1. Taking stock of eight years of
implementation of Seguro Popular in
Mexico:
Lessons for developing countries
Mariano Bosch, Belen Cobacho and Carmen Pagés

2. This paper
 Study the health and labor market effects of Seguro Popular, a
large program intended to provide free health care to the
uninsured population in Mexico.
 The analysis of this program is interesting because:
• Its size: Intention is to reach more than 50 million individuals
• Its representative of a wider trend: towards the expansion of non-
contributive programs in developing economies.
 We focus on this program with the aim of extracting lessons for
other developing economies.

3. The Context
 Social insurance systems have been unable to reach wide
coverage in developing economies:
– Design issues: in most cases exclude the self-employed, a large share
of employment in developing co.
– Evasion: low compliance with health and pensions contributions.
 Inability to reform SS has led to a trend towards an
unprecedented expansion of non-contributive programs side
by side SS
– Health
– Pensions
– And more.

4. Health
 Mexico: Seguro Popular
 Colombia: Subsidized Regime

5. Non contributory Pensions
0
5
10
15
20
25
1970 1975 1980 1985 1990 1995 2000 2005 2010
Year
Universal pensions
Complementary pensions
Targeted pensions
Countries with any non-
contributory pension

6. Parallel non contributory regimes
+
 Rapid expansion of
coverage in economies
where formalization is
slow.
-
 The combo SS/non
contributive program
reinforces incentives to
be in informal
employment or OLF.
 Poverty traps
 Resource misallocation
 Output loss

7. Large flows
This is particularly the case in economies with large
formal to informal flows:
 Mexico
– Large transitions between formal and informal workers:
– 1 out of 4 formal workers will be in informal within a year.

8. This paper
 We review the empirical evidence on the consequences of the
introduction of Seguro Popular (SP)
 We find:
– SP has been a very effective tool in reducing catastrophic health
expenditures for Mexican families.
– No health effects yet.
– An increase in informal employment of 0.4 - 1 % points of total
employment, or equivalently between 160.000 and 400.000
workers. Some incentives to transit to inactivity.

9. Outline
 Description of Seguro Popular
 Effects on health services provision and
health outcomes
 Effects on the labor market
 Discussion

10. Mexican Health Institutions
 As in many other Latin American countries, access to health
care in Mexico is intimately linked to formal work status.
 Mexico’s health care system was born in 1943. Two institutions
were created for formal sector workers.
– IMSS for private workers
– ISSSTE for public workers
 In parallel to IMSS and ISSSTE, the Secretaría de Salud y
Asistencia (SSA) was created to serve all individuals outside
the formal sector.
 Discussion

11. Mexican Health Institutions
By 2000:
 Nearly 50 % of the Mexican population, amounting to
47 million people, was not insured through either
IMSS or ISSSTE.
 The SSA estimated that 2 - 4 million families (10 - 20
% of the total population) suffered catastrophic and
impoverishing health care expenses every year.

12. The Reform
 In 2003 reform in Mexico establishing the System for Social Protection
in Health (Sistema de Protección Social en Salud, SSPH)
– SP was functioning as a pilot since late 2001, although expenditure in non-
contributory Health was increasing even before SP was born.
 One of the objectives of the reform was creating and guaranteeing an
explicit basic universal package of services to uninsured households
– Only households not covered by Social Security health services are eligible for
the program .
 Another aim of the reform was to promote a “culture of co-payment”
– The poorest 20% of the households were declared exempt from all payments
– By 2010 the poorest 40% of the households were exempt from any payment, and
families up to the seventh decile with at least 1 child younger than 4 years of age
were also exempt.

13. Implementation of SP
 The SP was implemented in stages across states and
municipalities.
– The program had begun with a pilot phase in five states in 2002
 The rules of operation stated that the program had to
implemented in localities with high poverty and/or localities
with indigenous populations, but the localities also needed to
have health facilities in close range.
 The expansion of the SP was very fast. By 2010, the SP had
more than 43 million affiliates (Secretaría de Salud, 2010).
– Although this number could be an overestimate. According to the 2010
Census there are only 26.2 million

14. Implementation of SP

15. Implementation of SP
 A crucial issue in the impact evaluations of SP is how this roll
out across municipalities was undertaken
– In principle poorest municipalities should receive it first
– Barros (2009) shows how political and logistical concerns seem to have
driven the implementation sequence of the SP program across states in
Mexico
– Díaz-Cayeros et al. (2006) argue that municipalities in smaller states
were given preference to achieve full coverage of the SP.
– Bosch and Campos (2010) show that bigger municipalities implemented
SP earlier, but variables related to the income, number of uninsured,
and industrial structure were not significant in predicting the timing of
affiliation of a municipality to SP
– Aterido et al. (2011) show that the growth rate of informality prior to the
introduction of SP does not predict the arrival of SP either
 Most studies rely on the rollout of the SP.

16. Outline
 Description of the program
 Effects on health services provision and
health outcomes
 Effects on the labor market
 Discussion

17. Effects on Health
 A series of articles where published in The Lancet , 2006
suggesting the program had been a success
– Frenk et al. (2006): The reform allowed a substantial increase in public
investment in health, and it realigned incentives towards a better quality
service

18. Effects on Health
 The two most systematic studies up to date on the effects of
the SP on health outcomes.
– King et al. (2009), using an experimental approach
– Barros (2009), using triple differences in differences
 Both find that
– The SP significantly reduced health catastrophic
expenditures of Mexican households.
– No significant effects on health outcomes, at least so
far.

19. King et al (2007)
 King et al. (2007, 2009) introduce an experimental component to
evaluate the program.
 During the early stages of the roll out of SP, King et al. (2007)
were able to randomly assign encouragements and extra health
expenditures to particular health clusters.
 The find:
– A 23% reduction from baseline in catastrophic health
expenditures.
– Contrary to expectations, they found no effects on medication
spending, health outcomes, or utilization.

20. Barros (2009)
 Barros (2009) estimates the impact of the SP program by analyzing
differences over three dimensions.
– Program intensity target, measured as the ratio of the total number of
households agreed (by the federal and state governments)
– Across time
– SP eligibility (IMSS or not)
 He finds
– SP increased non-health expenditures or savings by 4.2 percentage points,
and encouraged beneficiaries to seek care.
– While at baseline SP-eligible households were 10.4 percentage points less likely
than Social Security covered households to seek care due to financial concerns,
the effect of SP has been to close this gap by about 40%.
– However the effect on health status is negligible.

21. Health Effects
 In sum
– Increase in health expenditures for the uninsured
– The SP significantly reduced health catastrophic
expenditures of Mexican households.
– No significant effects on health outcomes, at least so
far

22. Outline
 Description of the program
 Effects on health services provision and
health outcomes
 Effects on the labor market
 Discussion

23. Effects on the Labor Market
 A pre-requisite to join the SP was that the worker was not
covered by SS.
 Concern that workers and firms could opt out of SS and join
SP.
 This issue is of particular relevance in Mexico because
evidence of a large degree of mobility between formal (the
insured) and the informal (the uninsured).
– Around 25% of formal workers are found in the informal sector or in
non-employment a year later.

24. Three main concerns of the shift to Informality
 Fiscal concerns
 Reduction in coverage in other SS benefits (bundling)
– If workers or firms (or both) choose to avoid social security contributions
because SP provides affordable health care, it triggers the loss of all
other benefits as well
 “informalization” of firms and workers may lead to resource
misallocation & productivity losses.
– Reduction in the scale of production.
– Too many resources allocated to low scale unproductive firms. (subsidy
for health care for informal firms)

25. Effects on the Labor Market
Very active field with diverse results.
 Early studies: No effect
– (Gallardo-García, 2006; Esquivel and Ordaz, 2008; Barros, 2009;
Campos and Knox, 2010; Aguilera, 2011).
 With data covering up to 2010 some studies find small negative
and non-significant effects on the share of informality. (Azuara
and Marinescu, 2010; Duval and Smith, 2011).
 Other find significant adverse effects
– (Aterido et al., 2010; Bosch and Campos, 2010; Bosch and Cobacho,
2011; Pérez-Estrada, 2011).

26. Effects on the labor market
 Almost all studies use the roll-out of the SP
 Different datasets
– ENIGH (Income and Expenditure Survey)
– ENE-ENOE (Household Survey)
– IMSS data
– ENAMIN (Micro Firm Sample)
 Different municipality coverage
– ENE-ENOE: around 400
– IMSS: 1300
– ENAMIN: 70

27. Studies using the ENE-ENOE
 Azuara and Marinescu (2010),
– 1995 -2010
– They find no significant effect of the SP when they analyze the whole employed
population (0.8% points)
 Aterido et al. (2010)
– The exploit the panel dimension—better control for unobservables.
– SP has generated adverse effects on SS affiliation of around 0.4-0.7 percentage
points of total employment.
– Effects on labor flows: 3.1 percentage point reduction of flow into formality. Little
movement from formality to salaried informality, but some increase to self-
employment and inactivity.
– Reduction of wages of informal workers
 Pérez-Estrada (2011)
– Confirms earlier studies that the SP decreased the share of formal employment
by around 1 % point coupled with a 15% decrease of relative informal wages

28. Some groups are more sensititve
Some groups are more sensititve
 Azuara and Marinescu (2010)
 Significant decrease (1 % point ) in the share of formal employment for less educated
workers (with less than 9 years of schooling).
 Aterido, Hallward and Pagés (2011)
– Second earners more sensitive to SP than primary earners (heads of household)
– Households were head has lower than secondary education, particularly if head is a woman.
– Large households.
 Bosch and Campos (2010)
– The effects are much stronger in small firms.

29. Other datasets
 Parker and Scott (2008)
– Mexican Family Life Survey (MxFLS) 2002–2005,
– Find a disincentive effect in rural municipalities: beneficiaries of the SP 13 to 15
percentage points less probable to be formal in rural areas, and about 7
percentage points in urban areas.
 Bosch and Campos-Vazquez (2010)
– IMSS data
– After three years of the initial implementation of the SP, the level of registration is
4% lower than it should have been for both employers and employees in firms of
less than 50 employees.
 Bosch and Cobacho (2011)
– ENAMIN
– On average, a 10% increase in the time of exposure to the SP decreases the
share of formal employment of young workers by 1.1 and 2.1 percentage points
relative to workers between 25-35 and older than 35, respectively.

30. Other datasets
 Aguilera (2011)
– Employs the social security administrative database of the pension
system (Sistema de Ahorro para el Retiro, BDSAR),
– She then matches this database provided by the randomized
experiment implemented by King et al. (2007) described above at the
zip code level.
– The results suggest that SP has no impact in the short run on formal
employment or the probability of entering or leaving the formal sector.
– Very few municipalities

31. In all
 While studies analyzing the early years of SP did not find
significant effects of the program in the labor market…recent
studies do find effects on informality.
 However the extent of this effect is a matter of debate.
– Some studies suggest 0
– ENE-ENOE suggests around 0.4-1 percentage points. Equivalent to
160.000 to 400.000 thousand jobs. With different degrees of precision.
– IMSS data suggests around 300.000 jobs.

32. We try to ,
 We use, the ENE-ENOE, IMSS and Census data to shed some
light on the differences
 We estimate the following equation using diferent datasets
mmsm Xy   2000,ln(exp)

33. Estimates
Table 3: Effect of the SP on the growth rate of the number of formal workers 2000-2010
Census, IMSS and ENE-ENOE
(1) (2) (3) (4) (5) (6)
Log quarters -0.1280*** -0.1356*** -0.1581*** -0.1168* -0.1794*** -0.3222
of exposure (0.028) (0.025) (0.049) (0.061) (0.064) (0.203)
Dataset Census Census Census IMSS IMSS ENE-ENOE
Sample Census IMSS ENE-ENOE IMSS ENE-ENOE ENE-ENOE
Municicipalities 2,332 1,373 428 1,373 428 428
Controls
States Fixed Effects X X X X X X
Mun. Characteristics X X X X X X
*** Significant at 1%. ** Significant at 5%. * Significant at 10%.
Notes: The table shows the results of estimating equation (1) where the dependent variable is the growth
rate of the number of formal workers by municipality and the independent variable is the log of the number
of quarters the municipality has been exposed to the SP. “Dataset” indicates the source of data. “Sample”
indicates the restriction on the number of municipalities available to the Census, IMSS and ENE-ENOE
datasets. All regressions include state fixed effects. “Mun. Characteristics” refer to municipality
characteristics in the 2000 Census and include 16 industry shares, share of dependents, share of females,
years of schooling, log population, urban dummy, human development index, and share of informal
workers in 2000. Errors are clustered at the state level and regressions are weighted using municipality
population in the year 2000.

34. ENE-ENOE vs IMSS data-.2
0
.2.4.6
-.2 0 .2 .4 .6
Formal employment growth 2000-2009 (IMSS)
-4-2
024
Formalemploymentgrowth2000-2009(ENE-ENOE)
-4 -2 0 2 4
Formal employment growth 2000-2009 (IMSS)

35. Effects on the Labor Market
Table 4: Effect of the SP on the growth rate of the share of formal workers 2000-2010
Census, IMSS and ENE-ENOE
(1) (2) (3) (4)
Log quarters of exposure -0.0578** -0.0453** -0.0220 -0.1880
(0.022) (0.021) (0.033) (0.132)
Dataset Census Census Census ENE-ENOE
Sample Census IMSS ENE-ENOE ENE-ENOE
Municicipalities 2,332 1,373 428 428
Controls
States Fixed Effects X X X X
Municipality Characteristics X X X X
*** Significant at 1%. ** Significant at 5%. * Significant at 10%.
Notes: The table shows the results of estimating equation (1) where the dependent variable is the
growth rate of the share of formal workers by municipality and the independent variable is the log
of the number of quarters the municipality has been exposed to the SP. “Dataset” indicates the
source of data. “Sample” indicates the restriction on the number of municipalities available to the
Census, IMSS and ENE-ENOE datasets. All regressions include state fixed effects. “Municipality
characteristics” refer to municipality characteristics in the 2000 Census and include 16 industry
shares, share of dependents, share of females, years of schooling, log population, urban dummy,
human development index, and share of informal workers in 2000. Errors are clustered at the state
level and regressions are weighted using municipality population in the year 2000.

36. Outline
 Description of the program
 Effects on health services provision and health outcomes
 Effects on the labor market
 Discussion

37. What have we learned?
 SP has promoted a rapid increase of coverage and a reduction
in health related expenses.
 Difficulties of creating a culture of prepayment. Provides a note
of caution on systems based on copayment or matching
contributions.

38. What have we learned?
 SP has increased the cost of employing low skilled labor in
large formal firms and has reduced the relative cost of doing so
in small less productive ones.
– The available evidence so far points at an increase at the share of
informality between 0.4 and 1 percentage points which is equivalent to
160.000 to 400.000 workers
 SP has also increased transitions towards inactivity. Income
effects.

39. Is this effect large or small?
 The estimates in the literature are for the short to medium run
but estimates provided by this paper based on the intensity of
exposure suggest that they will increase overtime.
 Estimates based on the roll out of SP are likely to
underestimate the size of the effects, because non-contributory
expenditures were increasing even before SP started. We only
capture the effects of the resources labeled as SP.
 The SP is by no means the only social program in Mexico
– It only represents 25% of the expenditure on non-contributory programs.

40. What have we learned?
 The recent trends in the regions to entitle informal workers
(and their families) with a series of non-contributory benefits is
likely to further distort the labor market. Poverty traps for low
income workers?
– Camacho et al. (2009) show that the expansion of social programs in
the early nineties in Colombia inadvertently created incentives for
people to become informal.
 Colombia: Inequality of benefits based only on condition of
employment is not sustainable.
 While we favor universalization of coverage, this needs to be
achieved with more integrated programs. Reform of the social
security systems is overdue…