95

14%
(subsidy to informality)
0.9

wi

0.87
non salaried labor: CSI + NCSI

Informales
bajo SSNClabor
Cost of non-salaried
to family firms and oneperson firms

0.85

Source:
Antn, Hernndez and Levy (2012).

0.8

10

20

30

40

50

Number of workers

The upward sloping line captures the costs of the profit maximizing combination of legal and illegal
salaried contracts, given the fines in Mexicos laws, and a function (.) that reproduces the legal/illegal
20
composition of employment observed in the 2008 Economic Census.

Size and type distribution of firms and employment, 2013

Formal

Mixed

Informal

Total

2.0
0.9
1.0
0.3

3.5
1.1
0.9
0.2

86.2
2.4
1.2
0.2

91.6
4.5
3.1
0.7

4.2

5.7

90.1

100.0

1.4
1.6
5.2
13.6

2.5
2.1
4.7
13.3

35.7
3.3
5.5
10.2

39.6
7.9
15.3
37.2

21.8

22.4

55.8

100.0

Firms
[0-5]
[6-10]
[11-50]
[50+]
Total
Workers
[0-5]
[6-10]
[11-50]
[50+]
Total

Informal firms are less productive than formal ones

TFP in Mexican firms: informal vs. formal

Other evidence

Number of employees
0-5

6-10

11-50

50+

Fajnzylber, Maloney and Montes Rojas

(2009) with Mexican data, and Fajnzylber,
Maloney and Montes Rojas (2011) with
Brazilian data, both find that increases in
formality increase productivity.

Pages [IDB (2010)] scattered data from

other Latin American countries.

Jung and Tran (2012) simulate the effects

on NCSI pensions in Brazil, and find a
reallocation of resources from the informal
to the informal sector that reduces output
by 4%.

However, data for proper measurement is

very scarce in many LA countries.

-10

-20

-30

-40

-50

Informal firms with salaried

workers (ilegal)

-60

Informal family firms (legal)

-70

Comparisons of firm-level TFP for a universe of

3.6 million firms grouped at 6-digit level, 2008.
Source: Busso, Fazio and Levy (2012).

22

The policy problem

Stop promoting

Improve social welfare

informality

We need to increase
productivity and
wident the tax base!

We need to expand the

coverage of social insurance to
all households!
( .with whatever combination of
programs and sources of revenue)

The nature of this trade-off varies from country to country and program
to program, and while more research is needed, it is clear that a dual social
insurance architecture can hurt productivity and growth.
23

3. Social assistance

Objetive:
Redistribution

Objetive:
Insurance against risks
Formal
Informal

NonPoor

Poor

CCTs are a LA
innovation to
redistribute while
investing in human
capital.

This is a row issue,

Not a column issue.

25

Will CCTs break the intergenerational transmission of

poverty ?
increase human capital

Two conditions:
better income opportunities, especially more
productive jobs with higher real wages

These two conditions are related, but they are separate. Unless the poor earn
higher wages with their own efforts, CCTs run the risk of being a permanent
scheme to transfer income to the poor, rather than a temporary investment in their
human capital.
It is critical to highlight that CCTs are not:
job-creating programs,
programs to respond to temporary shocks in output
substitutes for social insurance programs.
26

Incentive compatibility between CCTs and [CSI + NCSI] programs

NCSI programs should target workers by labor status, not by income level.
Focusing NCSI programs on poor workers will create stronger incentives for them to
be informally employed vis--vis non-poor workers.
CCTs should target workers by income levels, not by labor status. Focusing CCTs on
being informally employed will make it more difficult for poor workers to be
formally employed.
KEY POINTS:
The formal-informal status of poor workers is endogenous to the incentive
structure implicit in CCT and [CSI+NCSI] programs.
Poor workers may end up locked in informal jobs in low productivity activities.
Young poor workers may be healthier and more educated than their parents as a
result of CCTs but they will not access better jobs.

Evidence from CCTs with benefits conditional on informal status

Effect of Ecuadors Bono de Desarrollo Humano (BDH) on Affiliation to CSI
(panel data, women 35 to 65 years of age)

Share of women enrolled in IESS

Base line, 2002,

all households without BDH

2008-2009, 6 years of BDH

with
without

TThreshold to qualify for BDH

Source: Bosch and Schady (2013).

After six years, the BDH reduced formal employment of working women at the threshold by 15%.

Other evidence

Econometric evidence of CCT programs that condition benefits on informal status:

Amarante et al. (2011), Uruguay, Plan de Atencion Nacional a la Emergencia Social
(PANES), program reduces formal employment among men by 7.5%.
Ministerio do Desenvolvimento Social e Combate a Fome, (2012), Brazil, the Bolsa
Familia, program reduces formal employment among beneficiaries.

4. Concluding remarks

Social insurance

The combination of CSI and NCSI is bad social policy (lowers the efficacy of insurance) and bad
economic policy (hurts productivity and the tax base).
Countries need to escape from the dilemmas created by the CSI-NCSI dichotomy. Broadly, they
need to transit towards unified regimes. There are strong equity and efficiency reasons for
universalism.
In matters of social insurance two principles should apply: (i) risks that are common to all forms of
employment should be financed from a revenue source that is independent of any form of
employment; (ii) risks that are specific to a given form of employment should be financed with a
revenue source that is specific to that form of employment.
This is a tall order, as it inevitably involves tackling fiscal issues (what revenue sources should
replace CSI contributions?) and challenging long-held paradigms (dont CSI contributions
redistribute income from firms to workers, or from capital to labor?).
Social assistance/poverty
Income transfer programs for the poor (like CCTs) should avoid conditioning on poor workers
status in the labor market.
In parallel, poor workers should be protected against risks through the same mechanisms as all
other workers (i.e., by the same social insurance programs).

General
When discussing social protection, it is essential to make a distinction between social
insurance and social assistance/poverty programs. We need a clear view of the
objectives of each program, and of the instruments used to pursue them.
Extension of coverage and improvements in quality have been long attracted the
attention of those concerned with social protection in developing countries.
LAs experience shows that it is also central to consider the incentives implicit in social
programs, associated with the rules for qualifying for benefits, and the revenue
sources used to finance benefits. Who qualifies for what? Who pays for what? How do
households, firms and workers react to those rules and differences in revenue sources?
Debate centers on architecture of social protection, not on individual programs. An
integral view that ensures incentive compatibility across all programs is essential.
We need to go beyond the usual impact evaluation of individual programs and develop
a view of how all the pieces of the puzzle fit together. Reforms of individual programs,
or creation of new ones, can be risky without a more systemic view.
These issues need urgent attention, as countries may be constructing Welfare States in
economies characterized by permanent informality, weak fiscal basis and low
productivity growth.

Thank you.