## wpiea2020099-print-pdf

## Source details

**Canonical URL:** [wpiea2020099-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020099-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020099-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020099-print-pdf.pdf.json)

---

### Introduction: context, theory, and empirical background
- Cash transfer programs emerged in the mid-1990s in Latin America and have been implemented in almost 70 low- and middle-income countries (World Bank 2017).
- During the Covid-19 crisis many governments are resorting to social protection programs like cash transfer programs to sustain livelihoods.
- Theoretical channels:
  - Transfers may reduce labor supply through an income effect and a high tax rate on marginal earnings (Kanbur et al., 1994; Borjas, 2005). The larger the transfers and the larger the marginal tax rate on earnings, the larger the reduction of labor supply is expected to be (Moffit 2002; Borjas 2005).
  - Transfers could increase labor supply if they relax liquidity and credit constraints and cover costs of job search (Baird, McKenzie and Özler 2018).
  - In contexts with large informal sectors, transfers may reduce formal labor market participation while leaving informal labor participation unchanged because informal income is hard to detect.
- Empirical evidence summarized:
  - Kabeer and Waddington (2015): only one out of eight reviewed studies finds negative impacts on labor force participation.
  - Bastagli et al. (2016): three out of eight studies point to an increase in labor force participation; one points to a reduction.
  - Banerjee et al. (2017): meta-analysis of seven RCTs finds “no significant effect of belonging to a transfer program on employment or hours of work in any of the seven programs”.
  - Selected Bolsa Família evidence: mixed findings across studies; some small positive effects on participation, some small decreases in hours worked (usually for women), and evidence of lower probability of leaving formal jobs for beneficiaries (Santos et al. 2016).
- Paper’s contribution:
  - Leverages an unannounced change in the eligibility cutoff and municipal-level rationing of entry to estimate causal impact of Bolsa Família on formal employment.
  - Key finding: participation in Bolsa Família had a positive effect on formal employment, stronger for younger beneficiaries.
  - Interpretation compatible with models incorporating job search costs, credit and liquidity constraints, and decision-making under scarcity (Mullainathan and Shafir 2013). Also consistent with evidence on long-term positive effects of CCTs (Barham, Macours and Maluccio 2018).

### Beneficiary selection, program features, and data construction
- Administrative and operational features:
  - Ministry of Citizenship (previously Ministry of Social Development – MDS) defines eligibility and authorizes payments.
  - In 2018 over a fifth of Brazil’s population was beneficiary of the program, which had a budget of about BRL 30 billion (equivalent to 0.4 percent of GDP).
  - Eligibility requires registration in the Single Registry for Social Programs (Single Registry, or Cadastro Unico para Programas Sociais, CadUnico).
  - Eligibility criterion: per capita household income as declared in the Single Registry (sum of all incomes of all members divided by number of members).
  - Municipal-level quotas: an estimate of the number of families in poverty or extreme poverty was established for each municipality; once the quota is reached, entry becomes more difficult.
  - Verification: self-declared income is cross-checked with administrative records, including RAIS; income from informal work is virtually impossible to detect.
  - Policy note: introduction of the “retorno garantido” in 2012 ensured a right to reinsertion for families who left the program to accept a job offer.
- Data sources and merging:
  - Single Registry (version extracted December 2008): approximately 70 million individual registers, yielding 34.3 million people of working age (working age defined as those aged between 18 and 65 (men) or 60 (women)); covers more than 25 million low-income families (about 70 million people).
  - Historical Database of Benefits of Bolsa Família: concatenation of 96 monthly payrolls from January 2004 to December 2011; over the period families in the program increased from 5.9 to 12.3 million; total of 19.8 million families who entered the program at some point; 7.43 million of which had left by the end of the period; contains exact date of entry and (eventual) exit for each family.
  - RAIS (Annual Social Information Report): in December 2011 RAIS reported 46.3 million formal employment relations; paper uses six annual RAIS versions from 2006 to 2011.
  - The three datasets were merged using the unique identifier “social identification number” present in all of them.
- Entry/exit counts (Table 1, 2004–2011):
  - 2004: Entry = 6,003,602; Exit = 78,573
  - 2005: Entry = 2,135,331; Exit = 219,642
  - 2006: Entry = 3,480,427; Exit = 1,028,380
  - 2007: Entry = 1,180,200; Exit = 937,245
  - 2008: Entry = 940,455; Exit = 1,333,643
  - 2009: Entry = 2,543,647; Exit = 1,849,878
  - 2010: Entry = 1,759,186; Exit = 1,145,725
  - 2011: Entry = 1,719,744; Exit = 839,958
  - Total: Entry = 19,762,592; Exit = 7,433,044

### Sample characteristics and descriptive findings
- Sample: working-age individuals registered in the Single Registry (December 2008).
- Participation groups (2006 and 2011):
  - BB (beneficiaries in both 2006 and 2011) = 35 percent.
  - BN (beneficiaries in 2006 but not in 2011) = 21 percent.
  - NB (not beneficiaries in 2006 but beneficiaries in 2011) = 16 percent.
  - NN (not beneficiaries in 2006 nor in 2011) = 28 percent.
- Selected group profiles (Table 2; percentages where indicated):
  - BB: Number working age individuals = 11,972,421; Living in North & NE = 70.2; Rural = 39.1; W/kids < 16yo = 79.6; Female HH = 58.4; Primary schooling = 49.6; Black = 74.7.
  - BN: Number working age individuals = 7,295,443; Living in North & NE = 45.6; Rural = 26.8; W/kids < 16yo = 64.2; Female HH = 56.8; Primary schooling = 38.7; Black = 65.1.
  - NB: Number working age individuals = 5,503,518; Living in North & NE = 58.9; Rural = 30.9; W/kids < 16yo = 71.7; Female HH = 61.7; Primary schooling = 38.0; Black = 73.1.
  - NN: Number working age individuals = 9,555,750; Living in North & NE = 43.5; Rural = 27.0; W/kids < 16yo = 48.4; Female HH = 56.1; Primary schooling = 36.9; Black = 65.8.
- Labor market participation dynamics (2006–2011):
  - Three quarters of working age individuals in the Single Registry (about 26 million people) never worked in the formal sector between 2006 and 2011.
  - Over 80 percent of BB had no formal employment at any point over the six-year period, compared to 73 percent of NN.
  - In BB, 5 percent had a continuous presence in formal employment; in NN, 12 percent.
  - Growth in formal labor market participation: 68 percent for individuals who were in the program (BB) vs. 30 percent for those who were not (NN).
- Age-specific growth in formal labor market participation (2006–2011):
  - Age 18 and 19: increase of 213 percent (overall); for BB, almost four-fold increase noted.
  - [20,30): 59 percent total; BB 96 percent; NB 48 percent; BN 68 percent; NN 40 percent.
  - [30,40): 26 percent total; BB 44 percent; NB 20 percent; BN 30 percent; NN 15 percent.
  - [40,50): 11 percent total; BB 21 percent; NB 9 percent; BN 14 percent; NN 5 percent.
  - 50+: -17 percent total; BB -11 percent; NB -15 percent; BN -16 percent; NN -20 percent.

### Identification strategy and estimation
- Core identification:
  - Regression discontinuity not used due to evidence of manipulation around eligibility cutoff.
  - Exploits an unannounced change in eligibility in July 2007: income eligibility cutoff raised from BRL100 per capita to BRL120.
    - Almost 73 thousand families previously ineligible became eligible.
    - Only about one third of them became beneficiaries due to Central Government budget allocation and municipal quotas.
  - Treatment group: households that became eligible and entered the program after the change.
  - Control group: households that became eligible but did not enter the program.
  - Analysis restricted to families that updated information before the announcement to avoid strategic manipulation.
  - Local Average Treatment Effects (LATE) estimated.
- Sample for causal analysis:
  - 72,781 individuals aged 18–65 in 2007 whose household became eligible due to the cutoff change.
  - Table 4 (entry after July 2007 change) counts:
    - (100,120]: No BFP = 49,148; BFP = 23,633; Total = 72,781.
    - (120,137]: No BFP = 40,183; BFP = 1,429; Total = 41,612.
    - Total = 89,331 No BFP; 25,062 BFP; 114,393 Total.
  - Geographic variation in take-up: share of newly eligible that became beneficiaries ranges from 1.3 percent in the Federal District to 48.7 percent in the state of Goiás.
- Estimation approach:
  - Propensity score matching (kernel weights) used to balance observables between treated and control groups.
  - Matching achieved balance: t-tests cannot reject equal means; chi-squared test cannot reject overall balance.
  - Additional checks: correlation tests show no evidence of correlation between municipal quota levels and growth in formal sector employment (correlation between -0.02 in 2010-2011 and 0.02 in 2008-2009).

### Outcomes, main results, and heterogeneity
- Outcome definitions:
  - "formal at least once": appeared at least once in RAIS over a certain number of years.
  - "formal throughout": appeared in RAIS every month throughout a certain number of years.
- Time periods: results primarily reported for 2010 and 2011; other windows considered include 2007-2011 and 2009-2011.
- Main LATE estimates (selected entries from Table 6; Δ LATE reported with Standard Errors):
  - All:
    - Formal at least once in 2010 or 2011: No BFP = 0.29; BFP = 0.33; Δ LATE = 0.047 (Standard Error (0.004)).
    - Formal consecutively between 2010 and 2011: No BFP = 0.22; BFP = 0.25; Δ LATE = 0.033 (Standard Error (0.004)).
  - Women:
    - Formal at least once in 2010 or 2011: No BFP = 0.204; BFP = 0.251; Δ LATE = 0.046 (Standard Error (0.005)).
    - Formal consecutively between 2010 and 2011: No BFP = 0.145; BFP = 0.171; Δ LATE = 0.026 (Standard Error (0.004)).
  - Men:
    - Formal at least once in 2010 or 2011: No BFP = 0.423; BFP = 0.462; Δ LATE = 0.039 (Standard Error (0.006)).
    - Formal consecutively between 2010 and 2011: No BFP = 0.343; BFP = 0.382; Δ LATE = 0.040 (Standard Error (0.006)).
- Age-differentiated LATE (Table 7; Δ LATE with Standard Errors) — probability differences for being formal in 2010/2011:
  - [18,25): Formal at least once: Δ LATE = 0.048 (Standard Error (0.010)); Formal consecutively: Δ LATE = 0.030 (Standard Error (0.009)).
  - [25,35): Formal at least once: Δ LATE = 0.064 (Standard Error (0.007)); Formal consecutively: Δ LATE = 0.047 (Standard Error (0.007)).
  - [35,45): Formal at least once: Δ LATE = 0.037 (Standard Error (0.007)); Formal consecutively: Δ LATE = 0.025 (Standard Error (0.006)).
  - [45,55): Formal at least once: Δ LATE = 0.009 (Standard Error (0.010)); Formal consecutively: Δ LATE = 0.006 (Standard Error (0.009)).
  - [55,65]: Formal at least once: Δ LATE = 0.011 (Standard Error (0.013)); Formal consecutively: Δ LATE = -0.002 (Standard Error (0.011)).
- Key interpretations:
  - Positive and statistically significant effects for the whole sample and for men and women separately.
  - Strongest impacts concentrated among younger cohorts, notably [25,35).
  - Attachment to the formal market is limited: share appearing throughout the period substantially smaller, especially for women.
- Mechanisms discussed:
  - Cash transfers may finance job-search costs and ease transitions into formal jobs.
  - Indirect psychological benefits (reduced scarcity, improved cognitive processing and decision-making) may improve labor market decisions.
  - Lack of detectable impact for older age groups ([45,65]) consistent with "scarring" effects of prolonged informal employment making late transitions harder.

### Robustness and alternative specifications
- Difference-in-differences (DID) estimates (Table 8):
  - 2007 baseline: No BFP = 0.200; BFP = 0.226; Δ = 0.026 (***).
  - 2010: No BFP = 0.233; BFP = 0.283; Δ = 0.050 (***).
  - 2011: No BFP = 0.239; BFP = 0.293; Δ = 0.054 (***).
  - Δ (2007→2010): No BFP Δ = 0.033; BFP Δ = 0.057; DID = 0.024 (***).
  - Δ (2007→2011): No BFP Δ = 0.039; BFP Δ = 0.067; DID = 0.028 (***).
- DID combined with propensity score matching (Table 9):
  - 2007: No BFP = 0.223; BFP = 0.226; Δ = 0.003.
  - 2010: No BFP = 0.263; BFP = 0.283; Δ = 0.020 (***).
  - 2011: No BFP = 0.271; BFP = 0.293; Δ = 0.022 (***).
  - Δ (2007→2010): No BFP Δ = 0.040; BFP Δ = 0.057; DID = 0.017 (***).
  - Δ (2007→2011): No BFP Δ = 0.048; BFP Δ = 0.067; DID = 0.019 (***).
- Robustness conclusion: DID and DID+PSM estimates are positive and statistically significant, supporting main LATE findings.

### Conclusions and policy-relevant implications
- Main empirical finding:
  - Bolsa Família has a positive effect on formal labor market participation.
  - Estimated average treatment effects on the treated (ATT):
    - Beneficiaries are 4.7 percentage points more likely to be found at least once in a formal job in 2010 or 2011 (ATT = 0.047).
    - Beneficiaries are 3.3 percentage points more likely to be found in a formal job throughout 2010–2011 (ATT = 0.033).
  - Effects statistically significant for the whole sample, for men and women separately, and for younger age groups ([25,35) and [35,45)).
- Policy interpretation:
  - Results counter a simple theoretical prediction of negative labor supply effects from cash transfers in contexts with large informal sectors.
  - Positive effects are consistent with transfers helping finance job-search costs and with indirect benefits via improved psychological well-being and decision-making.
  - No detectable impact for older age groups ([45,65]) aligns with scarring effects that hinder late transitions to formal employment.
- Overall implication:
  - Modest conditional cash transfers, when combined with mechanisms that reduce search costs and support human capital accumulation (through conditionalities), can be associated with increased transitions into formal employment, particularly among younger cohorts.

*Italic: Source — wpiea2020099-print-pdf*

### REFERENCES ____________________________________________________________20

### wpiea2020099-print-pdf - REFERENCES ____________________________________________________________20

### Introduction: context, theory, and empirical background
- Cash transfer programs emerged in the mid-1990s in Latin America and have been implemented in almost 70 low- and middle-income countries (World Bank 2017).
- During the Covid-19 crisis many governments are resorting to social protection programs like cash transfer programs to sustain livelihoods.
- Theoretical channels:
  - Transfers may reduce labor supply through an income effect and a high tax rate on marginal earnings (Kanbur et al., 1994; Borjas, 2005). The larger the transfers and the larger the marginal tax rate on earnings, the larger the reduction of labor supply is expected to be (Moffit 2002; Borjas 2005).
  - Transfers could increase labor supply if they relax liquidity and credit constraints and cover costs of job search (Baird, McKenzie and Özler 2018).
  - In contexts with large informal sectors, transfers may reduce formal labor market participation while leaving informal labor participation unchanged because informal income is hard to detect.
- Empirical evidence summarized:
  - Kabeer and Waddington (2015): only one out of eight reviewed studies finds negative impacts on labor force participation.
  - Bastagli et al. (2016): three out of eight studies point to an increase in labor force participation; one points to a reduction.
  - Banerjee et al. (2017): meta-analysis of seven RCTs finds “no significant effect of belonging to a transfer program on employment or hours of work in any of the seven programs”.
  - Bolsa Família Program (BFP) evidence:
    - Oliveira and Soares (2012): five out of eight studies suggest a small positive effect on participation; one finds a negative impact. Small decrease in hours worked observed in three out of five studies, usually for women.
    - de Brauw et al. (2015): “no aggregate effects on labor market participation” and “no effect on total household work hours.”
    - Barbosa and Corseuil (2014): “the program has no impact on the occupational choice of beneficiaries between formal and informal jobs”.
    - Santos et al. (2016): probability of leaving a formal job is between 7 and 10 percent lower for beneficiaries.
- Paper’s contribution:
  - Leverages an unannounced change in the eligibility cutoff and municipal-level rationing of entry to estimate causal impact of Bolsa Família on formal employment.
  - Key finding: participation in Bolsa Família had a positive effect on formal employment, stronger for younger beneficiaries.
  - Interpretation compatible with models incorporating job search costs, credit and liquidity constraints, and decision-making under scarcity (Mullainathan and Shafir 2013). Also consistent with evidence on long-term positive effects of CCTs (Barham, Macours and Maluccio 2018).

### Beneficiary selection and link to the formal labor market
- Administrative and operational features:
  - Ministry of Citizenship (previously Ministry of Social Development – MDS) defines eligibility and authorizes payments.
  - In 2018 over a fifth of Brazil’s population was beneficiary of the program, which had a budget of about BRL 30 billion (equivalent to 0.4 percent of GDP).
  - Eligibility requires registration in the Single Registry for Social Programs (Single Registry, or Cadastro Unico para Programas Sociais, CadUnico).
  - Eligibility criterion: per capita household income as declared in the Single Registry (sum of all incomes of all members divided by number of members).
  - Municipal-level quotas: an estimate of the number of families in poverty or extreme poverty was established for each municipality based on Demographic Census data (Hellman 2015); once the quota is reached, entry becomes more difficult.
- Registration and verification:
  - A family member (generally a woman aged 16 or more) provides detailed household information during Single Registry registration.
  - Families must update Single Registry information at least once every two years or whenever relevant changes occur.
  - Self-declared income is cross-checked with administrative records, including the Annual Social Information Report (RAIS). Income from jobs registered in RAIS is used to determine eligibility; income from informal work is virtually impossible to detect.
- Policy note from source text:
  - Footnote: “Indeed, the introduction of the ‘retorno garantido’ (guaranteed return) in 2012 that ensure a right to reinsertion in the program to families who had willingly left the program to accept a job offer was an acknowledgement of this.”

### Data sources and construction of analytical datasets
- Three administrative records used and how they were constructed/merged:
  - Single Registry:
    - Used by about 30 social programs.
    - Contains information on more than 25 million low-income families (about 70 million people, a third of the population).
    - Covers nearly the totality of the country's poorest population.
    - Version extracted in December 2008 contains approximately 70 million individual registers, yielding 34.3 million people of working age (working age defined as those aged between 18 and 65 (men) or 60 (women)).
  - Historical Database of Benefits of Bolsa Família:
    - Constructed by concatenating 96 monthly payrolls from January 2004 to December 2011 under a computational routine in R.
    - Over the period, number of families in the program increased from 5.9 to 12.3 million.
    - Total of 19.8 million families who entered the program at some point; 7.43 million of which had left by the end of the period.
    - The Historical Database contains exact date of entry and (eventual) exit for each family since program start.
  - RAIS (Annual Social Information Report):
    - Most comprehensive administrative record on formal employment in Brazil; covers public and private sectors.
    - Information provided by employers.
    - In December 2011, RAIS reported 46.3 million formal employment relations.
    - Paper uses six annual RAIS versions from 2006 to 2011.
- Merging:
  - The three datasets were merged using the unique identifier “social identification number” present in all of them.
- Table 1 – Number of families entering and exiting Bolsa Família Program (2004-2011):
  - Entry / Exit
  - 2004: 6,003,602 / 78,573
  - 2005: 2,135,331 / 219,642
  - 2006: 3,480,427 / 1,028,380
  - 2007: 1,180,200 / 937,245
  - 2008: 940,455 / 1,333,643
  - 2009: 2,543,647 / 1,849,878
  - 2010: 1,759,186 / 1,145,725
  - 2011: 1,719,744 / 839,958
  - Total: 19,762,592 / 7,433,044
  - Sources: Authors’ calculation, based on the Single Registry and the Historical Database of Benefits.

### Bolsa Família beneficiaries and formal employment (summary of initial findings and sample characteristics)
- Working-age population in Single Registry (December 2008):
  - 34.3 million people of working age (defined as those aged between 18 and 65 (men) or 60 (women)).
  - Almost half were in the North East Region and were in households with a couple and kids (referenced Figure 1).
- Dataset coverage and scope:
  - Single Registry: ~70 million individual registers.
  - Historical Database: monthly coverage January 2004 to December 2011 (96 monthly payrolls).
  - RAIS: 46.3 million formal employment relations reported in December 2011; six annual files 2006–2011 used.
- Analytical advantage:
  - Exact dates of program entry/exit and linkage to administrative formal employment records enable causal analysis exploiting eligibility cutoff changes and municipal rationing.

*Italic: Source — wpiea2020099-print-pdf - REFERENCES ____________________________________________________________20*

### 1.8 million between 2006 and 2011, from 4.17 million to almost 6.

### wpiea2020099-print-pdf - 1.8 million between 2006 and 2011, from 4.17 million to almost 6.

### Sample, classification, and key descriptive statistics
- Population: working age individuals registered in the Single Registry (December 2008).
- Four groups by Bolsa Familia participation in 2006 and 2011:
  - BB: beneficiaries in both 2006 and 2011 — 35 percent.
  - BN: beneficiaries in 2006 but not in 2011 — 21 percent.
  - NB: not beneficiaries in 2006 but beneficiaries in 2011 — 16 percent.
  - NN: not beneficiaries in 2006 nor in 2011 — 28 percent.
- Table 2 profile (working age individuals registered on December 2008) — selected entries:
  - BB: Number working age individuals = 11,972,421; Living in North & NE = 70.2; Rural = 39.1; W/kids < 16yo = 79.6; Female HH = 58.4; Primary schooling = 49.6; Black = 74.7.
  - BN: Number working age individuals = 7,295,443; Living in North & NE = 45.6; Rural = 26.8; W/kids < 16yo = 64.2; Female HH = 56.8; Primary schooling = 38.7; Black = 65.1.
  - NB: Number working age individuals = 5,503,518; Living in North & NE = 58.9; Rural = 30.9; W/kids < 16yo = 71.7; Female HH = 61.7; Primary schooling = 38.0; Black = 73.1.
  - NN: Number working age individuals = 9,555,750; Living in North & NE = 43.5; Rural = 27.0; W/kids < 16yo = 48.4; Female HH = 56.1; Primary schooling = 36.9; Black = 65.8.
- Labor market participation dynamics 2006–2011:
  - Three quarters of working age individuals in the Single Registry (about 26 million people) never worked in the formal sector between 2006 and 2011.
  - Over 80 percent of BB had no formal employment at any point over the six-year period, compared to 73 percent of NN.
  - In BB, 5 percent had a continuous presence in formal employment; in NN, 12 percent.
  - Growth in formal labor market participation: 68 percent for individuals who were in the program (BB) vs. 30 percent for those who were not (NN).
- Age-specific growth in formal labor market participation (2006–2011):
  - Age 18 and 19: increase of 213 percent (overall); for BB, almost four-fold increase noted.
  - [20,30): 59 percent total; BB 96 percent; NB 48 percent; BN 68 percent; NN 40 percent.
  - [30,40): 26 percent total; BB 44 percent; NB 20 percent; BN 30 percent; NN 15 percent.
  - [40,50): 11 percent total; BB 21 percent; NB 9 percent; BN 14 percent; NN 5 percent.
  - 50+: -17 percent total; BB -11 percent; NB -15 percent; BN -16 percent; NN -20 percent.

### Identification strategy
- Core challenge: unobserved counterfactual outcomes (treatment effect (Yi1 − Yi0) unobservable).
- Regression discontinuity not used due to evidence of manipulation around eligibility cutoff.
- Exploited an unannounced change in eligibility in July 2007: income eligibility cutoff raised from BRL100 per capita to BRL120.
  - Almost 73 thousand families previously ineligible became eligible.
  - Only about one third of them became beneficiaries due to Central Government budget allocation and municipal quotas.
- Treatment group: households that became eligible and entered the program after the change.
- Control group: households that became eligible but did not enter the program.
- Analysis restricted to families that updated information before the announcement (to avoid strategic manipulation).
- Local Average Treatment Effects (LATE) estimated.
- Table 4 (entry after July 2007 change):
  - (100,120]: No BFP = 49,148; BFP = 23,633; Total = 72,781.
  - (120,137]: No BFP = 40,183; BFP = 1,429; Total = 41,612.
  - Total = 89,331 No BFP; 25,062 BFP; 114,393 Total.
- Geographic variation: share of newly eligible that became beneficiaries ranges from 1.3 percent in the Federal District to 48.7 percent in the state of Goiás.
- Checks for correlation between municipal quota levels and growth in formal sector employment: no evidence of correlation found (correlation between -0.02 in 2010-2011 and 0.02 in 2008-2009).
- Estimation approach: propensity score matching (kernel weights) to balance observables between treated and control groups.
  - Matching achieved balance: t-tests cannot reject equal means; chi-squared test cannot reject overall balance.

### Outcomes and main results
- Sample for causal analysis: 72,781 individuals aged 18–65 in 2007 whose household became eligible due to the cutoff change.
- Outcome variables:
  - "formal at least once": appeared at least once in RAIS over a certain number of years.
  - "formal throughout": appeared in RAIS every month throughout a certain number of years.
- Time periods considered: 2007-2011 and 2009-2011; primary reported results focus on presence in RAIS in 2010 and 2011.
- Table 6 — Probability of being formal (No BFP, BFP, Δ LATE) — selected entries:
  - All:
    - Formal at least once in 2010 or 2011: No BFP = 0.29; BFP = 0.33; Δ LATE = 0.047 (Standard Error (0.004)).
    - Formal consecutively between 2010 and 2011: No BFP = 0.22; BFP = 0.25; Δ LATE = 0.033 (Standard Error (0.004)).
  - Women:
    - Formal at least once in 2010 or 2011: No BFP = 0.204; BFP = 0.251; Δ LATE = 0.046 (Standard Error (0.005)).
    - Formal consecutively between 2010 and 2011: No BFP = 0.145; BFP = 0.171; Δ LATE = 0.026 (Standard Error (0.004)).
  - Men:
    - Formal at least once in 2010 or 2011: No BFP = 0.423; BFP = 0.462; Δ LATE = 0.039 (Standard Error (0.006)).
    - Formal consecutively between 2010 and 2011: No BFP = 0.343; BFP = 0.382; Δ LATE = 0.040 (Standard Error (0.006)).
- Age-differentiated LATE (Table 7) — Probability differences (Δ LATE) for being formal in 2010/2011:
  - [18,25): Formal at least once: Δ LATE = 0.048 (Standard Error (0.010)); Formal consecutively: Δ LATE = 0.030 (Standard Error (0.009)).
  - [25,35): Formal at least once: Δ LATE = 0.064 (Standard Error (0.007)); Formal consecutively: Δ LATE = 0.047 (Standard Error (0.007)).
  - [35,45): Formal at least once: Δ LATE = 0.037 (Standard Error (0.007)); Formal consecutively: Δ LATE = 0.025 (Standard Error (0.006)).
  - [45,55): Formal at least once: Δ LATE = 0.009 (Standard Error (0.010)); Formal consecutively: Δ LATE = 0.006 (Standard Error (0.009)).
  - [55,65]: Formal at least once: Δ LATE = 0.011 (Standard Error (0.013)); Formal consecutively: Δ LATE = -0.002 (Standard Error (0.011)).
- Interpretation of magnitude and gender:
  - Positive and significant effects for both men and women.
  - Attachment to formal market limited: share appearing throughout the period substantially smaller, especially for women.
  - Strongest impacts concentrated among younger cohorts, notably [25,35).
- Mechanisms discussed:
  - Cash transfers may help meet job search costs and ease transitions into formal jobs.
  - Indirect psychological benefits (reduced scarcity, improved cognitive processing and decision-making) may improve labor market decisions.

### Robustness checks
- Difference-in-differences (DID) estimates (simple DID, Table 8):
  - 2007 baseline: No BFP = 0.200; BFP = 0.226; Δ = 0.026 (***).
  - 2010: No BFP = 0.233; BFP = 0.283; Δ = 0.050 (***).
  - 2011: No BFP = 0.239; BFP = 0.293; Δ = 0.054 (***).
  - Δ (2007→2010): No BFP Δ = 0.033; BFP Δ = 0.057; DID = 0.024 (***).
  - Δ (2007→2011): No BFP Δ = 0.039; BFP Δ = 0.067; DID = 0.028 (***).
- DID combined with propensity score matching (Table 9):
  - 2007: No BFP = 0.223; BFP = 0.226; Δ = 0.003.
  - 2010: No BFP = 0.263; BFP = 0.283; Δ = 0.020 (***).
  - 2011: No BFP = 0.271; BFP = 0.293; Δ = 0.022 (***).
  - Δ (2007→2010): No BFP Δ = 0.040; BFP Δ = 0.057; DID = 0.017 (***).
  - Δ (2007→2011): No BFP Δ = 0.048; BFP Δ = 0.067; DID = 0.019 (***).
- Robustness conclusion: DID and DID+PSM estimates are positive and statistically significant, supporting main LATE findings.

### Conclusions and policy-relevant implications
- Empirical finding: Bolsa Familia has a positive effect on formal labor market participation.
  - Estimated average treatment effects on the treated (ATT):
    - Beneficiaries are 4.7 percentage points more likely to be found at least once in a formal job in 2010 or 2011 (ATT = 0.047).
    - Beneficiaries are 3.3 percentage points more likely to be found in a formal job throughout 2010–2011 (ATT = 0.033).
  - Effects statistically significant for the whole sample, for men and women separately, and for younger age groups ([25,35) and [35,45)).
- Policy interpretation and mechanisms:
  - Results counter a simple theoretical prediction of negative labor supply effects from cash transfers in contexts with large informal sectors.
  - Positive effects are consistent with cash transfers helping to finance job-search costs and with indirect benefits via improved psychological well-being and decision-making.
  - No detectable impact for older age groups ([45,65]), consistent with "scarring" effects of prolonged informal employment making late transitions to the formal sector harder.
- Overall implication: modest conditional cash transfers, when combined with mechanisms that reduce search costs and support human capital accumulation (through conditionalities), can be associated with increased transitions into formal employment, particularly among younger cohorts.

*Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020099-print-pdf.pdf*

### REFERENCES

### REFERENCES

### Cash transfers and social protection
- Attah, Ramlatu, Valentina Barca, Andrew Kardan, Ian Macauslan, Fred Merttens and Luca Pellerano. Can social protection affect psychosocial wellbeing and why does this matter? Lessons from cash transfers in Sub Saharan Africa. The Journal of Development Studies 52(8), 2016.
- Bastagli, F., J. Hagen-Zanker, L. Harman, V. Barca, G. Sturge, T. Schimidt and L. Pellerano. Cash Transfers: what does the evidence say? A rigorous review of Programme impact and of the role of design and implementation features. London: Overseas Development Institute (ODI), 2016.
- World Bank. Closing the gap: The State of Social Safety Nets 2017. World Bank Group, 2017.
- Lentz, Erin C., Christopher B. Barrett and John Hoddinott. Food Aid and Dependency: Implications for Emergency Food Security Assessments. World Food Programme, Emergency Needs Assessment Branch (ODAN). December 2005.
- Mullainathan, Sendhil and Shafir, Eldar. Scarcity: why having too little means too much. New York: Henry Holt, 2013.

### Effects of cash transfers on labor market outcomes and labor supply
- Baird, Sarah, David McKenzie and Berk Özler. The effects of cash transfers on adult labor market outcomes. IZA Journal of Development and Migration Vol. 8 No. 1, 2018.
- Banerjee, Abhijit, Rema Hanna, Gabriel Kreindler and Benjamin A. Olken. Debunking the Stereotype of the Lazy Welfare Recipient: Evidence from Cash Transfer Programs Worldwide. World Bank Research Observer, Vol.32, Issue 2, August 2017.
- De Brauw, Alan, Daniel Gilligan, John Hoddinott and Shalini Roy. Bolsa Familia and household labor supply. Economic Development and Cultural Change Vol. 63, n° 3, 2015.
- Moffitt, R. A. Welfare programs and labor supply. In: A. J. Auerbach and M. Feldstein (org.). Handbook of Public Economics, Vol. 4. North Holland: Elsevier, 2002.
- Kanbur, Ravi, Michael Keen and Matti Tuomala. Labor Supply and Targeting in Poverty Alleviation Programs. The World Bank Economic Review Vol. 8, n° 2, 1994.
- Borjas, G. J. Labor Economics. New York: McGraw-Hill/Irwin, 2005.
- Oliveira, Luis F. B. and Sergei O. S. Soares. O que se sabe sobre os efeitos das transferências de renda sobre a oferta de trabalho. Texto para Discussão n° 1738. Brasília: IPEA, 2012.
- Soares, Fabio V. Do informal workers queue for formal jobs in Brazil? Texto para Discussão n° 1021. Brasília: IPEA, 2004.
- Barbosa, Ana L. H. and Corseuil, Carlos H. Bolsa Família, escolha ocupacional e informalidade no Brasil. Texto para Discussão No. 1948. Rio de Janeiro: IPEA, 2014.
- Barros, Ricardo P., Mirela de Carvalho, Samuel Franco and Rosane Mendonça. A focalização do Programa Bolsa Família em perspectiva comparada. In: Lucia Modesto & Jorge Castro (eds). Bolsa Família 2003-2010: avanços e desafios. Brasília: Ipea, 2010.
- Barros, Ricardo P., Samuel Franco and Rosane Mendonça. Discriminação e segmentação no mercado de trabalho e desigualdade de renda no Brasil. Texto para Discussão No. 1288. Rio de Janeiro: IPEA, 2007.
- Santos, D. B., Alexandre R. Leichsenring, Naércio Menezes-Filho and Wesley M. da Silva. Os Efeitos do Programa Bolsa Família sobre a Duração do Emprego Formal das Pessoas Pobres. Conference Paper, XL Encontro da Associação Nacional dos Programas de Pós-Graduação e Pesquisa em Administração, September 2016

### Conditional cash transfers, schooling, learning, and earnings
- Barham, Tania, Karen Macours, and John A. Maluccio. Are conditional cash transfers fulfilling their promise? Schooling, learning, and earnings after 10 years. Mimeo, 2017.
- Silva, Joana, Flavio Cireno & Rafael Proença. Improving Learning Outcomes Through Social Assistance: Regression-Discontinuity Evidence from Brazil. Mimeo, 2016.
- Kabeer, Naila, and Hugh Waddington. Economic impacts of conditional cash transfer programmes: a systematic review and meta-analysis. Journal of Development Effectiveness, Vol. 7, No. 3, 2015.
- Hellmann, Aline Gazola. “How does Bolsa Familia Work? Best practices in the Implementation of Conditional Cash Transfers Programs in Latin America and the Caribbean”. Social Protection and Health Division, IDB, Technical Note n. IDB-TN-856, 2015.

### Eligibility manipulation, targeting, and program implementation
- Camacho, A., and E. Conover. Manipulation of social program eligibility. American Economic Journal: Economic Policy Vol. 3 No. 2, 2011.
- Firpo, Sergio, R. Pieria, E. Pedroso Jr. & A. P. Souza. Evidence of eligibility manipulation for conditional cash transfer programs. EconomiA, Vol. 15, No. 3, 2014.
- Camargo, Camila F., Claudia Curralero, Elaine Lício and Joana Mostafa. Perfil socioeconômico dos beneficiários do Programa Bolsa Família: o que o Cadastro Único revela? In: T. Campello & M. Nery. Programa Bolsa Família – Uma década de inclusão e cidadania. Brasília: IPEA, 2013.

### Miscellaneous analyses and reviews
- Barros, Ricardo P., Mirela de Carvalho, Samuel Franco and Rosane Mendonça. A focalização do Programa Bolsa Família em perspectiva comparada. In: Lucia Modesto & Jorge Castro (eds). Bolsa Família 2003-2010: avanços e desafios. Brasília: Ipea, 2010.
- Duncan, Greg J., Kathleen M. Harris and Johanne Boisjoly. Time Limits and Welfare Reform: New Estimates of the Number and Characteristics of Affected Families. Social Service Review, Vol. 74, No. 1, 2000.

*Content from wpiea2020099-print-pdf - REFERENCES*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020099-print-pdf.pdf_