Skip to main content Accessibility help

Impact Evaluation of Chile Solidario: Lessons and Policy Recommendations


This article evaluates the impact of the Chile Solidario anti-poverty programme. The evaluation is based on propensity score matching and a difference-in-difference estimator along with databases of Social Assistance Committee forms. The results show a positive but small impact on employment and housing along with a slightly negative impact on self-generated income. They also suggest that gains tend to be concentrated in the first phase, during which beneficiaries work with a family support professional, and that these benefits may not be sustainable. Participant families show absolute gains in income and employment, but these may be attributed to environmental conditions rather than the programme; this raises doubts about the premise that these families were initially marginalised from the economy and social networks.

Este artículo evalúa el impacto del programa anti-pobreza Chile Solidario. La evaluación se basa en la nivelación por puntaje de propensión y un estimador de diferencia-en-diferencias junto a bases de datos que contienen los formularios del Comité de Asistencia Social. Los resultados muestran un impacto pequeño pero positivo sobre el empleo y la vivienda junto a un pequeño impacto negativo sobre el ingreso del autoempleo. También sugieren que las ganancias tienden a concentrarse en la primera fase durante la cual los beneficiarios trabajan con un profesional del Apoyo Familiar y que tales beneficios puede que no sean sostenibles. Las familias participantes muestran ganancias absolutas en ingreso y empleo, aunque esto pueden atribuirse a condiciones ambientales en vez del programa, lo que genera dudas acerca de la premisa de que estas familias estuvieron desde un principio marginadas de las redes económicas y sociales.

O artigo avalia o impacto do programa anti-pobreza Chile Solidário. A avaliação basea-se no escore de propensão e o estimador de diferença-em-diferenças junto com os bancos de dados que possuem as fichas do Comitê de Assistência Social. Os resultados demonstram um impacto positivo, porém pequeno sobre empregos e moradia além de um impacto levemente negativo sobre a renda auto-gerada. Também sugerem que ganhos tendem a concentrar-se na primeira fase durante a qual os beneficiários trabalham junto a um profissional de Ajuda Familiar e que estes benefícios podem não ser sustentáveis. Famílias participantes apresentaram ganhos absolutos em renda e empregos, mas estes podem ser atribuídos a condições contextuais ao invés do programa, fato que levanta dúvidas acerca da premissa de que estas famílias eram inicialmente marginalizadas das redes econômicas e sociais.

Hide All

1 Raczynski, Dagmar, Sistema Chile Solidario y la política de protección social de Chile (Santiago: CIEPLAN, 2008).

2 Emanuela Galasso, ‘Alleviating Extreme Poverty in Chile’, unpubl. Development Research Group report (Washington, DC: World Bank, 2006); Marcela Perticara, ‘Análisis cuantitativo de impacto del sistema Chile Solidario’, unpubl. report (Santiago: MIDEPLAN, 2007); Pedro Carneiro and Emanuela Galasso, ‘Lessons from the Evaluation of CS’, unpubl. Development Research Group report (Washington, DC: World Bank, 2008). The results of these evaluations are still in development. See

3 The baseline uses 2003 CASEN survey households as well as a sample of households participating in CS because it was not clear that the CASEN sample would include sufficient participant households. A second round of the panel survey was conducted in 2004, during which a control group was selected based on households not included in the 2003 CASEN survey. Households were to be paired according to probability of selection for CS, estimated according to a set of variables related to this event. In addition, a new set of CS participants – those who entered the programme between 2003 and 2004 – was added on the condition that they had participated in the baseline CASEN survey, which led to a reduced number of cases. A similar treatment was used for the third round of the panel implemented in 2006.

4 This affects most of the techniques for estimating programme impact. However, discontinuous regression and other tools can work in the absence of the baseline. Of the studies cited in note 2, only that of Galasso and Carneiro uses this methodology. The main limitation of this approach is that it considers only the results for the population situated around the point of discontinuity.

5 The Social Assistance Committee was a municipality-level institution that was in charge of applying central government guidelines for targeting social policies at the local level.

6 See Appendix for a fuller explanation.

7 See Fiszbein, Ariel, Schady, Norbert Rüdiger and Ferreira, Francisco H. G., Conditional Cash Transfers: Reducing Present and Future Poverty (Washington, DC: World Bank, 2009), chap. 4.

8 Consultores, Focus, Caracterización y evaluación del vínculo entre el apoyo familiar y las personas participantes del Programa Puente (Santiago: Focus Consultores, 2004); Trucco, Daniela and Nun, Eleonora, Sistematización de evaluaciones cualitativas del Programa Puente y Sistema de Protección Chile Solidario (Santiago: UNDP, 2008).

9 See Focus Consultores, Caracterización y evaluación; and Trucco and Nun, Sistematización de evaluaciones cualitativas.

10 Trucco and Nun, Sistematización de evaluaciones cualitativas.

11 Based on Ruz, Miguel Ángel and Palma, Julieta, Análisis del proceso de elaboración e implementación del sistema Chile Solidario (Santiago: Instituto de Asuntos Públicos, Universidad de Chile, 2005). See also Osvaldo Larrañaga and Dante Contreras, ‘Chile Solidario y el combate a la pobreza’, in Osvaldo Larrañaga and Dante Contreras (eds.), Las nuevas políticas de protección social en Chile (Santiago: Uqbar, 2010), pp. 43–76.

12 CASEN surveys, MIDEPLAN, 1990–2009.

13 A transcription of the parliamentary discussion is available at

15 However, there is some ambiguity in this regard given that the official discourses tend to privilege the goal of ending extreme poverty, which is measured as a lack of income. See Larrañaga and Contreras (eds.), Las nuevas políticas de protección social.

16 Below we provide a discussion of the validity of this observation in light of the results.

17 The amount of the SUF in July 2007 was CL$ 5,393 (US$ 10) for each beneficiary minor; PASIS fluctuated between CL$ 44,000 (US$ 83) and CL$ 51,000 (US$ 96) depending on the beneficiary's age. The potable water subsidy is higher for households that do not participate in CS.

18 These values are based on the July 2007 exchange rate.

19 The amount of the CS grant averages US$ 20 for the first two years. The amount of the Bolsa Familia (Family Allowance) in Brazil ranges from US$ 42 to US$ 98 for households in extreme poverty. Mexico's Oportunidades Programme offers a grant based on the number of students and seniors in the family that can exceed US$ 200 per month. The numbers correspond to purchasing power parity US dollars and were obtained from Sergei Suarez Dillon Soares et al., ‘Conditional Cash Transfers in Brazil, Chile and Mexico: Impacts upon Inequality’, International Policy Centre for Inclusive Growth Working Paper no. 35 (New York: UNDP, 2007).

20 Larrañaga, Osvaldo, Focalización de programas sociales en Chile: el sistema CAS, Social Protection Center, Latin America and Caribbean Region (Washington, DC: World Bank, 2003).

21 In 2006, the CAS form was replaced by the Social Protection Form.

22 Raczynski, Sistema CS y la política de proteccion social.

23 A family with a low CAS score that belongs to a municipality in which the CASEN survey underestimates extreme poverty may be excluded from the programme, and its slot could be taken by a family with a higher CAS score from a different municipality. This sample error results in an overestimation of the impoverished population.

24 ‘Treated’ refers to participants who entered the CS programme in the first cohort.

25 The CASEN database includes all CAS form variables so that the CAS score can be calculated in CASEN households.

26 See Andrés Hojman, ‘Evaluando el Programa CS: resultados utilizando el Panel CS y lecciones para la evaluación’, unpubl. MSc diss., Universidad de Chile, 2008; and Fernando Hoces, ‘Evaluación de impacto del Sistema CS: estudio de la trayectoria de los impactos utilizando el panel administrativo’, unpubl. MSc. diss., Universidad de Chile, 2008.

27 The propensity score matching methodology has a long track record in quasi-experimental impact evaluations (see Rajeev Dehejia and Sadek Wahba, ‘Propensity Score Matching Methods for Non-Experimental Causal Studies’, NBER Working Paper no. 6829 (Washington, DC: National Bureau of Economic Research, 1998); Heckman, James, Ichimura, Hidehiko and Todd, Petra, ‘Characterizing Selection Bias Using Experimental Data’, Econométrica, 66: 5 (1998), pp. 1017–98; Rosenbaum, Paul R. and Rubin, Donald B., ‘Constructing a Control Group Using Multivariate Matched Sampling Methods that Incorporate the Propensity Score’, American Statistician, 39 (1985), pp. 33–8. See Appendix for a fuller explanation.

28 For control group households, the person who would have received the grant is chosen (the household head in a single-parent case and the female partner in a two-parent case).

29 The result is consistent with those reported by Galasso and Perticara, whose studies indicate positive results for income in rural areas.

30 The correlation coefficient is 0.985.

31 The behaviour of the control subjects is consistent with developments in the economy. Between 2002 and 2003, GDP grew by 16.2 per cent and employment increased by 10.6 per cent.

32 See Focus Consultores, Caracterización y evaluación; and Trucco and Nun, Sistematización de evaluaciones cualitativas.

33 Trucco and Nun, Sistematización de evaluaciones cualitativas.

34 This reveals the role that psycho-social support can play in programme achievements. However, the databases available do not allow for a comprehensive evaluation of the family support workers’ contribution. An in-depth study of the role played by these professionals is ongoing.

35 See Neilson, Christopher, Contreras, Dante, Cooper, Ryan and Hermann, Jorge, ‘The Dynamics of Poverty in Chile’, Journal of Latin American Studies, 40: 1 (2008), pp. 257–73.

36 Only households with CAS forms are considered in both years.

37 Desarrollo, Asesorías para el, Evaluación de programas de la oferta pública en convenio con el Sistema Chile Solidario (Santiago: MIDEPLAN, 2006); and Necesidades y aspiraciones de las familias que han finalizado la etapa de apoyo psicosocial del Sistema de Protección Social Chile Solidario (Santiago: MIDEPLAN, 2005).

38 Irma Arriagada and Charlotte Mathivet, Los programas de alivio a la pobreza Puente y Oportunidades: una mirada desde los actores, Serie Políticas Sociales no. 134 (Santiago: ECLAC/CEPAL, 2007).

* The authors thank Ricardo Paes de Barros, Luis Díaz, Rodrigo Herrera, Heidi Berner, Fernando Hoces and Andrés Hojman for their comments. They are grateful to participants in the Chile Solidario Evaluation Seminar organised by the Ministry of Planning and Cooperation and the Seminar of the Association of Chilean Economists. The authors would also like to acknowledge funding granted by the Iniciativa Cientifica Milenio (Millennium Scientific Initiative) for Centro de Microdatos Project no. NS100041. Osvaldo Larrañaga is writing in his private capacity; this article should not be attributed in any way to the United Nations, related organisations or UN member states. Any errors or omissions are the sole responsibility of the authors.

Recommend this journal

Email your librarian or administrator to recommend adding this journal to your organisation's collection.

Journal of Latin American Studies
  • ISSN: 0022-216X
  • EISSN: 1469-767X
  • URL: /core/journals/journal-of-latin-american-studies
Please enter your name
Please enter a valid email address
Who would you like to send this to? *



Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed