Poverty, NBI, and IPM/TPM reproducibility

Purpose

enemduR includes infrastructure for income poverty, NBI, and IPM/TPM workflows. These workflows are designed for official-source alignment, reproducible local analysis, and official benchmark comparison, not for declaring institutional validation.

This vignette explains how to interpret those workflows at a portfolio level. It uses placeholder paths and environment variables only. It does not include, download, or request official microdata.

Data and authorization

Official ENEMDU microdata and official reference materials must be obtained by the user from authorized sources. Local scripts should read those files from user-controlled paths, not from committed repository data.

For reproducibility scripts, prefer environment variables such as:

Sys.getenv("ENEMDU_PERSONAS_PATH")
Sys.getenv("ENEMDU_HOGARES_PATH")
Sys.getenv("ENEMDU_VIVIENDA_PATH")

These values can point to authorized local files without exposing sensitive or restricted data in the repository.

Income poverty

Income poverty indicators require explicit and auditable poverty lines. The package does not derive poverty lines implicitly, does not ingest CPI series as part of the KPI wrapper, and does not treat poverty as a generic binary recode.

A manual workflow should state the poverty line, the extreme poverty line, the period, and the source note.

microdata <- enemdu_read_data(
  path = Sys.getenv("ENEMDU_PERSONAS_PATH"),
  survey_type = "mensual",
  period = "2025-12"
)

poverty_kpis <- enemdu_kpi_income_poverty(
  data = microdata,
  period = "2025-12",
  mode = "manual",
  poverty_line = 92.40,
  extreme_poverty_line = 52.07,
  line_source = "INEC poverty bulletin, December 2025: poverty line 92.40 and extreme poverty line 52.07.",
  survey_type = "mensual"
)

For the December 2025 reproducibility workflow, the package provides a helper that estimates national and urban-rural outputs and compares them with retained benchmark rows.

poverty_reproducibility <- enemdu_run_poverty_reproducibility(
  data = microdata,
  period = "2025-12",
  survey_type = "mensual",
  strict = FALSE
)

Official benchmark comparison output is an audit aid. It does not establish official institutional validation.

NBI

NBI workflows are profile-specific. The package includes infrastructure for the ENEMDU 2025 annual profile, including component, flag, KPI, benchmark, and local reproducibility helpers.

The NBI reproducibility helper works from final NBI components by default. Those components must be present, audited, and appropriate for the selected profile.

annual_people <- enemdu_read_data(
  path = Sys.getenv("ENEMDU_PERSONAS_ANUAL_PATH"),
  survey_type = "anual",
  period = "2025"
)

nbi_reproducibility <- enemdu_run_nbi_reproducibility(
  data = annual_people,
  period = "2025",
  benchmark_set = "nbi_2025_anual",
  survey_type = "anual",
  component_vars = c("comp1", "comp2", "comp3", "comp4", "comp5"),
  strict = FALSE
)

When person and housing files are delivered separately, a controlled source join should preserve person rows before component derivation.

people_path <- Sys.getenv("ENEMDU_PERSONAS_ANUAL_PATH")
housing_path <- Sys.getenv("ENEMDU_VIVIENDA_ANUAL_PATH")

people <- enemdu_read_data(
  path = people_path,
  survey_type = "anual",
  period = "2025"
)

housing <- enemdu_read_data(
  path = housing_path,
  survey_type = "anual",
  period = "2025"
)

nbi_sources <- enemdu_join_nbi_sources(
  person_data = people,
  household_data = housing
)

This is a reproducibility scaffold. It is local reproducibility evidence, not a claim that local outputs have been officially validated.

IPM/TPM

IPM/TPM workflows are also profile-specific. The current infrastructure is oriented to the ENEMDU 2025 annual profile and includes component, flag, KPI, benchmark comparison, and local reproducibility helpers.

ipm_microdata <- enemdu_read_data(
  path = Sys.getenv("ENEMDU_PERSONAS_ANUAL_PATH"),
  survey_type = "anual",
  period = "2025"
)

ipm_reproducibility <- enemdu_run_ipm_reproducibility(
  data = ipm_microdata,
  period = "2025-12",
  survey_type = "anual",
  by = "area",
  build_components = FALSE,
  build_flags = TRUE,
  missing_component_policy = "complete_case",
  strict = FALSE
)

The workflow can consume already-built score and flag variables, or it can build flags from registered component columns when those inputs are available. It does not make new official validation claims.

Local evidence versus official validation

The repository retains local reproducibility evidence for December 2025 poverty, NBI, and IPM/TPM workflows. That evidence documents how local workflows compare against retained official benchmark rows under explicit inputs and contracts.

This is different from official institutional validation.

Official benchmark comparisons are audit aids. They help reviewers inspect alignment, differences, tolerances, missing inputs, and workflow assumptions. They are not official endorsement and should not be described as official validation unless the relevant official authority explicitly authorizes or confirms that status.

Where package outputs include validation metadata, the appropriate status should remain:

official_validation_status = "not_officially_validated"

unless official validation has actually been authorized, run, and documented.

Practical interpretation

Use these workflows to:

• make assumptions explicit;
• keep poverty lines auditable;
• keep NBI and IPM/TPM profile rules visible;
• compare local estimates against retained benchmark rows;
• document tolerances and differences;
• prepare evidence for review.

Do not use these workflows to:

• reconstruct official status without authorization;
• publish official results from package outputs alone;
• treat benchmark alignment as institutional endorsement;
• hide missing inputs or profile-specific assumptions.

The package provides analytical infrastructure. Review, authorization, and official publication decisions remain outside the package.