Replication code and derived data for: "Trust, Not Ideology: Institutional Credibility and the Climate Tax Gap in Brazil"
Description
This dataset contains the Stata replication package for the article "Trust, Not Ideology: Institutional Credibility and the Climate Tax Gap in Brazil". The article uses a nationally representative survey conducted by the Datafolha Institute in March 2026 (N = 2,004, face-to-face interviews with post-stratification weights) to document a "climate tax gap" in Brazil: 68% of respondents accept higher taxes earmarked for public services, but only 49% accept an equivalent tax for climate adaptation. The analysis indicates that the more important constraint on climate willingness to pay (WTP) is institutional trust rather than political ideology. Causal mediation analysis (Imai–Keele–Tingley) is consistent with a suppression effect, in which the ideology coefficient reverses sign once institutional trust is accounted for (ACME = −0.36); doubly robust treatment-effects estimators (PSM, IPW, AIPW) indicate that the negative effect of perceived federal government inaction on climate WTP (ATT ≈ −7 percentage points) is concentrated among right-leaning respondents and is largely absorbed by the trust channel. Contents: (i) three Stata do-files reproducing all tables and figures in the article and in the Supplementary Material, covering survey-weighted ordered probit models, seemingly unrelated estimation (suest), causal mediation with sensitivity analysis, generalized ordered logit, heteroskedastic ordered probit, bivariate probit, treatment-effects estimation with common-support and balance diagnostics, Best Linear Projection of CATEs, Oster (2019) bounds, a suggestive IV-probit exercise with placebo falsification, and bias-corrected bootstrap; (ii) four estimation logs documenting the full output of every reported estimate; (iii) a README describing the file structure, execution order, and the mapping from each table and figure to the code section that generates it. Important: the raw individual-level survey microdata are not yet included. They were collected by the Datafolha Institute under a contractual agreement that currently does not permit public redistribution of the individual-level records. The author is working with FAPESP and the Datafolha Institute to make an anonymized version of the microdata publicly available in a future version of this dataset; until then, access to the microdata requires authorization from the Datafolha Institute, and reasonable requests can be directed to the corresponding author. The deposited code and logs are sufficient to verify the construction of all variables and every estimate reported in the article. Software: Stata 19; user-written packages psacalc, ivreg2, gologit2, oglm. Funding: São Paulo Research Foundation (FAPESP), grant 2019/16970-5.