Does Expanding Health Insurance Beyond Formal-Sector Workers Encourage Informality? Measuring the Impact of Mexico s Seguro Popular

Size: px

Start display at page:

Download "Does Expanding Health Insurance Beyond Formal-Sector Workers Encourage Informality? Measuring the Impact of Mexico s Seguro Popular"

Florence Bishop
5 years ago
Views:

1 Does Expanding Health Insurance Beyond Formal-Sector Workers Encourage Informality? Measuring the Impact of Mexico s Seguro Popular Reyes Aterido (WB-DECMG) Mary Hallward-Driemeier (WB-FPDCE) Carmen Pagés (IDB) Washington, May 19, 2011.

2 Introduction Large informal sector in most developing economies Informal sector associated with poverty, low productivity and limited protection against old age poverty, health risks or unemployment. Social programs are often contributory, e.g. employees/employers pay taxes towards unemployment insurance, health insurance, pensions etc. Enforcement of labor taxes is unfeasible in informal sector. Increasing attempts to expand coverage of social programs beyond the formal sector by introducing alternative, non-contributory programs

3 Introduction (II) Programs are well intentioned but Now there two types of people: 1) Some pay for protection with their direct labor taxes, 2) others get protection for free. Which group you are in does not depend on income, but on the type of job. Does getting it for free encourage workers to shift to the informal sector? Informal sector employment of last resort/choice?

4 Introduction(III) Mounting evidence that informal sector is not necessarily sector of last resort for many workers. Maloney 2004: Very high mobility particularly in the unskilled market and in Mexico. Being self-employed is attractive for many unskilled workers (Madrigal and Pagés, 2009, IDB, 2009) Workers pay part of SS contributions in the form of lower wages. (Heckman and Pagés, 2004) Myopia and low pension literacy reduce incentives to pay for social security.

5 Outline Seguro popular Review of the literature Data Identification Strategy Endogeneity of SP expansion? Results

6 Seguro Popular Social Security (SS) covers about 40% of Labor Force. Self-employed not obliged to contribute. In 2001, government launched Seguro Popular to improve health care access to all Mexicans not covered by SS. Goal is to provide health insurance to everyone lacking SS coverage.

7 Seguro Popular (II) Co-financed by state and federal governments In principle, contributions are made according to income. In practice, no one pays (only 2% pay even when 65% of beneficiaries are non-poor). Condition to qualify: To not have SS affiliation. SP coverage: Extends to spouse/partner, children under 18 or up to 25 if studying or dependents. It also covers, children of relatives, parents, & grand parents if living in the same household. SS covers only children up to 25.

8 Relative benefits of SP and SS Covered SP SS Primary earner If not a formal sector employment If registered with IMSS and pay SS If self-employed contributions Spouse Yes Yes Children less than 25 Yes if less than 18; if students Yes is less than 18; if students Other dependents Parents, grandparents, relative s children if in same household and dependent Other benefits No Yes, pension, unemployment insurance No Costs None if income is in lower 2 declines (In practice, non-poor don t pay) SS contributions

9 millions millions Important uptake of the program. Overall numbers of those in SS still increased with introduction of SP Figure 1. Households and Individuals with Social Security Large increase in those affiliated with SP Figure 2. Number of Individuals and Households with Seguro Popular 2002q1 2003q1 2004q1 2005q1 2006q1 2007q1 2008q1 2009q1 date source: Labor Surveys individual with ss household ss covered 2002q3 2003q3 2004q3 2005q3 2006q3 2007q3 2008q3 date individuals with SP (millions) Source: Seguro Popular Registry for municipalities in Labor Surveys households with SP (millions)

10 0 share By 2009, SP had arrived in all municipalities Figure 3. Share of Municipalities with Seguro Popular 2002q1 2003q1 2004q1 2005q1 2006q1 2007q1 2008q1 2009q1 date 5 households with SP 2% population with SP 5% population with SP Source: Seguro Popular Registry for municipalities in Labor Surveys

11 Literature Review (I) There are a number of papers in the literature Some of them find effects of SP on informality. However, there is room to examine this issue again, addressing some of the shortcomings of the existing studies.

12 Literature Review (II): Little effect of SP Campos-Vazquez and Knox(08). Aggregate data from 33 cities ENE Also use individual data from 136 municipalities for They do not find significant effects on informality. Barros (08). Repeated cross sections of ENIGH (00,04,05,06) and measures SP according to state intensity target (stated objective in 2007). Finds no effect.

13 Literature Review (III): Some effect in DF Juarez (2008). Examines effect of non-contributory health program rolled out in the DF to examine effect on informality. Uses cross sectional data from ENE , Diff-in-diff comparing DF with neighborhood municipalities not covered by the program. Finds that women are more likely to be in informal sector after the program in the covered municipalities But: Does not use the longitudinal aspect of the panel it does not control for individual and/or household characteristics. Diff-in-diff over one single district, over one period. Effect could be picking up other effects.

14 Literature Review (IV): Some effects in aggregate data Bosch et al (2010). Use SS registry data aggregated at the municipality level Positive effect on informality at later stages of the program (07). Larger effects in small firms (relative to micro firms) But: Does not control for individual characteristics; We do not know whether households are abandoning formality, or new entrants are getting into informal jobs.

15 Literature Review (V): Evidence from Colombia Related programs in other countries: Camacho et al (2010). Effect of introduction of SR in Colombia. Study introduction of SR in Colombia. SR is partly financed with 1pp of payroll and is less comprehensive than CR. But unlike CR, SR covers every individual member of the household. 2-5% increase in informality post SR. Diff in diff. But: cross sectional analysis Colombia does not have good panel data. Thus, cannot follow people or families over time. Identification is not great. There is only one round difference between the introduction of program across cities. Therefore, much of the variance is before/after.

16 Data used in this paper ENE: ,quarterly ENOE: , quarterly We follow households over time, using the panel dimension of the data. We have about 100,000 hh per period. We treat the two panels as separate given that there are some discountinuities. We measure coverage of SP in each municipality and quarter according to the SP registry of beneficiaries data.

17 Summary statistics Table 1. Summary Statistics of the Main Variables Variable N Mean S.D. Min Max Households Covered by Social Security 2,926, Number of children 0-5 2,931, Number of children ,931, Number of adults 66 or older 2,931, large household (more than 4) 2,931, Household heads M ale 2,924, Age 2,924, Education: secondary or more 2,924, Individuals Social Security 8,955, Wage-employed with social security 8,966, Self-employed 8,966, Wage-employed without social security 8,966, Other informal employment 8,955, Not in the labor market 8,966, Employees Working in small firm 1-5 employees 2 2,520, Working in medium firm 6-50 employees 2 2,520, Working in large firm more than 50 employees 2 2,520, Source: Author's calculations based on Mexico Labor Survey 2000q2-2009q2.

18 Identification Strategy It relies on the variation of the SP rollout across municipalities and time. We are comparing how the probability of formality (SS) of households changes as SP becomes available, relative to the behavior in municipalities where it is not yet available, or less prevalent.

19 Identification Strategy We estimate: P(SS ijt ) = a i + λz t + βsp jt +ψx ijt + ε ijt, where E[ε ijt \ jt ]=0, i, identifies the household j, identifies the municipality t, time period a i, household fixed effects Z t, is a vector of time dependent variables: X ijt, is a vector of household variables that change over time. SP jt, tracks the evolution of SP by municipality and period. O,1 if introduced logsp, number of affiliates by SP registry Errors clustered at the household level (Bertrand et al, 2004)

20 Little evidence that SP was rolled out in municipalities with higher or growing informality Table 2. Rollout of Seguro Popular (1) (2) (3) (4) (5) (6) Sample Municipality and period Dependent Variable SP SP SP SP SP SP share ss coverage (1yr lag) [ ] share ss (1yr lag) [ ] share ss coverage (2yr lag) [ ] share ss (2yr lag) [ ] growth share ss coverage (2yrs) [ ] growth share ss (2yrs) [ ] period fe yes yes yes yes yes yes municipality fe yes yes yes yes yes yes Observations R-squared

21 SP lowers formality on average; effect only declines at relatively high levels of intensity Table 4. Impact of Seguro Popular on Formality 1 (1) (2) (3) (4) (5) (6) SP [ ]*** [ ]*** [ ]*** logsp [ ]*** [ ]*** [ ]*** logspsq [ ]*** [ ]*** [ ]*** male [ ]*** [ ]*** [ ]*** [ ]*** [ ]*** [ ]*** less than secondary [ ]*** [ ]*** [ ]*** [ ]*** [ ]*** [ ]*** secondary [ ]*** [ ]*** [ ]*** [ ]*** [ ]*** [ ]*** terciary [ ]*** [ ]*** [ ]*** [ ]*** [ ]*** [ ]*** Panel control Yes Yes Yes Yes Yes Yes Log GDP; inflation Yes Yes No Yes Yes No State trends No Yes No No Yes No Quarterly time dummies No No Yes No No Yes Observations R-squared Household level; Municipalities with more than 50 observations in sample Standard errors clustered at the municipality level in brackets,* significant at 10%; ** at 5%; *** at 1%

22 Impact remains controlling for unobserved heterogeneity; less effect of changes in growth of rollout Table 5. Impact of Seguro Popular on Formality 1 (fixed effects) (1) (2) (3) (5) (6) (7) SP [ ]*** [ ]* [ ] logsp [ ] [ ] [ ] logspsq [ ]* [ ] [ ] Log GDP; inflation Yes Yes No Yes Yes No State trends No Yes No No Yes No Quarterly time dummies No No Yes No No Yes Observations Number of id_i R-squared Household level; Municipalities with more than 50 observations in sample Standard errors clustered at the municipality level in brackets,* significant at 10%; ** at 5%; *** at 1%

23 2.5 Impact of Seguro Popular on Formality Over Time : Regression Coefficients and 10% Confidence Intervals (cross Section)

24 Effects larger for less educated households and larger households Table 9: Effects by type of household according to characetristics household head 1 SP [se] Obs. R-sq. (1) HH head less than secondary [ ]*** (2) HH head secondary or more [ ] (3) HH head less than secondary and male [ ]*** (4) HH head less than secondary and female [ ]*** (5) HH head secondary or more and male [ ] (6) HH head secondary or more and female [ ] (7) HH head less than 30 years old [ ] (8) HH head years old [ ]*** (9) HH head 50 or more years old [ ]** (10) Small household [ ] (11) Large household [ ]*** HH controls include: gender, age, education of head,number of children 0-5 years old, 6-13 years old, of adults 66 or older Controls include: panel control, municipality trend, quarterly log GDP, quarterly inflation Standard errors clustered at the municipality level in brackets * significant at 10%; ** significant at 5%; *** significant at 1%

25 Impact larger in larger firms Table 8: Effects by type of household according to firm size 1 logsp [se] logspsq [se] Obs. R-sq (1) firm <=5 employees [ ] [ ]** (2) firm 6-50 employees [ ]*** [ ]*** (3) firm more than 50 employees [ ]*** [ ]*** Individual level HH controls include: gender, age, education of head,number of children 0-5 years old, 6-13 years old, adults 66 or older Controls include: panel control, municipality trend, quarterly log GDP, quarterly inflation Standard errors clustered at the municipality level in brackets * significant at 10%; ** significant at 5%; *** significant at 1%

26 Effects larger for wage earners not in SS than self-employed Table 7: Impact of SP on Formality by Labor Market Ocupation 1 (1) (2) logsp [ ]*** [ ]*** logspsq [ ]*** [ ]*** selfempt [ ]*** logsp*selfempt [ ]*** logspsq*selfempt [ ]*** wgemp_nosst [ ]*** logsp*wgemp-nosst [ ] logspsq*wgemp-nosst [ ]*** panel control yes yes Control yes yes Observations R-squared Other categories (unemployed, unpaid, not in labor force are also less affected compared to wage employees not in SS)

27 Conclusions (I) Informality increases as SP is deployed. However the estimated effects may not appear large: of the order of 0.5 to 1.2 pp on average. Effects increase in larger municipalities and over time. Larger Households are more sensitive to switch, as are the less educated. Evidence that SP has little effect on selfemployed or those out of labor force. Larger effect is on discouraging wage employees not in SS to join SS.

28 Conclusions (II) Small deadweight loss? Lower bound of true effects we only follow individuals for 5 quarters Results suggest effects increase over time. Should be weighted against benefits of integral SS reforms. Lower distortions associated with labor payroll taxes Yet, likely increased distortion on VAT side.

Does Expanding Health Insurance beyond Formal-Sector Workers Encourage Informality?

Public Disclosure Authorized Policy Research Working Paper 5785 WPS5785 Public Disclosure Authorized Public Disclosure Authorized Does Expanding Health Insurance beyond Formal-Sector Workers Encourage