Labor indicators in enemduR

Purpose

enemduR is an analytical infrastructure package for reproducible work with Ecuadorian ENEMDU microdata.

This guide documents the initial labor-indicator workflow implemented in:

enemdu_kpi_employment()

The function produces stable, long-format outputs that can be consumed by reports, Quarto documents, dashboards, validation scripts, and institutional analytical pipelines.

This document is written as a user-facing guide. It does not replace the official INEC methodology and it does not change any estimator.

What this module does

The labor module provides a reproducible workflow to:

read official ENEMDU microdata;
build labor flags from the consolidated labor-status variable;
estimate labor totals and labor rates using the survey design;
evaluate representativity and precision;
control official design domains;
compare results against official INEC labor-market tabulations.

The initial labor implementation is intentionally conservative. It uses the consolidated ENEMDU variable:

condact

It does not reconstruct labor status from raw questionnaire variables.

Current methodological scope

The current implementation covers the initial labor-indicator block based on condact.

The confirmed condact categories used by the package are:

Code	Meaning
0	Population younger than 15 years
1	Adequate or full employment
2	Time-related underemployment
3	Income-related underemployment
4	Other non-full employment
5	Unpaid employment
6	Unclassified employment
7	Open unemployment
8	Hidden unemployment
9	Economically inactive population

Sector indicators use:

secemp

when required.

Main function

enemdu_kpi_employment(
  data,
  survey_type,
  group_vars = NULL,
  domain_scope = c("observed", "design")
)

Input data

Official recent ENEMDU microdata are treated operationally as .sav files.

enemdu_read_data() supports:

.sav as the operational primary format for recent official ENEMDU microdata;
.dta as an interoperability format;
.csv as an interoperability format.

Example:

micro_mensual <- enemdu_read_data(
  path = "path/to/enemdu_persona_2026_03.sav",
  survey_type = "mensual"
)

Survey types

The current workflow supports three survey types:

`survey_type`	Official design domains used by the package
`mensual`	National and urban-rural area
`trimestral`	National, urban-rural area, and five main city domains
`anual`	National, urban-rural area, five main city domains, and 24 provinces

Design variables

The package uses the following core survey-design variables:

upm
estrato
fexp

These variables are part of the package-level survey design contract.

Domain scope

enemdu_kpi_employment() includes the argument:

domain_scope = c("observed", "design")

`domain_scope = "observed"`

This is the default behavior.

It keeps all observed analytical domains present in the data after estimation.

Use it when the user wants exploratory or analytical-domain outputs, especially when working with custom groups.

Example:

kpi_observed <- enemdu_kpi_employment(
  data = micro_trimestral,
  survey_type = "trimestral",
  group_vars = "ciudad",
  domain_scope = "observed"
)

`domain_scope = "design"`

This mode estimates using the complete microdata and then filters the output to recognized design domains.

It does not prefilter the microdata before estimation.

This distinction is important because prefiltering the microdata to a domain can degrade or distort the survey design used by the estimator.

Example:

kpi_area <- enemdu_kpi_employment(
  data = micro_mensual,
  survey_type = "mensual",
  group_vars = "area",
  domain_scope = "design"
)

Important distinction: `ciudad` versus `dominio`

The variable ciudad must not be assumed to be equivalent to the official published city-domain tabulations.

For official city-domain benchmark comparison, the package workflow uses:

dominio

This is especially important for quarterly and annual official tabulations that publish the five main cities as official domains.

The benchmark-compared city domains are:

Quito;
Guayaquil;
Cuenca;
Machala;
Ambato.

For official city-domain benchmark comparison:

use dominio;
do not use ciudad as an automatic substitute.

Basic national example

micro_mensual <- enemdu_read_data(
  path = "path/to/enemdu_persona_2026_03.sav",
  survey_type = "mensual"
)

kpi_mensual_nacional <- enemdu_kpi_employment(
  data = micro_mensual,
  survey_type = "mensual",
  domain_scope = "design"
)

kpi_mensual_nacional

Area example

kpi_mensual_area <- enemdu_kpi_employment(
  data = micro_mensual,
  survey_type = "mensual",
  group_vars = "area",
  domain_scope = "design"
)

Quarterly city-domain example

micro_trimestral <- enemdu_read_data(
  path = "path/to/enemdu_persona_2026_I_trimestre.sav",
  survey_type = "trimestral"
)

kpi_trimestral_dominio <- enemdu_kpi_employment(
  data = micro_trimestral,
  survey_type = "trimestral",
  group_vars = "dominio",
  domain_scope = "observed"
)

domain_scope = "observed" is used here because the official city-domain benchmark comparison is performed through the derived dominio variable.

Annual province example

micro_anual <- enemdu_read_data(
  path = "path/to/BDDenemdu_personas_2025_anual.sav",
  survey_type = "anual"
)

kpi_anual_provincia <- enemdu_kpi_employment(
  data = micro_anual,
  survey_type = "anual",
  group_vars = "prov",
  domain_scope = "design"
)

Output structure

The function returns a long-format tibble.

The output includes one row per indicator and domain combination.

Typical fields include:

indicator_id;
indicator_label;
estimate;
survey-design fields;
precision and representativity fields;
domain fields when group_vars is used.

The exact number of columns may evolve as the package adds metadata, but the contract is to keep the output analytical, long-format, and suitable for downstream reporting.

Implemented labor indicator registry

The implemented indicators are documented in:

enemdu_labor_indicator_registry()

Example:

labor_registry <- enemdu_labor_indicator_registry()

labor_registry

The registry documents indicator identifiers, labels, estimator type, universe, required variables, scale, and domain policy.

Official benchmark comparison workflow

The package includes helper functions for reading and comparing official INEC labor-market tabulations:

enemdu_read_official_labor_tabulados()
enemdu_compare_labor_tabulados()

Example:

oficial_mensual <- enemdu_read_official_labor_tabulados(
  path = "path/to/202603_Tabulados_Mercado_Laboral_CSV.zip",
  survey_type = "mensual"
)

comparacion_mensual_nacional <- enemdu_compare_labor_tabulados(
  estimates = kpi_mensual_nacional,
  official = oficial_mensual,
  official_period = "mar-26",
  domain_group = "Nacional",
  domain_label = "Total",
  tolerance_count = 1,
  tolerance_rate = 0.0006,
  strict = TRUE
)

Benchmark comparison tolerances

The benchmark comparison workflow uses:

tolerance_count = 1
tolerance_rate = 0.0006

The rate tolerance is expressed in package scale.

Therefore:

0.0006 = 0.06 percentage points

This tolerance is used because official tabulations publish rounded percentages.

The tolerance is only used for benchmark comparison against published tabulations. It does not change the internal estimator.

Published dash handling

Official tabulations may publish - in some cells.

The benchmark comparison workflow accepts - as a match only under all of the following conditions:

the official value is exactly -;
the package scale is count;
the package estimate is not missing;
the package estimate is equal to zero or within the count tolerance from zero.

This adjusted status is documented as:

match_official_dash_zero

This rule must not be applied globally to rates, coefficients of variation, standard errors, or other non-count measures.

Province label alias

The annual official tabulation may label one province as:

Santo Domingo

The corresponding labelled value in the official microdata is:

Santo Domingo de los Tsáchilas

For benchmark comparison purposes, the accepted alias is:

Santo Domingo -> Santo Domingo de los Tsáchilas

This is a label-alignment rule. It is not a statistical transformation.

Benchmark comparison coverage documented

The initial labor block has documented official benchmark comparisons against official INEC labor-market tabulations for the following cuts and domains. These comparisons are local reproducibility evidence; they do not imply official institutional validation, certification, endorsement, or homologation.

Monthly March 2026

Benchmark-compared domains:

national;
urban area;
rural area.

Quarterly First Quarter 2026

Benchmark-compared domains:

national;
urban area;
rural area;
Quito;
Guayaquil;
Cuenca;
Machala;
Ambato.

Annual 2025

Benchmark-compared domains:

national;
urban area;
rural area;
Quito;
Guayaquil;
Cuenca;
Machala;
Ambato;
24 provinces.

What is not covered yet

The current labor module does not yet cover:

reconstruction of condact from raw questionnaire variables;
benchmark comparison for all historical periods;
benchmark comparison for poverty indicators;
benchmark comparison for income aggregates;
benchmark comparison for non-labor indicators;
publication of official INEC results from package outputs.

Recommended workflow

A safe analytical workflow is:

micro <- enemdu_read_data(
  path = "path/to/official_microdata.sav",
  survey_type = "mensual"
)

labor <- enemdu_kpi_employment(
  data = micro,
  survey_type = "mensual",
  domain_scope = "design"
)

labor_registry <- enemdu_labor_indicator_registry()

For official comparison:

official <- enemdu_read_official_labor_tabulados(
  path = "path/to/official_labor_tabulados.zip",
  survey_type = "mensual"
)

comparison <- enemdu_compare_labor_tabulados(
  estimates = labor,
  official = official,
  official_period = "mar-26",
  domain_group = "Nacional",
  domain_label = "Total",
  tolerance_count = 1,
  tolerance_rate = 0.0006,
  strict = TRUE
)

Portfolio summary

The labor module demonstrates that enemduR is not only a convenience wrapper. It includes:

survey-design-aware estimation;
explicit domain control;
a formal labor-indicator registry;
official-tabulation parsers;
reproducible validation against published INEC outputs, interpreted here as documented benchmark comparison and local reproducibility evidence rather than official institutional validation;
local reproducibility evidence from benchmark comparison against published INEC outputs;
documented handling of publication-format differences;
a clear separation between analytical computation and reporting reconciliation.

This makes the package suitable as an analytical infrastructure component for institutional reporting, Quarto pipelines, dashboards, technical reports, and reproducible statistical review.

Purpose

What this module does

Current methodological scope

Main function

Input data

Survey types

Design variables

Domain scope

domain_scope = "observed"

domain_scope = "design"

Important distinction: ciudad versus dominio

Basic national example

Area example

Quarterly city-domain example

Annual province example

Output structure

Implemented labor indicator registry

Official benchmark comparison workflow

Benchmark comparison tolerances

Published dash handling

Province label alias

Benchmark comparison coverage documented

Monthly March 2026

Quarterly First Quarter 2026

Annual 2025

What is not covered yet

Recommended workflow

Portfolio summary

`domain_scope = "observed"`

`domain_scope = "design"`

Important distinction: `ciudad` versus `dominio`