5 Access Data

Published

July 20, 2026

5.1 Overview

Approved files are distributed through datalibweb, in two collections:

Collection	Content
`FDPRAW`	Raw survey microdata, questionnaires and documentation
`FDP`	Harmonized microdata (50by35 schema) and harmonization programs

The harmonized welfare module in the FDP collection is called WELF.

5.2 Who can access data

To access the data, users must be registered on datalibweb and have permission to access the FDPRAW and FDP collections. Access is set to private by default on datalibweb. Each user needs to subscribe to access the datasets they need and be approved by admin. Admin can change access to public, so that all (internal) datalibweb users can access it without user level subscription.

5.3 File naming conventions

datalibweb survey identifiers guarantee replicability: versions are never replaced — new releases get a new version number, and previous vintages remain available. Always record the full survey ID used in an analysis.

Raw (master) data in FDPRAW:

CCC_YYYY_SSSS_V##_M

CCC — 3-letter ISO/WDI country code (e.g. COL)
YYYY — year data collection started
SSSS — survey acronym (e.g. GEIH, EHCVM, RHS)
V##_M — version of the master (original) data; first release is V01

Harmonized data in FDP:

CCC_YYYY_SSSS_V##_M_V##_A_FDP

V##_A — version of the harmonization (adaptation), independent of the master version
FDP — the collection of harmonized surveys

Module files append the module name to the survey ID. The 50by35 harmonized welfare file (V01) for Colombia is:

COL_2023_GEIH_V01_M_V01_A_FDP_WELF.dta

5.4 Using datalibweb

Two clients are available: the datalibweb Stata package and the dlw R package. Both require the World Bank intranet/VPN, permission to access the collections, and the same datalibweb token used by PRIMUS (valid 30 days, renewed on the datalibweb website).

dlw_set_token("<your datalibweb token>")   # stored in the system keyring

datalibweb, token(<your datalibweb token>) version(2)

5.5 Checking the catalog

R
Stata

library(dlw)

# all collections (servers) you can access
dlw_server_catalog()

# what is in a collection
dlw_server_catalog("FDP")
dlw_server_catalog("FDPRAW")

# everything available for one country, across collections
dlw_country_catalog("COL")

# files for one country within a collection
dlw_server_inventory("COL", server = "FDP")

* interactive browser: click through region -> country -> collection
datalibweb

* list what is available for a country-year in a collection; when raw
* data is requested, datalibweb lists the individual files to pick from
datalibweb, country(COL) years(2023) type(FDP) clear

5.6 Loading data

R
Stata

library(dlw)

# harmonized 50by35 welfare module (WELF) from the FDP collection

welf <- dlw_get_data(
  country_code = "COL",
  filename     = "COL_2023_GEIH_V01_M_V01_A_FDP_WELF.dta",
  server       = "FDP",
  year         = 2023
)

# a raw file from the FDPRAW collection; year and survey (acronym) disambiguate
# the specific survey vintage — the R equivalent of Stata's surveyid()
raw <- dlw_get_data(
  country_code = "COL",
  year         = 2023,
  survey       = "GEIH",
  filename     = "individual_data_2023.dta",
  server       = "FDP"
)

* harmonized 50by35 welfare module (WELF) from the FDP collection;
* the latest master and harmonization versions are loaded by default
datalibweb, country(COL) years(2023) type(FDP) mod(WELF) clear

* pin specific vintages for replicability
dlw, country(COL) years(2023) type(FDP) mod(WELF) verm(01) vera(01) survey(GEIH) clear

* a raw file from the FDPRAW collection
datalibweb, country(COL) years(2023) type(FDPRAW) ///
    surveyid(COL_2023_GEIH_V01_M) filename(individual_data_2023.dta) clear

5.7 Calculating SR indicators

R
Stata

library(dlw)
library(pipr)

# harmonized 50by35 welfare module (WELF) from the FDP collection

welf <- dlw_get_data(
  country_code = "COL",
  filename     = "COL_2023_GEIH_V01_M_V01_A_FDP_WELF.dta",
  server       = "FDP",
  year         = 2023
)

# CPI and PPP conversion factors from PIP (public)
cpi_pip <- get_aux("cpi", ppp_version = 2021)
ppp_pip <- get_aux("ppp", ppp_version = 2021)

# calculate SR indicator ----

ccode <- welf$code[1]
yr    <- welf$year[1]

# CPI: ratio normalized to 1 in the PPP base year (2021), for the survey year
cpi_value <- cpi_pip$value[cpi_pip$country_code == ccode &
                             cpi_pip$year == yr &
                             cpi_pip$data_level == "national"]

# PPP: ICP 2021 conversion factor (LCU per international $)
ppp_value <- ppp_pip$value[ppp_pip$country_code == ccode &
                             ppp_pip$data_level == "national"]

# poverty line in $/day 2021 PPP (placeholder — pending confirmation)
zline <- 3.00

# self-reliance share: weighted proportion above the line using welfare_self
sr_share <- weighted.mean(
  welf$welfare_self / 365 / cpi_value / ppp_value >= zline,
  welf$weight
)

# poverty headcount: weighted proportion below the line using welfare
headcount <- weighted.mean(
  welf$welfare / 365 / cpi_value / ppp_value < zline,
  welf$weight
)

sprintf("%s %d - Self-reliance share: %.4f  Headcount: %.4f ($%.2f/day 2021PPP)", ccode, yr, sr_share, headcount, zline)

* harmonized 50by35 welfare module (WELF) from the FDP collection;
* the latest master and harmonization versions are loaded by default
datalibweb, country(COL) years(2023) type(FDP) mod(WELF) clear

* calculate SR indicator
local ccode = code[1]
local year  = year[1]

* CPI and PPP conversion factors from PIP (public)
preserve
pip tables, table(cpi) ppp_year(2021) clear
keep if country_code=="`ccode'" & year==`year' & data_level=="national"
local cpival = value[1]
restore

preserve
pip tables, table(ppp) ppp_year(2021) clear
keep if country_code=="`ccode'" & data_level=="national"
local pppval = value[1]
restore

* poverty line in $/day 2021 PPP (placeholder — pending confirmation)
local zline 3.00

* self-reliance share: weighted proportion above the line using welfare_self
gen byte _sr = (welfare_self/365/`cpival'/`pppval') >= `zline'
qui sum _sr [aw=weight]
local sr : di %9.6f r(mean)

* poverty headcount: weighted proportion below the line using welfare
gen byte _poor = (welfare/365/`cpival'/`pppval') < `zline'
qui sum _poor [aw=weight]
local hc : di %9.6f r(mean)

drop _sr _poor
di "`ccode' `year' - Self-reliance share: `sr'  Headcount: `hc' ($`zline'/day 2021PPP)"

For citation guidance and further options (e.g. PPP conversion on load), see the datalibweb help file in Stata and the datalibweb guidelines; questions to the datalibweb team at datalibweb@worldbank.org.

--- title: "Access Data" date: today --- ## Overview Approved files are distributed through **datalibweb**, in two collections: | Collection | Content | |---|---| | `FDPRAW` | Raw survey microdata, questionnaires and documentation | | `FDP` | Harmonized microdata (50by35 schema) and harmonization programs | The harmonized welfare module in the `FDP` collection is called **`WELF`**. ## Who can access data To access the data, users must be registered on datalibweb and have permission to access the `FDPRAW` and `FDP` collections. Access is set to private by default on datalibweb. Each user needs to subscribe to access the datasets they need and be approved by admin. Admin can change access to public, so that all (internal) datalibweb users can access it without user level subscription. ## File naming conventions datalibweb survey identifiers guarantee replicability: versions are **never replaced** — new releases get a new version number, and previous vintages remain available. Always record the full survey ID used in an analysis. **Raw (master) data** in `FDPRAW`: ``` CCC_YYYY_SSSS_V##_M ``` - `CCC` — 3-letter ISO/WDI country code (e.g. `COL`) - `YYYY` — year data collection started - `SSSS` — survey acronym (e.g. `GEIH`, `EHCVM`, `RHS`) - `V##_M` — version of the **master** (original) data; first release is `V01` **Harmonized data** in `FDP`: ``` CCC_YYYY_SSSS_V##_M_V##_A_FDP ``` - `V##_A` — version of the harmonization (**adaptation**), independent of the master version - `FDP` — the collection of harmonized surveys **Module files** append the module name to the survey ID. The 50by35 harmonized welfare file (V01) for Colombia is: ``` COL_2023_GEIH_V01_M_V01_A_FDP_WELF.dta ``` ## Using datalibweb Two clients are available: the [`datalibweb` Stata package](https://github.com/worldbank/datalibweb) and the [`dlw` R package](https://github.com/worldbank/dlw). Both require the World Bank intranet/VPN, permission to access the collections, and the same datalibweb token used by PRIMUS (valid 30 days, renewed on the datalibweb website). ### Authentication Register the datalibweb token (from the datalibweb website) once every 30 days: ::: {.panel-tabset group="language"} ## R ```r dlw_set_token("<your datalibweb token>") # stored in the system keyring ``` ## Stata ```stata datalibweb, token(<your datalibweb token>) version(2) ``` ::: ## Checking the catalog ::: {.panel-tabset group="language"} ## R ```r library(dlw) # all collections (servers) you can access dlw_server_catalog() # what is in a collection dlw_server_catalog("FDP") dlw_server_catalog("FDPRAW") # everything available for one country, across collections dlw_country_catalog("COL") # files for one country within a collection dlw_server_inventory("COL", server = "FDP") ``` ## Stata ```stata * interactive browser: click through region -> country -> collection datalibweb * list what is available for a country-year in a collection; when raw * data is requested, datalibweb lists the individual files to pick from datalibweb, country(COL) years(2023) type(FDP) clear ``` ::: ## Loading data ::: {.panel-tabset group="language"} ## R ```r library(dlw) # harmonized 50by35 welfare module (WELF) from the FDP collection welf <- dlw_get_data( country_code = "COL", filename = "COL_2023_GEIH_V01_M_V01_A_FDP_WELF.dta", server = "FDP", year = 2023 ) # a raw file from the FDPRAW collection; year and survey (acronym) disambiguate # the specific survey vintage — the R equivalent of Stata's surveyid() raw <- dlw_get_data( country_code = "COL", year = 2023, survey = "GEIH", filename = "individual_data_2023.dta", server = "FDP" ) ``` ## Stata ```stata * harmonized 50by35 welfare module (WELF) from the FDP collection; * the latest master and harmonization versions are loaded by default datalibweb, country(COL) years(2023) type(FDP) mod(WELF) clear * pin specific vintages for replicability dlw, country(COL) years(2023) type(FDP) mod(WELF) verm(01) vera(01) survey(GEIH) clear * a raw file from the FDPRAW collection datalibweb, country(COL) years(2023) type(FDPRAW) /// surveyid(COL_2023_GEIH_V01_M) filename(individual_data_2023.dta) clear ``` ::: ## Calculating SR indicators ::: {.panel-tabset group="language"} ## R ```r library(dlw) library(pipr) # harmonized 50by35 welfare module (WELF) from the FDP collection welf <- dlw_get_data( country_code = "COL", filename = "COL_2023_GEIH_V01_M_V01_A_FDP_WELF.dta", server = "FDP", year = 2023 ) # CPI and PPP conversion factors from PIP (public) cpi_pip <- get_aux("cpi", ppp_version = 2021) ppp_pip <- get_aux("ppp", ppp_version = 2021) # calculate SR indicator ---- ccode <- welf$code[1] yr <- welf$year[1] # CPI: ratio normalized to 1 in the PPP base year (2021), for the survey year cpi_value <- cpi_pip$value[cpi_pip$country_code == ccode & cpi_pip$year == yr & cpi_pip$data_level == "national"] # PPP: ICP 2021 conversion factor (LCU per international $) ppp_value <- ppp_pip$value[ppp_pip$country_code == ccode & ppp_pip$data_level == "national"] # poverty line in $/day 2021 PPP (placeholder — pending confirmation) zline <- 3.00 # self-reliance share: weighted proportion above the line using welfare_self sr_share <- weighted.mean( welf$welfare_self / 365 / cpi_value / ppp_value >= zline, welf$weight ) # poverty headcount: weighted proportion below the line using welfare headcount <- weighted.mean( welf$welfare / 365 / cpi_value / ppp_value < zline, welf$weight ) sprintf("%s %d - Self-reliance share: %.4f Headcount: %.4f ($%.2f/day 2021PPP)", ccode, yr, sr_share, headcount, zline) ``` ## Stata ```stata * harmonized 50by35 welfare module (WELF) from the FDP collection; * the latest master and harmonization versions are loaded by default datalibweb, country(COL) years(2023) type(FDP) mod(WELF) clear * calculate SR indicator local ccode = code[1] local year = year[1] * CPI and PPP conversion factors from PIP (public) preserve pip tables, table(cpi) ppp_year(2021) clear keep if country_code=="`ccode'" & year==`year' & data_level=="national" local cpival = value[1] restore preserve pip tables, table(ppp) ppp_year(2021) clear keep if country_code=="`ccode'" & data_level=="national" local pppval = value[1] restore * poverty line in $/day 2021 PPP (placeholder — pending confirmation) local zline 3.00 * self-reliance share: weighted proportion above the line using welfare_self gen byte _sr = (welfare_self/365/`cpival'/`pppval') >= `zline' qui sum _sr [aw=weight] local sr : di %9.6f r(mean) * poverty headcount: weighted proportion below the line using welfare gen byte _poor = (welfare/365/`cpival'/`pppval') < `zline' qui sum _poor [aw=weight] local hc : di %9.6f r(mean) drop _sr _poor di "`ccode' `year' - Self-reliance share: `sr' Headcount: `hc' ($`zline'/day 2021PPP)" ``` ::: For citation guidance and further options (e.g. PPP conversion on load), see the `datalibweb` help file in Stata and the [datalibweb guidelines](https://github.com/worldbank/datalibweb); questions to the datalibweb team at `datalibweb@worldbank.org`.