Banque Nationale De Belgique Sa · EURONEXT:BNB
Banque Nationale De Belgique Sa — Technical Presentation (May 2025)
59 slides from Banque Nationale De Belgique Sa's technical presentation, May 26, 2025
Slide 1: The Macroeconomic Effects of Cash Transfers: Evidence from Brazil
T HE M ACROECONOMIC E FFECTS OF C ASH T RANSFERS : E VIDENCE FROM B RAZIL Arthur Mendes1 Wataru Miyamoto2 Thuy Lan Nguyen3 Steven Pennings1 Leo Feler4 1 World Bank 2 HKU 3 FRBSF 4 Numerator National Bank of Belgium May 26, 2025 Disclaimer: The views expressed here are of the authors and not of SF Fe
Slide 2: Motivation: Cash Transfers
M OTIVATION : C ASH T RANSFERS ▶ A fiscal policy tool used for poverty reduction and countercyclical stimulus ▶ 500m recipients before COVID-19 to 1.4bn in 2020-21 ▶ Important to understand the aggregate impact of transfers ▶ Macro effects on local GDP are theoretically ambiguous → empirics needed ▶
Slide 3: This Paper
T HIS PAPER Question: ▶ What are short-run macro impacts of cash transfers in developing countries on local GDP and employment? ▶ Can a standard open-economy New Keynesian model rationalize the results? Approach: ▶ Estimate the short-run macro effects of Brazil’s Bolsa Familia (BF) Program ▶ World’s
Slide 4: Relative Transfer Multipliers
R ELATIVE T RANSFER M ULTIPLIERS Identification: ▶ Bartik IV approach ▶ Heterogeneous exposure of states to changes in BF at the national level Results Preview: ▶ Large and persistent increase in relative state-level GDP (M ≈ 2 − 3.5) ▶ Mostly driven by non-traded goods. ▶ ↑ informal employment and
Slide 5: Contribution and Related Literature
C ONTRIBUTION AND R ELATED LITERATURE ▶ Contribution: One of the first estimates macro effect of cash transfers on GDP, total employment in developing countries ▶ Advanced countries: ▶ Large literature on (i) government purchase multiplier (ii) MPC from transfers ▶ Pennings (2021) estimates transfer
Slide 6: Brazil's Bolsa Familia Program
B RAZIL’ S B OLSA FAMILIA P ROGRAM ▶ Monthly cash transfers to poor and extremely poor families. Two types: ▶ Basic benefit: extremely poor families receive R$50/month (at start of program) ▶ Variable benefit: poor & extremely poor families receive an extra R$15 per child ▶ Means tested and conditio
Slide 7: Data
D ATA Sample: 2004–2019. Annual macro variables for 27 states. ▶ State gross value added: IBGE ▶ Deflated by state GDP deflator (real GDP) ▶ State formal employment: RAIS ▶ State informal employment: harmonizing PNAD and Census ▶ Definition: No formal labor contract/tax ID/social security contributi
Slide 8: Main Specification
M AIN S PECIFICATION ys,t − ys,t−1 bs,t − bs,t−1 ′ = β0 + ηt + ηs + Xs,t δ + ε s,t (1) ys,t−1 ys,t−1 ▶ (ys,t − ys,t−1 )/ys,t−1 is real gross value added (GVA) growth in state s ▶ (bs,t − bs,t−1 )/ys,t−1 are ∆BF transfers to state s as a share of lagged state GVA ▶ β 0 is the relative transfer multip
Slide 9: Identification Challenges
I DENTIFICATION C HALLENGES ▶ Main concern: Reverse Causality — BF is an anti-poverty program: ▶ Adverse shock (↓ ε s,t ) → more poverty → more transfers to residents of state s ▶ OLS can be downward biased, we use 2SLS Bartik-Style instrument ▶ Secondary concern: Omitted variable bias ▶ Time & stat
Slide 10: Intuition for Instrument
I NTUITION FOR INSTRUMENT ▶ Heterogeneous state exposure to national changes in BF .3 Change in Bolsa Familia, % national/state GDP -.2 -.1 0 .1 .2 2004 2006 2008 2010 2012 2014 2016 2018 Brazil North South 10
Slide 11: Instrumental Variable
I NSTRUMENTAL VARIABLE ▶ Instrument using the state expose shares bs,2004 /ys,2004 γs = bN,2004 /yN,2004 ▶ If national transfers increase 1% of GDP, transfers to state s increase by γs ▶ Identifying assumption: No unobserved factor that is both (i) correlated with national BF transfers in the time s
Slide 12: γs: Differential Exposure of States to National BF
γs : D IFFERENTIAL E XPOSURE OF S TATES TO N ATIONAL BF 8 6 2004 Bolsa Share 4 2 Benchmark exposure = 1 0 RO AC AM RR PA AP TO A PI CE RN PB PE AL SE BA G ES RJ SP PR SC RS S T GO DF M M M M N NE MW SE S 95% CI ▶ First stage γs in poorer areas in NE and N are larger than S and SE ▶ Very relevant ins
Slide 13: Relative Output Multiplier
R ELATIVE O UTPUT M ULTIPLIER (1) (2) (3) (4) (5) (6) (7) (8) Dependent variable: A. Output growth B. Non-tradable output C.Tradable output Parsimonious Benchmark OLS 2SLS OLS 2SLS Parsimonious Benchmark Parsimonious Benchmark GDP multiplier 2.37 2.00 3.06 3.22 4.03 3.03 -0.73 1.04 Clustered SEs (st
Slide 14: Comparison with Literature: Output Multipliers
C OMPARISON WITH L ITERATURE : O UTPUT M ULTIPLIERS Our paper: BF output multiplier M = 2 − 3.5 ▶ Nakamura & Steinsson (2014): M = 1.5 defense spending in US states ▶ Pennings (2021): M = 1.5 permanent social security transfer across US states (M = 1/3 for temporary transfer) ▶ Egger et al (2022): M
Slide 15: Cumulative Multipliers
C UMULATIVE M ULTIPLIERS ▶ Estimate: ∑hj=0 ys,t+yj −ys,t−1 = β h ∑hj=0 bs,t+yj −bs,t−1 + ηs + ηt s,t−1 s,t−1 (1) (2) (3) (4) (5) (6) (7) (8) (7) (8) Dependent variable: Contemporaneous 2 years 3 years 4 years 5 years Pars Bench Pars Bench Pars Bench Pars Bench Pars Bench GDP multiplier 2.00 3.22 1.1
Slide 16: Output per Capita and Population
O UTPUT PER CAPITA AND P OPULATION (1) (2) (3) (4) Dependent variable: Output Per Capita Population Pars Bench Pars Bench Multiplier 1.16 2.96 0.57 0.50 Clustered SEs (state) (1.06) (1.35)** (0.14)*** (0.17)*** Wild bootstrap 90% CI Clustered by states & years [-0.9, 3.7] [1.5, 4.5] [0.3, 0.9] [0.3,
Slide 17: Labor Market: Specification of Jobs Multiplier
L ABOR M ARKET: S PECIFICATION OF J OBS M ULTIPLIER j j Emps,t − Emps,t−1 j (bs,t − bs,t−1 ) /R$100, 000 = β0 + ηs + ηt + es,t EmpFWE s,t−1 EmpFWEs,t−1 ▶ For all j ∈ {Formal, Informal, FWE} ▶ Informal employment is almost half of total employment ▶ Formal-wage equivalent empl. (Winf /Wf = 0.5) : FWE
Slide 18: Labor Market
L ABOR M ARKET (1) (2) (3) (4) (5) (6) Dependent variable: Formal Employment Informal Employment FWE Employment Pars Bench Pars Bench Pars Bench Jobs per R$100K -2.91 -1.37 4.73 4.16 -0.73 0.57 Clustered SEs (state) (0.88)*** (1.40) (1.58)*** (1.96)** (1.20) (1.52) Wild bootstrap 90% CI Clustered by
Slide 19: Labor Market in Different Sectors
L ABOR M ARKET IN DIFFERENT S ECTORS (3) (4) (5) (6) Dependent variable: Tradable Employment Non-tradable Employment Pars Bench Pars Bench 2SLS 2SLS 2SLS 2SLS Formal Employment -0.97 0.02 -1.89 -1.86 Clustered SEs (state) (0.28)*** (0.26) (0.88)** (1.14) Wild bootstrap 90% CI Clustered by states & y
Slide 20: Multipliers Vary with Economic Slack
M ULTIPLIERS VARY WITH E CONOMIC S LACK Output Multiplier FWE Labor Productivity 10 12 8 10 8 8 6 6 6 4 4.1 4.8 4 2 4 3.8 3.2 0 -0.2 3.3 2 2.2 -2 2.5 -2.9 2 -4 0 -6 -2 -8 0 Bench High unemp Low unemp Bench High unemp Low unemp Bench High unemp Low unemp Total Employment Formal Employment Informal Em
Slide 21: Robustness Check: Additional Controls
R OBUSTNESS CHECK : A DDITIONAL C ONTROLS (1) (2) (3) (4) (5) (6) (7) (8) (9) Additional controls to benchmark Dependent variable: Ramey 2021 Lags Commodity price Min wage Other fed transfers Tax revenues Lagged state VA gr. Population Poverty ALL GDP multiplier 2.25 2.56 2.48 3.35 3.55 3.44 3.40 2.
Slide 22: Robustness Check: Other IVs
R OBUSTNESS CHECK : O THER IV S (2) (4) (6) (8) Instrument: Baseline (2004 Bolsa share) Estimated Gamma 2019 Bolsa Share 2004-2019 Avg Bolsa Share Benchmark Benchmark Benchmark Benchmark GDP Multiplier 3.22 4.14 5.51 4.57 Clustered SEs (state) (1.35)** (1.51)*** (1.69)*** (1.39)** Wild bootstrap 90%
Slide 23: Robustness Check: Prices
R OBUSTNESS CHECK : P RICES (1) (2) (3) (4) Dependent variable: GDP Deflator Inflation Services Inflation Pars Bench Pars Bench Inflation 2.45 5.45 -0.68 1.99 Clustered SEs (state) (2.23) (3.49) (2.36) (3.03) Bootstrap 90% CI Clustered by states & years [-1.6, 5.7] [1.4, 9.5] [-4.0, 2.8] [-1.9, 6.0]
Slide 24: Anticipation Effects
A NTICIPATION E FFECTS ys,t − ys,t−1 bs,t − bs,t−1 b − bs,t+h−1 = β0 + β h s,t+h + ηs + ηt + es,t (2) ys,t−1 ys,t−1 ys,t−1 ▶ No anticipation effects for output. 24
Slide 25: Interpretation of the Results
I NTERPRETATION OF THE R ESULTS Can open-economy New Keynesian model rationalize the empirical findings? ▶ Pennings (2021), and Nakamura and Steinsson (2014) explain evidence in the US using Gali and Monacelli (2005) type model ▶ Do calibrated standard NK models need additional features for developi
Slide 26: Households: Ricardian
H OUSEHOLDS : R ICARDIAN ▶ Asset markets are incomplete ▶ A share of Ricardian HHs can borrow and save through a risk-free bond, maximizing ∞ t E0 ∑ ∏ β k U CRt , LRt t=0 k =0 ▶ Budget constraint: 2 Bt + 1 ϕB Bt + 1 B Pt CRt + ∗ + Pt − = Bt + WtL LRt + TRRt + ΠRt Rt 2 Pt P ss ▶ Consume both tradable
Slide 27: Households: Hand-to-Mouth
H OUSEHOLDS : H AND - TO - MOUTH ▶ The rest are hand-to-mouth, cannot save/borrow, and face financial stress ∞ E0 ∑ βt U CN N N t , Lt + D Mt t=0 ▶ Budget constraint: t = TRt + Wt Lt + Πt Pt CN N L N N Pt MNt = TRt + Wt Lt + Πt N L N N N 1 N N N N N D Mt = D exp −ϕD N Mt − Mss −D Css ▶ Financial str
Slide 28: The Rest of the Model
T HE REST OF THE MODEL ▶ Calvo-type price and wage rigidities ▶ Production function for firm i: YtT (i) = ATt LtA,T (i)αL and YtS (i) = ASt LtA,S (i)αL ▶ Monetary policy: standard Taylor rule R̂Bt = ρR R̂Bt−1 + (1 − ρR ) ϕπ π̂tB + ϕY ŶtB 28
Slide 29: Fiscal Policy: Transfer
F ISCAL P OLICY: T RANSFER ▶ Home transfer per output TRt TRRt TRN = (1 − ω ) +ω t Yss Yss Yss ▶ Only non-Ricardian HHs receive transfers ▶ Transfer shocks TRt = trt Pt Yss tr ˆ t−1 + etr ˆ t = ρtr tr t ▶ Transfers can be financed by foreign Ricardian HHs 29
Slide 30: Calibration
C ALIBRATION ▶ Discount rate β = 0.99 ▶ Inverse IES parameter σ = 2 ▶ Inverse Frisch elasticity ψ = 1 ▶ Home region size n = 1/27 ▶ Tradable share in C αC = 0.5 ▶ Home bias in tradable αT = 0.5 ▶ Elasticity of tradable and non tradable θC = 2 ▶ Elasticity of home and foreign goods θT = 2 ▶ Price and
Slide 31: Model Prediction
M ODEL PREDICTION 2.5 2 1.5 1 0.5 0 -0.5 1 2 3 4 5 Year Sectors 31
Slide 32: Discussion: Mechanism
D ISCUSSION : M ECHANISM ▶ Standard NK model explains some of the observed relative multipliers ▶ Can be larger if regions are less open ▶ Wealth effects on labor supply small/positive in micro-development literature (Baird et al 2018, Banerjee et al. 2020, Kaur et al 2022) ▶ Wealth effects reduce m
Slide 33: Conclusion
C ONCLUSION ▶ BF cash transfers have large and persistent impacts on relative state output. ▶ Impact multiplier is about 2.3 ▶ Employment effects are near zero as informal employment increases but formal falls ▶ An open-economy NK model has lower multipliers than data at all horizons ▶ Financial str
Slide 34: Extra Slides
EXTRA SLIDES 34
Slide 35: Model Prediction - Sectors
M ODEL PREDICTION - S ECTORS 1.5 1.5 1 1 0.5 0.5 0 0 1 2 3 4 5 1 2 3 4 5 Year Year Back 35
Slide 36: Multipliers with Slack - Old Result
M ULTIPLIERS WITH S LACK - OLD RESULT ys,t − ys,t−1 bs,t − bs,t−1 = β ū × 1st + β u × (1 − 1st ) ū ū + ηs + ηt + ε s,t , ys,t−1 ys,t−1 (1) (2) (3) (4) (5) Output Formal employment Informal employment FWE Labor productivity Low unemployment 1.70 3.56 7.68 3.34 3.24 (1.08) (2.22) (3.82)** (3.18) (1
Slide 37: Cumulative Multipliers
C UMULATIVE M ULTIPLIERS h ys,t+j − ys,t−1 h bs,t+j − bs,t−1 ∑ ys,t−1 = βh ∑ ys,t−1 + ηs + ηt + ε s,t , j=0 j=0 (1) (2) (3) (4) (5) Horizon: Contemporaneous 1 year 2 years 3 years 4 years 2.30 2.29 1.92 1.81 1.33 Cumulative Multipliers (1.04)** (1.11)** (1.22) (1.31) (1.39) State & year fixed effect
Slide 38: Labor Income
L ABOR I NCOME (1) (2) (3) Dependent variable: A. Total B. Formal C. Informal Labor income 2.94 1.18 1.76 (0.60)*** (0.44)*** (0.36)*** State & Year FEs YES YES YES Observations 425 425 425 First-stage statistics 82.4 82.4 82.4 Back 38
Slide 39: Alternative Instruments: Calibrated Exposure
A LTERNATIVE I NSTRUMENTS : C ALIBRATED E XPOSURE ▶ Classic Bartik Instrument has calibrated (not estimated) exposure in year T ▶ Alternative first stage: bs,ty−bs,t−1 = ϕ(ψs bN,ty−bN,t−1 ) + ... + ε s,t with exposure s,t−1 N,t−1 b b based on rel. size of BF transfers (share of GDP) in base year T:
Slide 40: Placebo Tests
P LACEBO T ESTS ▶ Non-zero multiplier might indicate spurious relationship (1) (2) (3) (4) Dependent variable: A. Output multiplier B. Formal employment Benchmark Placebo Benchmark Placebo (2004-2019) (1988-2003) (2004-2019) (1988-2003) Multiplier 2.30 0.43 4.68 0.24 (1.04)** (0.92) (1.94)** (1.67)
Slide 41: Anticipation
A NTICIPATION ys,t+h − ys,t−1 bs,t − bs,t−1 = βh + ηs + ηt + ε s,t , ys,t−1 ys,t−1 Back 41
Slide 42: GDP Deflator Inflation
GDP DEFLATOR INFLATION (1) (2) (3) (4) Dependent variable: Total Non-tradables OLS 2SLS OLS 2SLS GDP deflator -0.8 -0.4 1.60 1.52 (2.58) (3.10) (2.77) (3.04) State & year FEs YES YES YES YES Observations 421 421 415 415 First-stage statistics 10.3 12.4 Back 42
Slide 43: Bolsa Familia Program
B OLSA FAMILIA P ROGRAM ▶ Federally funded program, but joint federal-municipal implementation HHs,t Benefitss,t Bs,t = Quotas,t × × Quota Quotas,t HHs,t Coverage Average benefit per HH where Bs,t is the Bolsa Familia transfer benefit in state s at time t. ▶ Quotas: Max # of beneficiaries’ enrollmen
Slide 44: Quotas
Q UOTAS ▶ Set at the federal level using state & municipal based census and HH survey (PNAD) ▶ Updated infrequently with lag 2-3 years ▶ In 2004: based on 2001 PNAD ▶ In 2006-7: revised based on 2004 PNAD ▶ In 2009: revised based on new poverty map estimate with 2006 PNAD ▶ In 2012: revised based on
Slide 45: Coverage
C OVERAGE ▶ Municipals may enroll up to quota ▶ Quota takes time to implement, so coverage lags quota ▶ Coverage is based on rolling questionnaires ▶ More endogenous - can respond quickly to change in local poverty ▶ Rules change over time: ▶ Income threshold (2006) ▶ 2 years above income ceiling (2
Slide 46: Average Benefits
AVERAGE B ENEFITS ▶ From R$80 to 130 between 2007–2014 ▶ 2011: 50% rise in children’s benefit ▶ Mostly exogenous ▶ 2012: Benefits to overcome poverty ▶ Poverty gap payment of difference between income and poverty line ▶ 2012 implement likely exogenous (based on income level, not change) ▶ 2013+ vari
Slide 47: Brazil Economy Overview
B RAZIL ECONOMY OVERVIEW Output growth BF transfer Average BF Formal employment Informal employment Poverty rate Population growth rate (% of output) growth rate growth rate (% of Pop) growth rate National 2.23 3.40 1.80 3.14 0.95 7.51 1.09 (3.0) (2.3) (1.5) (3.5) (4.0) (0.4) (1.2) North 3.24 5.3 2.
Slide 48: Model Intuition
Cumulative multiplier MODEL INTUITION Targeting & 1/3 Htm «s@+ Flex wage & no targeting Years after shock 48
Slide 49: Model Main Elements
M ODEL M AIN E LEMENTS ▶ Ricardian households (j ∈ [ω, 1]) ∞ " # L′ (j )1+ φ E0 ∑ βt ln ct′ (j) − t t=0 1+φ Pt ct′ (j) = Ph,t bt′ (j) − Rt−1 Ph,t−1 bt′ −1 (j) + Wt (j) Lt′ (j) + Ph,t Trt′ + Ph,t Πt′ + χt ▶ Hand-to-mouth households (j ∈ [0, ω ]) ∞ " # L ′′ (j)1+ φ E0 ∑ βt ln ct′′ (j) − t t=0 1+φ Pt c
Slide 50: Model Main Elements
M ODEL M AIN E LEMENTS ▶ Consumption: " θ −1 θ −1 # θ θT−1 1 ′ Tθ 1 ′ Tθ T ct′ = α θT ch,t T + (1 − α ) θT cf ,t T ct = (1 − ω ) ct′ + ωct′′ ▶ Resource constraint 1−n ⋆ Yh,t = ch,t + c + gh,t n h,t Back 50
Slide 51: Model Main Elements
M ODEL M AIN E LEMENTS ▶ Monetary policy h i R b t−1 + (1 − ρr ) ϕπ π b t = ρr R bt + ϕy Y bt ▶ Fiscal policy ▶ Transfers given to hand-to-mouth housdholds in home state ▶ Transfers financed by Ricardian households in foreign state Back 51
Slide 52: Model Main Elements
M ODEL M AIN E LEMENTS ▶ Sticky prices: πbh,t = βEt π bh,t+1 + κ m cch,t Z 1 σX −1 σ σX−1 X Yh,t = Yh,t (i) σX di 0 Yh,t (i) = Lh,t (i) Z 1 Lh,t = Lh,t (i) di 0 ▶ Sticky wages bw,t = βEt π π bw,t+1 − λb µt Z 1 ϵ ϵw−1 w 1− ϵ1w Lt = Lt ( j ) dj 0 Back 52
Slide 53: Back-of-the-Envelope Relative Output Multiplier
B ACK - OF - THE - ENVELOPE R ELATIVE O UTPUT M ULTIPLIER ▶ Literature uses simple rule of thumb to transform formal employment multipliers into output multipliers ▶ Naive calculation based on simple production function approach with only formal employment M = 1.5 ▶ Ad-hoc adjustment for informality
Slide 54: Bolsa Familia Program
B OLSA FAMILIA P ROGRAM Back 54
Slide 55: Calibration
C ALIBRATION ▶ Mostly standard ▶ Discount rate: 0.99 ▶ Home bias: 0.69 ▶ Frisch labor elasticity: 1 ▶ ES between home and foreign: 2 ▶ HtM share: 0.54 (Bracco et al. (2022) for Brazil) ▶ Sticky prices/wages: 0.75 (in quarterly frequency) Back 55
Slide 56: Unemployment in Brazil
U NEMPLOYMENT IN B RAZIL Back 56
Slide 57: Unemployment in the Northeast
U NEMPLOYMENT IN THE N ORTHEAST Back 57
Slide 58: Estimates for Years with Policy Reforms
E STIMATES FOR YEARS WITH POLICY REFORMS (1) (2) (3) Dependent variable: Total output Formal employment FWE employment Multipliers 3.92 0.36 9.09 (1.44)*** (2.34) (4.67)* State & year fixed effects YES YES YES Observations 240 236 238 First-stage F statistics 93.09 79.68 71.14 Notes: Clustered stand
Slide 59: Alternative Standard Errors - Output Multiplier
A LTERNATIVE S TANDARD E RRORS - O UTPUT M ULTIPLIER A. Output growth B. Non-tradable output C. Tradable output Parsimonious Benchmark OLS 2SLS OLS 2SLS OLS 2SLS OLS 2SLS GDP multiplier 2.07 2.30 2.90 3.14 2.07 2.30 -0.35 -0.35 Clustered SEs by state (1.12)* (1.04)** (1.68)* (1.55)** (1.12)* (1.04)*
Banque Nationale De Belgique — Technical Presentation (May 2025) on 6ix