Investments in Human Capital: Long-term Effects of Progresa- Oportunidades on Poverty and Migration in Rural Mexico

Abstract

This paper evaluates the effects on poverty reduction and migration of a conditional cash transfer program in Mexico named Oportunidades (previously Progresa). This program, the first in Latin America and the most imitated of its kind, was intended to increase human capital, which would eventually translate into poverty reduction. Linear and non-linear panel models are used to explore whether there are any such effects, and the implications for the effectiveness and evaluation strategies of the program. No significant effects of Progresa-Oportunidades in reducing income poverty or affecting international migration were found at the rural level. However, there is a weak effect in the case of domestic migration. As data continues to become available through Mexico's official Secretariat of Social Development, there will be more opportunities for further exploration of the relationship between this program, poverty alleviation, and migration outcomes. Limitations and further recommendations for this study are also discussed.

Headnote

Abstract

This paper evaluates the effects on poverty reduction and migration of a conditional cash transfer program in Mexico named Oportunidades (previously Progresa). This program, the first in Latin America and the most imitated of its kind, was intended to increase human capital, which would eventually translate into poverty reduction. Linear and non-linear panel models are used to explore whether there are any such effects, and the implications for the effectiveness and evaluation strategies of the program. No significant effects of Progresa-Oportunidades in reducing income poverty or affecting international migration were found at the rural level. However, there is a weak effect in the case of domestic migration. As data continues to become available through Mexico's official Secretariat of Social Development, there will be more opportunities for further exploration of the relationship between this program, poverty alleviation, and migration outcomes. Limitations and further recommendations for this study are also discussed.

Keywords: Conditional Cash Transfers; Poverty; Migration;Mexico.

(ProQuest: ... denotes formulae omitted.)

Poverty has become an intrinsic part of rural Mexico. The Mexican government has followed different strategies in order to palliate the burden of domestic poverty. During World War I, the Mexican authorities and the United States government agreed to send Mexicans to work in the railroad and agricultural fields in the United States. However, the lack of effective coordination and planning contributed to an increasing flow of undocumented immigrants, which raised strong opposition for the continuation of the program from different labor groups (Romo, 1975). By the mid 1930s, the restoration of the ejido system as well as the redistribution of land, during the presidency of Lázaro Cárdenas, contributed significantly in the reduction of landless laborers, which declined more than 50 percent from 1934-1940 (Thiesenhunsen, 1996). However, Ibarra Mendivil (1996) argues that Mexican land reform was incomplete since the ejido program remained locked into peasant technologies.

In the 1940's the Mexican government embarked again in a joint project with the United States authorities and implemented the "Mexican Farm Labor Program", also known as "Bracero Program." According to Fennelly (2007), compared to the temporary work program during WWI, the Bracero program was far more effective at alleviating poverty in Mexico. However, like the previous one, this program had the unintended effect of increased undocumented migration. Besides, as pointed by different authors, migration, particularly international migration, has intrinsic social costs such as disrupted families, rapid urbanization, and political tensions (Massey et al., 1993; Massey & Espinosa, 1997; Garza, 1999; William & Massey, 2002; Angelucci, 2005; Rubalcava & Teruel, 2005; Stecklov et al., 2005; Oliver, 2009).

Most recently, during of the 1980s, Mexico's government strategy for poverty alleviation consisted basically of domestic efforts rather than new agreements to send Mexican workers to its northern neighbor. Among these, the National Solidarity Program (Pronasol), 1989 - 1994, was an explicit attempt to combat poverty, but ineffective in accomplishing its goal of reaching the poorest. Besides, it was mostly used to strengthen political control of the Institutional Revolutionary Party (PRI) (Levy, 1991; Székely, 1998; Diaz-Cayeros, Estévez & Magaloni, 2006). It was succeeded in 1997 amid an economic crisis by the Program for Education, Health and Nutrition (Progresa). Among the main differences, Progresa was based on a specific targeting strategy rather than on Pronasol's open and very broad selection of beneficiaries. In addition, Progresa was a more transparent public policy, less likely to be subject to discretionary decisions. It also required co-responsibility.

In 2002 the Progresa changed its name to Oportunidades and increased its coverage. By 2005, the program had reached its statutory limit of five million households, but after sixteen years of its implementation in rural areas there is the question of how effective the program has been in breaking up the intergenerational poverty cycle (Escobar Latapí, 2009).

When we look at the official poverty measures we can assess the relevance of the problem. From 1996 to 2006, the three official lines registered a general negative trend, meaning a reduction in the number of Mexicans living in poverty (Figure 1). However, since 2008 there has been a constant increase in all lines. The last measures reveal that in 2012 around half of the population in Mexico was living in asset poverty, and near to 20 percent of it was living below the food poverty line.

One of the questions explored in this research is the extent to which Progresa-Oportunidades (PO) has alleviated rural poverty. Another is whether there have been unintended effects such as giving incentives to people in rural areas to move to urban areas, creating pockets of poverty there. Also of interest is whether the program provides greater incentives for people to undertake the even riskier decision to migrate to the United States searching for better-paid jobs.

The Program and its Evaluations

Basically PO was created as a conditional cash transfer.2 This means that beneficiaries have to comply with certain program requirements such as school attendance, regular clinic checkups and nutritional meetings. The amount of benefits per household depends on different factors related to household composition.3

Since its first evaluations, PO showed significant impact on different poverty dimensions. Specifically, Skoufias & McClafferty (2001) and found a relatively significant effect of PO on income poverty reduction: from 1997 to 1999 the number of people living in poverty fell 10 percent; the depth of poverty went down 30 percent and the poverty severity index dropped 45 percent.4 However, it is not clear the extent to which these poverty measures are comparable to the three official poverty lines (food, capabilities, and asset).

In a more extensive impact evaluation study, Skuofias (2005) provides evidence for the effect of PO on individuals's income and human capital. For instance, PO has positive effects on both school enrolment and educational attainment. More recent studies support these findings (de Janvry et al., 2006; de Brauw & Hoddinott, 2011; Gertler, Patrinos & Rubio- Codino, 2012). Skuofias (2005) also found positive impacts of PO on individuals's health and wellbeing. Barham (2011) discussed this in the case of children, while Barham and Rowberry (2013) explored the effects on health conditions of the elderly.

Skuofias (2005) also shows that PO has important effects on children's productivity. On the same note, de Janvry et al. (2006) found that although PO positively affect school enrolment, it does not prevent child work. Reggio (2011) expands on this by showing gender differences. Skuofias (2005) does not assess the effect of PO on poverty based on income measures, however, he found a positve impact of the program on household's food expenditure. In a later study, Attanasio et at. (2013) shows similar findings.

As for the effects of the program on migration, previous research does not provide clear evidence. An analysis of official evaluations reveal different and inconclusive findings: Parker & Scott (2001) found that migration rates tend to be lower for program beneficiaries, but Rubalcava & Teruel (2005) found a positive effect on both domestic and international migration. The difference between these two studies lies in the comparison group used for the analysis.

Non official studies also provide contradictory findings. For instance, Stecklov et. al (2005) found that PO has no effect on international migration, but with a positive effect on domestic migration. However, Angelucci (2005) used a different specification and reached opposite findings with the PO having a positive impact on migration to the United States and no effect on domestic migration. Also, Azuara (2009) used an alternative proxy to operationalize total migration in rural areas fully covered by the program as the distance to transport infrastructure. He concludes that reductions in the beneficiary population in rural areas are due to PO since the aid received is used for migration purposes, but he fails to provide clear evidence.

Data, Hypotheses, and ModelSpecifications

This research uses a dataset that is part of a series of household surveys collected since 1997. The last wave was gathered in 2007, ten years after the program's implementation. This one in particular has not been fully explored until now, and allows for assessing long-term effects of PO on migration and poverty.

The design and implementation was in charge of the PO administration with the collaboration of national and internationally well-known institutions such as the National Public Health Institute (Instituto Nacional de Salud Pública, INSP), the Center for Research on Social Anthropology (Centro de Investigaciones y Estudios Superiores en Antropología Social, CIESAS), the International Food Policy Research Institute (IFPRI), the Center for Economic Research and Teaching (Centro de Investigación y Docencia Económicas, CIDE), and the National Council for the Evaluation of Social Development Policy (Consejo Nacional de Evaluación de la Política de Desarrollo Social, CONEVAL).

Since the PO provides significant cash transfers that alter households' disposable income, it is expected to reduce income poverty. Thus, the first hypothesis is:

Hypothesis 1: PO has reduced income poverty in rural areas of Mexico after 10 years of implementation.

It also is of interest to test the hypothesis of Levy (2008) on whether PO is a "good intended" policy with "bad outcomes". Therefore we test for the effect of PO on poverty after controlling for the effects of other relevant covariates mentioned in the literature (Haughton & Kandker, 2009). Our second hypothesis would be:

Hypothesis 2: The effect of PO on poverty is significant after controlling for the effect of other social programs, infrastructure, and geography.

Also, it is of relevance the conflict between the findings of Stecklov et al. (2005) and those of Angelucci (2005). For this reason, there are two competing hypotheses:

Hypothesis 3a (Stecklov et al., 2005): PO has had a positive effect on domestic migration in rural areas for the period of 1997 to 2007, without effects on international migration.

Hypothesis 3b (Angelucci 2005): PO has had a positive effect on international migration in rural areas for the period of 1997 and 2007, with no effects on domesticmigration.

Finally, Massey et al. (1994) in their theory of cumulative causation in migration argue that the existence of strong migrant networks could induce migration. Therefore the effects of PO on migration might be mediated by presence of migrant networks. In other words, when there are strong migrant networks, cash transfers like those provided by PO might represent an extra incentive to migrate for potential migrants. This would make us look at PO as a important determinant of migration. Thus, the fourth hypothesis can be stated as:

Hypothesis 4: The effect of Progresa-Oportunidades on domestic and international migration is mediated by the existence of migrant networks.

Two different model specifications are used to evaluate the previous hypotheses. The first specification assesses the program effect on poverty, while the second one evaluates the program impact on migration. The regression equation to evaluate the program effect on poverty is as follows:

...

Where,

Subscript h refers to the h-th household

Subscript j stands for the j-th coefficients

z: food poverty line for rural areas.

y: Monthly household income reported in the surveys.

t: Dummy variable allowing for comparison between the pretreatment measure or baseline period in 1997, and any of the subsequent post treatment measures in 1998, 1999, 2000, 2003, and 2007

PO: Whether the household receives the benefits of Progresa-Oportunidades

HM: Number of household members

AGE: Age of the head of the household in years.

GEN: Gender of the head of the household

E: Educational attainment of the head of the household

WORK: Working status of the head of the household

MAR: Marital status of the head of the household.

SP: Household with access to other social programs, such as Procampo, DIF, Temporary Job Program, Liconsa, retirement, or pension funds.

MIGD: Whether a household member migrated to work during the year of the survey, with D = M for migration within Mexico and D = U for migration to the United States

REM: Whether the household received remittances the year before the survey

SCH: Number of primary and secondary schools in the locality, as a proxy for infrastructure.

HEALTH: Number of health clinics in the locality (infrastructure proxy)

GEO: Local geographic factors such as droughts, plagues, flood, or hurricanes

e: Random error

Since the decision to migrate is operationalized through a categorical variable with individual beneficiaries either migrating or not, the specification of the migration equations requires a logit regression.

...

Where

Subscript i refers to the i-th migrant

Subscript h refers to the h-th household

Subscript j stands for the j-th coefficient

*** : Whether a household member migrated to work the year before the survey, with D = M for domestic migration, and D = U for migration to the United States

t: Dummy variable allowing for comparison between the pretreatment measure or baseline period in 1997, and any of the subsequent post treatment measures in 1998, 1999, 2000, 2003, and 2007

PO: Whether the household receives the benefits of Progresa-Oportunidades

FN: Existence of family networks

AGEM: Age in years of i-th migrant from the h-th household

GENM: Gender of the of the i-th migrant from the h-th household

EDUCM: Educational attainment of the i-th migrant from the h-th household

MARM: Marital status of the i-th migrant from the h-th household

KINM: Kinship of the i-th migrant with the head of the h-th household

ε: Random error

Table 1 provides a schematic description of the way each variable is operationalized.

Solution to the RegressionModels

In order to assess the outcomes of PO across time a consistent balanced panel was created by keeping only the respondents who were interviewed in all surveys. The regression outcomes for three different estimates are presented in Figure 2. Model 1 is the baseline for the Fixed Effects (FE) estimates.6 This model shows that a PO has on average a positive and significant effect on household's income. However, as the the extended models reveal, when controlling for other factors the effect of PO becomes negative, which would have meant that beneficiary households had on average a lower ratio of monthly income to the food poverty line. But as model 3 shows, the effect is not significant.

The set of year variables is included to control for the effect of time. All the years but 2007 are positive and significant. In other words, the average ratio of monthly household income to the food poverty line was higher in each of these years compared to that of 1997, the year of PO inception. Moreover, models 2 and 3 reveal the effect of PO benefits was only significant in 1999. These results seem to be counterintuitive since PO is intended to have a long term impact by helping its beneficiaries to break up with the intergenerational povertycycle.

In regards to household demographics, particularly those related to the head of the household, model 2 and 3 show that age has a positive and non-linear effect on the ratio of monthly household income to the food poverty line. In addition, gender is a significant factor, revealing that monthly household income tends to be higher when the head of the household is male. As expected, when the head of the household is employed monthly household income is higher than otherwise. In the case of marital status, it was found not to be relevant to explain changes in household income. In regards to education, the last two models show that monthly household income is on average higher when the head of the household has completed either basic or high school education than that of those individuals with no educational skill.7

In regards to households covariates, the estimates show that the larger the number of household members the higher the monthly household income. Further, households with access to other social programs had a higher ratio of monthly household income compared to those households without access to any of them. In the case of migration, it was found an important difference between domestic and international migration. Had a household member migrated domestically the year before the survey, the lower the monthly household income compared to households without domestic migration. As for the effect of international migrantion on poverty, it was found not significant. This provides empirical evidence for the hypothesis that people living in poverty conditions in rural areas tend to migrate more domestically than internationally.

Finally, the estimates show that none of the proxies for geographic factors and infrastructure are not significant. This is not say that these factors do not matter on a poverty profile, but better proxies should be considered.

PO Benefits and Migration

In the analysis of the effects of PO on migration the available data were desagregated at the individual level. Two separate dichotomous proxies were created for each type of migration: domestic and international. Migration equals 0 if no migration related to work was observed and 1 if an individual in the household migrated for labor reasons. A logit approach was followed in order to model the nonlinearity of the dependent variable.8

The regression outcome of the logit model is shown in Figure 3. As observed, the estimates shows that the probability of domestic migration is significantly lower for individuals in rural households with PO benefits compared to those in the control group. As to international migration, Model 2 reveals that there are no significant effects of PO on labor international migration.9

The Year variables capture the effect of time on migration. With 1998 as the reference for comparisons, Model 1 reveals that labor domestic migration was less likely to be observed in 1999, 2000, and 2007. In the case of Model 2, labor international migration was more likely to occur in 2000 and 2003.

The interactions of time and PO show that the effect of the program on domestic migration varied over time. Domestic migration for labor reasons was more likely to occur in households with PO benefits in 1999 and 2000 relative to similar potential migrants in the control group. However, as observed in Model 2 column, the program had no significant effects on international migration in any of the years under analysis. These findings would support the hypothesis that migration in rural areas is a 2-step decision. First, individuals tend to migrate domestically, and eventually they do it internationally. In addition, the results provide evidence in favor of Hypothesis 3a and against Hypothesis 3b.

The effect of social networks on rural migration, both domestic an international, was found positive and significant. In other words, the presence of relatives or friends who had migrated before was a decisive factor considered by individuals in their decision to migrate. This is in accordance with the network theory of migration. The presence and strength of migrant networks increases the expected returns and reduces both the costs and risks of migration. This in turn makes migration more likely to happen.

As for migrants demographic covariates, age has no significant effects on domestic migration. But age was found positive, significant, and nonlinear in the case of international migration. Gender was also found significant with males more likely to migrate both domestically and internationally than females.

With respect to the effect of education on the decision to migrate, individuals with high school diploma or lower educational attainments are more likely to migrate domestically for work, and less likely to do it internationally. In fact, the models suggest that education has in general a positive effect on domestic migration, but a negative or no effect at all on international migration, both for labor reasons. Interestingly, marital status tends to reduce the probability of migration either domestically or internationally of individuals living in rural areas.

Finally, kinship has different effects on the decision to migrate. For instance, it is more likely to observe that husbands/wives of the head of the household are more likely to migrate internationally in the search for job opportunities, while sons/daughters of the head of the household tend to migrate more domestically than internationally for the samereasons.

Conclusions

In this section the research hypotheses set down before are revisited. Hypothesis 1 was that PO had had a positive effect on income poverty reduction without controlling for other factors. The empirical findings support this hypothesis. This means that rural households that benefited from PO between 1997 and 2007 had, on average, a higher ratio of monthly household income to the food poverty line (lower income poverty) than rural households in the control group during the same period of analysis.

Hypothesis 2 also tested for the effect of PO on income poverty but now controlling for other relevant factors: access to other social programs as well as demographic variables. Surprisingly, PO had a weak effect on income poverty in rural households. Apparently the program significantly contributed to income poverty reduction in 1999 only. This invalidates the previous conclusions in regards to hypothesis 1.

The next set of hypotheses evaluates the possible effects of PO on both domestic and international labor migration. The models reveal that rural households with PO benefits were more likely to migrate domestically for labor reasons in 1999, 2000, and 2007 than rural households in the control group. No effects of the intervention were found in the case of international migration for labor. This supports Hypothesis 3a which is based on the arguments of Stecklov et al. (1995). The same findings provide no support for Hypothesis 3b, based on Angelucci (2005) who argues that PO increases international migration and has no effects on domestic migration.

Hypothesis 4 evaluates the network theory of migration which states that the decision to migrate is mediated by the existence and strength of migrant networks. The regression models provide support for this theory, since the proxies for the strength of the migrant networks are significant for domestic and international migration for labor reasons.

Overall, the regression outcomes reveal weak and ambiguous effects of PO on income poverty and on most forms of migration. At this point it is important to acknowledge limitations of the data as well as the study design. Some limitations may be related to the structure of the questionnaires, in particular the last two surveys in 2003 and 2007, which were significantly changed. In addition, using a dichotomous variable instead of a continuous one in the case of PO benefits restricts the scope of the analysis. The justification in this case is that data is only available for some years. Also, it would have been better to have individuals as the unit of analysis, but data at this level was again limited. The 2003 surveys reported access to PO benefits at the household level only. Policy evaluations based on data consistently collected tend to be more efficient. This in turn would ensure better policy outcomes, which might interest government officials.

An inconsistent periodicity of the information could be another source of problems. After 2000 the following survey was collected in 2003 and the next one in 2007. The gaps in between the surveys make it difficult for an evaluation of the effects of the program during those years. Not to mention that an up to date evaluation in rural areas is required, since the last one was performed almost 7 years ago.

Due to the limitations mentioned above, it would not be objective to conclude that PO has been ineffective in reducing poverty. What we know from the analysis is that PO does not have a consistent and significant impact on income poverty. But, as official and non-official evaluations suggest, the program has had a positive impact on other poverty dimensions such as education or health. This is something beyond the scope of this research.

The questions of how the accumulation of human capital through PO is translated into higher income or whether PO has altered income distribution in Mexico require further research.

Footnote

2 There is a food supplement as part of the nutritional component of Progresa.

3 Among the factors considered are age, number, and gender of members. It is explicitly intended to increase women's empowerment. Thus, cash transfers are preferentially provided to housewives, and scholarships are greater for girls than for boys, since girls show higher drop-out rates than boys.

4 The authors compared Progresa with an untargeted transfer. For the latter they found a reduction in poverty depth of 28 percent, while poverty severity index decreased 36 percent. Therefore, they conclude that Progresa has a relatively larger impact on povertyreduction.

5 Standardization of income by the food poverty line allows distinguishing those living in extreme poverty from those in a higher threshold. This is in accordance to the program's goal of reducing the number of Mexicans in extreme poverty conditions. A value of 1 means that monthly household income is at the food poverty line, a ratio of less than 1 means that monthly household income is below the food poverty line, and a ratio of more than 1 means that monthly household income is above the food poverty line. Besides, income as a continuous variable contains more information and is less likely to be subject to specification errors when modeled through regression analysis (Haughton & Khandker 2009).

6 The decision of using FE instead of Random Effects (RE) was based on the Hausman test, which revealed that RE estimates were inconsistent.

7 Based on a constitutional amendment signed by President Felipe Calderón in 2011, starting the 2012-2013 academic year, the basic mandatory education includes high school.

8 Random Effects (RE) was preferred over Fixed Effects (FE) since most of the variation of the time-varying variables age and marital status was due to betweenvariations.

9 Mainly to the UnitedStates

References

References

1. Angelucci, M. (2005). Aid Program's Unintended Effects: The Case of Progresa and Migration. July. Retrieved February 20, 2009, from http://dx.doi.org/10.2139/ssrn.868646

2. Attanasio, O., Di Maro, V., Lechene, V., & Phillips, D. (2013). Welfare consequences of food prices increases: Evidence from rural Mexico. Journal of Development Economics, 104, pp. 136-151. http://dx.doi.org/10.1016/j.jdeveco.2013.03.009

3. Azuara, O. (2009). Does Poverty Alleviation Increase Migration? Evidence from Mexico. Munich Personal RePEc Archive (MPRA). MPRA Paper. September 30. Retrieved August 29, 2009, fromhttp://mpra.ub.uni-muenchen.de/17599/

4. Barham, T. (2011). A healthier start: The effect of conditional cash transfers on neonatal and infant mortality in rural Mexico. Journal of Development Economics, 94 (1), pp. 74-85. http://dx.doi.org/10.1016/j.jdeveco.2010.01.003

5. Barham, T., & Rowberry, J. (2013). Living longer: The effect of the Mexican conditional cash transfer program on elderly mortality. Journal of Development Economics, 105, pp. 226-236.http://dx.doi.org/10.1016/j.jdeveco.2013.08.002

6. CIDE, & CIESAS (2005) Evaluation Documents. Oportunidades, External Evaluations. December 13, 2005. Retrieved March 3, 2010, from http://www.oportunidades.gob.mx:8010/en/docs/docs2001.php

7. CONEVAL (2009). Evolución de la pobreza en México. Power Point Presentation, Mexico City: CONEVAL, 2009. Retrived February 23, 2009, from http://www.coneval.gob.mx/rw/resource/coneval/med_pobreza/3967.pdf

8. CONEVAL (2007). Los Mapas de Pobreza en México. July 2007. Retrieved August 5, 2009, from http://www.coneval.gob.mx/SalaPrensa/Paginas/Comunicados_de_prensa/Comunicad o-No-0012007.aspx

9. CONEVAL (2005) Mapas de Pobreza 2005. Retrieved August 5, 2009, from http://www.coneval.gob.mx/rw/resource/coneval/med_pobreza/3967.pdf

10. de Brauw, A., & Hoddinott, J. (2011). Must conditional cash transfers programs be conditioned to be effective? The impact of conditioning transfers on school enrolment in Mexico. Journal of Development Economics, 96 (2), pp. 359-370. http://dx.doi.org/10.1016/j.jdeveco.2010.08.014

11. de Janvry, A., Finan, F., Sadoulet, E., & Vakis, R. (2006). Can conditional cash transfer programs serve as safety nets in keeping children at school and from working when exposed to shocks? Journal of Development Economics, 79(2), pp. 349-373. http://dx.doi.org/10.1016/j.jdeveco.2006.01.013

12. Diaz-Cayeros, A., Estévez, F., & Magaloni, B. (2006). The Struggle Against Poverty. In A.Diaz-Cayeros, F. Estévez & B.Magaloni (Eds.), Vote-Buying, Poverty and Democracy: The Politics of Social Programs in Mexico, 1989-2006 (pp. 2-44), Standford, California: StandfordUniversity.

13. Escobar Latapí, A. (2009). Mexico's Progresa-Oportunidades Programs: Where Do We Go from Here? In M.Jo Bane & R. Zenteno (Eds.), Poverty and Poverty Alleviation Strategies in North America (pp. 203-232), Boston: David Rockefeller Center for Latin American Studies.

14. Fennelly, K. (2007). U.S. Immigration: A Historical Perspective. The National Voter, pp. 4-7. Retrived February 9, 2009, from http://www.hhh.umn.edu/people/kfennelly/pdf/immigration_historical_perspective.pdf

15. Garza, G. (1999). Global economy, metropolitan dynamics and urban policies in Mexico. Cities, 16(3), pp. 149-170. http://dx.doi.org/10.1016/S0264-2751(99)00013-X

16. Gertler, P. J., Patrinos, H. A., & Rubio-Codino, M. (2012). Empowering parents to improve education: evidence from rural Mexico. Journal of Development Economics, 99(1), pp. 68-79. http://dx.doi.org/10.1016/j.jdeveco.2011.09.004

17. Haughton, J., & Khandker, S.R. (2009). Handbook on Poverty and Inequality. Washington, D.C.: The WorldBank.

18. Ibarra Mendivil, J. L. (1996). Recent Changes in the Mexican Constitution and their Impact on the Agrarian Reform. In L. Randall (Ed.), Reforming Mexico's Agrarian Reform (pp. 49-60), Armonk, New York: M.E. SharpePress.

19. Levy, S. (1991). Poverty Alleviation in Mexico. Policy, Research, and External Affairs working papers; no. WPS 679. Country operations. Washington, DC: World Bank. Retrived April 15, 2010, from http://documents.worldbank.org/curated/en/1991/05/699980/poverty-alleviation-mexico

20. Levy, S. (2008). Good Intentions, Bad Outcomes: Social Policy, Informality, and Economic Growth in Mexico. Washington, D.C.: The BrookingsInstitution.

21. Massey, D.S., & Espinosa, K.E. (1997). What's Driving Mexico-U.S. Migration? A Theoretical, Empirical, and Policy Analysis. The American Journla of Sociology, 102(4), pp. 939-999. http://dx.doi.org/10.1086/231037

22. Massey, D. S., Arango, J., Hugo, G., Kouaouci, A., Pellegrino, A., & Taylor, J. E. (1993). Theories of International Migration: A Review and Appraisal. Population and Devlopment Review, 19(3), pp. 431-466.http://dx.doi.org/10.2307/2938462

23. Massey, D. S., Goldring, L., & Durand, J. (1994). Continuities in Transnational Migration - An Analysis of 19 Mexican Communities. American Journal of Sociology, 99(6), pp. 1492-1533. http://dx.doi.org/10.1086/230452

24. Parker, S., & Scott, J. (2001). La Evaluación de los Impactos de Progresa sobre los Hogares: Encuestas de Evaluación (Encel). Evaluación del Programa de Educación, Salud y Alimentación (PROGRESA) 2001 . Evaluation, Mexico City: CIDE. Retrived June 3, 2010 from https://www.prospera.gob.mx/EVALUACION/es/wersd53465sdg1/docs/2001/cide_20 01_evaluacion_impacto.pdf

25. Reggio, I. (2011). The influence of the mother's power on her child's labor in Mexico. Journal of Development Economics, 96(1), pp. 95-105. http://dx.doi.org/10.1016/j.jdeveco.2010.07.002

26. Romo, R. (1975). Reponses to Mexican Immigration, 1910-1930. Aztlán: A Journal of Chicano Studies, 6(2), pp. 172-194.

27. Rubalcava, L. N., & Teruel, G.M. (2005). El efecto de Oportunidades sobre la dinámica demográfica de los hogares beneficiarios y las decisiones de migración de sus integrantes, en localidades rurales. In B. H. Prado & M. H. Ávila (Eds.), Evaluación externa de impacto del Programa Oportunidades 2004. Aspectos Económicos y Sociales (pp. 153-193). Cuernavaca, México: Instituto Nacional de Salud Pública, INSP, and Centro de Investigaciones y Estudios Superiores en Antropología Social, CIESAS.

28. Skuofias, E. (2005). PROGRESA and its impacts on the welfare of rural households in Mexico. Reserch Report 139, pp. 1-84. Washington, D.C.: International Food Policy Research Institute. Retrived March 9, 2009, from http://cdm15738.contentdm.oclc.org/utils/getfile/collection/p15738coll2/id/80436/file name/80437.pdf

29. Skuofias, E., & McClafferty, B. (2001). Is PROGRESA Working? Summary of the Results of an Evaluation by IFPRI. FCND Discussion Paper No. 118, pp. 1-65. Intenational Food Policy Research Institute, Washington, D.C.. Retrived March 9, 2009, from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.58.3872&rep=rep1&type=pdf

30. Stecklov, G., Winters, P., Stampini, M., & Davis, B. (2005). Do conditional cash transfers influence migration? A study using experimental data from the Mexican Progresa program. Demography, 42(4), pp. 769-790. http://dx.doi.org/10.1353/dem.2005.0037

31. Székely, Miguel (1998). The Economics of Poverty, Inequality and Wealth Accumulation in Mexico. New York: St. Martin's Press in association with St. Antony's College, Oxford.

32. Thiesenhunsen, W. C. (1996). Mexican Land Reform, 1934-91: Success or Failure?, In L. Randall (Ed.), Reforming Mexico's Agrarian Reform (pp. 35-47). M.E. Sharpe Press.

33. William, K., & Massey, D.S. (2002). The Culture of Mexican Migration: A Theoretical and Empirical Analysis. Social Forces, pp. 981-1004.