BEYOND LOAN VOLUMES: A MULTI-DIMENSIONAL CAUSAL EVALUATION OF BRAZIL'S MINHA CASA MINHA VIDA PROGRAM (2006-2023)
DOI:
https://doi.org/10.52152/9eptc380Ključne besede:
Housing policy evaluation, causal inference, MCMV, Brazil, interrupted time series, synthetic controlPovzetek
We provide the first multi-dimensional causal evaluation of Brazil's Minha Casa Minha Vida (MCMV) housing program using interrupted time series analysis, synthetic control methods, and state-level differencein- differences. Using Central Bank administrative data (2006-2023) and Ministry of Cities program records, we test four pre-registered hypotheses regarding credit expansion, social targeting, institutional sustainability, and counter-cyclical stabilization. Our event-study analysis shows MCMV increased housing credit by 969.3 billion reais (95% CI: 793.2-1,145.4) with no pre-intervention trends, representing system-wide expansion rather than substitution effects. However, targeting analysis reveals regressive outcomes: Track 1 (poorest families) received 39.4% of units versus 60% target (Kakwani progressivity index = 0.12, below 0.25 benchmark). During the 2014-2016 recession, MCMV lending elasticity to unemployment was -0.34 versus -0.89 for private markets (difference = 0.55, p<0.01), demonstrating counter-cyclical buffering. Institutional sustainability metrics show manageable fiscal exposure (0.78% GDP) with default rates (3.8% Track 1, 2.1% Track 2/3) below international benchmarks. State-level heterogeneity analysis reveals stronger program effects in low pre-program mortgage penetration states (β=0.23, p<0.05), consistent with credit constraintmechanisms. Our findings demonstrate that preferential interest rates generate volume expansion but require complementary institutional design to achieve equity objectives.
Prenosi
Objavljeno
Številka
Rubrika
Licenca
Avtorske pravice (c) 2025 Lex localis - Journal of Local Self-Government

To delo je licencirano pod Creative Commons Priznanje avtorstva-Nekomercialno-Brez predelav 4.0 mednarodno licenco.


