Selection meeting for PEP- AusAid Policy Impact Evaluation Research Initiative (PIERI)

Québec, Canada - June 2008

School attendance, labor suppy and formality

An impact evaluation of PANES

Verónica AmaranteRodrigo Arim

Gabriel Burdin

Gioia de Melo

Andrea Vigorito

Outline of the presentation:

I. Objectives of this research proposalII. Policy relevance

II.1 General description of PANESII.2 New interventions: reformulation of Asignaciones Familiares

III. Background information on the outcomes to evaluate:

III.1 Child labour and school attendanceIII.2 Households’ income and adult labor supplyIII.3 Informality

IV. MethodsIV.1 DataIV.2 Modelling and testing

V. Final remarks

I. Objectives of this research proposal

Undertake an impact evaluation of Ingreso ciudadano, a conditional cash

transfer component of PANES (Programa de Atención Nacional a la

Emergencia Social).

PANES was an ambitious anti-poverty plan implemented from April

2005 to December 2007 in Uruguay.

We will concentrate on some key outcomes:

- child labour and school attendance

- household income

- adult labour supply

- informality.

We aim at answering the following research questions:

-Did Ingreso Ciudadano increase child school attendance in children aged 12 to 17 years old?

-Did Ingreso Ciudadano reduce child labour (6-17)?

-What were the effects of the program on household income?

-Did Ingreso Ciudadano have an impact on adult labour supply? Did it have a differencial effect by sex and region?

-Has Ingreso Ciudadano had any impact on the probability of being a formal worker?

II. Policy relevance

II.1 General description of PANES

PANES was a temporary anti-poverty program composed by differentinterventions. The objectives of the plan were extremely ambitiousranging from poverty and indigency alleviation to improvements in socialparticipation and autonomy. Main component:

- Ingreso Ciudadano: conditional cash transfer. 75 USD per month,despite household size and conditional on school attendance forchildren aged 6 to 14 and health controls of women and children. OnDecember 2007, 74.500 households were receiving thistransfer(around 10% of total population)

Other components:

-Trabajo por Uruguay: workfare program. 6 month temporary employment program of six hours of work per day, including training activities. The household received two IC stipends a month. Inscription in the program was voluntary and selection was random. Around 16.000 persons were covered by this program.

- Rutas de Salida: program aimed at fostering the reinsertion of households in social life through different activities, including courses on civil and labor rights. One of the members of the household could attend these activities during a semester. The program was implemented in cooperation with NGOs and social organisations

- Food purchase card. Households with children aged 0 to 18 received a magnetic card that allowed them to purchase food and store items. The amount of the transfer was dependent on the number of children and pregnant women in the household, ranging 20 to 40 USD

- Target population: first quintile of households under the poverty line (around 8% of total population)

- Cost of the program was around 100 million USD per year.

- Eligibility: based on an income threshold (55 dollars per capita) and a proxy means test

- Information provided by households was checked with the social security records. If the household did not exceed the established threshold, it was visited by an interviewer

- Targeting: succesful (Banco Mundial, 2007). However, some non elegible households were treated

- Conditionalities were not really controlled, mainly due to coordination failures other institutional problems

II.2 New interventions: reformulation of Asignaciones

Familiares

The results of this evaluation are key inputs for the redesign of

Asignaciones Familiares, a permanent cash transfer that is being

Incepted at present

At the same time, the Uruguayan case can shed some new light on the

extent to which these programs can reduce poverty and indigence in

middle income countries

It can also illustrate their effects on other relevant outcomes, such as

secondary school attendance or informality in the labour market, that have

been scarcely addressed in previous research for the region

III. Background information on the outcomes to evaluate

III.1 Child labour and school attendance

Attendance rates for children aged 5 to 12 are close to 100% despite household income level (universal primary education since the earlier decades of the XX century)

Dropping out from secondary school is one of the more relevant problems in the Uruguayan educational system, particularly in low income strata

Although child work is not generalized even in deprived sectors, it turns to be a relevant alternative to school attendance. According to the latest available data (2006), 9.2% of children between 12 and 17 work (Arim y Salas, 2007). Employment rates are higher among children in extreme poverty (15%).

-Previous evaluations: success at increasing child school attendance, particularly at the primary school level (ECLAC, 2006; Coady, 2001; Coady and Parker, 2002; Skoufias and Parker, 2004), although compliance was difficult and expensive to control

- Child labour was reduced but not to the extent policy designers have originally expected, so schooling may partly have increased at the expense of a reduction of child leisure time (Ravallion and Wodon, 2000)

III.2 Households’ income and adult labour supply

-Previous evaluations: impacts in terms of poverty reduction have been meager due to the reduced amount of the transfers and their targeting on extreme poor households (ECLAC, 2006)

-Reduction in child labor could imply a reduction on the net impact on poverty

-Impacts have been higher on indigence and severity of poverty

-effects on adult labor market behavior: intensely debated, specially in industrialized countries. Some authors have argued that the relatively low amount of cash transfers in Latin America turn unlikely that these programs generate disincentives to work (Tabor, 2002).

-Substitution and income effects faced by beneficiary households. Feared outcome: substitution effect leads to a reduction in work effort, either in labour market participation or in hours of work

- In developed countries many programs have been reshaped in order to include conditions that ensure working effort among beneficiaries (Moffit, 2001; Tabor, 2002)

-Latin America: no reduction in labour force participation and effort after these plans were implemented (Attanasio et al, 2004)

III.3 Informality

High percentage of workers that do not contribute to the social security, although decreasing trend (from 40% in 2004 to 35% in 2006)

Informality (social security coverage)

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

45.0

2001 2002 2003 2004 2005 2006

Private workers All workers

- The per capita income threshold set by MIDES could act as an incentive against formalization

- Household income net of the transfer was monthly checked against the social security records and if the theshold was surpassed, households were temporary removed from the program

-Previous evidence found a negative impact on informality for Jefas y Jefes (Argentina) (Gasparini et al, 2007)

IV. Methods

IV.1 Data

- Administrative records of PANES: housing conditions, personalcharacteristics and income for all applicants. Contains baselineinformation about all outcomes included in this proposal

- Two follow up surveys. Sample of succesful and unsuccesful applicants. First wave: December 2006– March 2007. Second one: March 2008-May 2008. Can be merged with administrative records. Sample size: 2000 treatment households, 1000 controls

-National Household Survey asked about application and reception of Ingreso Ciudadano and alimentary card (2006- 2007)

- Other administrative records: social security data, data on births and deaths, data from educational system.

IV.2 Modelling and testing

- Identification strategy: a )regresssion discontinuity, as participation in the program was decided on the basis of a score (Gertler and Martínez 2006); b) differences in differences (information on outcomes in three different moments in time, except for child labor that was not included in the baseline)

- The sample of the follow up survey consists of an important mass of observations in a tight surrounding of the cut point (within a threshold of 2% around the cutting point)

-Comparison of very similar households that are situated above and below the cut-off point.

Outcome variables have to present a monotonous relation with the score before treatment. The jump in the previously monotonous relation is captured by the regression discontinuity method

 

a) Pre treatment b) Post treatment

Source: Gertler and Martínez (2006)

Standarized score (proxy means test) Household income (per capita)

-Graphical analysis

-Regression analysis

- Differences in differences

- Sensitivity of results to different functional forms (linear, quadratic, etc.)

- Weaknesses: results from near the cut-off can not be extrapolated to all households

XScoreDTreatmenty )(10

XInterTreatmentTimey 3210

V. Final remarks

- Research team from Universidad de la República has provided technical assistance to Ministerio de Desarrollo Social for the design of the proxy means and of the policy evaluation. They support this initiative

- Close link between research team and policy makers. Results will be taken into account for new policies that are being implemented at the moment