OECD parser walkthrough#

This notebook is the practical guide for parsing OECD tables in MARIO.

What this notebook covers#

  • where each OECD table family comes from;

  • when to pass a direct file path and when to pass a directory;

  • how dataset=, year=, and country= change the parser workflow;

  • the difference between ICIO, national IOT, and SUT parsing;

  • the current limitation that OECD tables are parsed as economic tables only.

Relevant source pages#

MARIO does not download OECD files automatically. For ICIO and national IOT, the expected workflow is to download the source files manually and then point the parser to the local files. For SUT, MARIO pulls the data directly from the official OECD SDMX API.

Main entry point#

For normal user workflows, the public entry point is:

  • mario.parse_oecd(...)

The same function currently supports:

  • OECD ICIO csv bundles;

  • OECD national total-table IOT csv files;

  • OECD SUT tables through the official SDMX API.

Key arguments#

The key public arguments are:

  • path: one local OECD file or one directory containing multiple OECD files;

  • dataset: choose "ICIO", "IOT", or "SUT";

  • year: used to disambiguate directories and required for dataset="SUT";

  • country: used to disambiguate national IOT files and required for dataset="SUT".

Direct file path vs directory path#

Use a direct file path when you already know the exact OECD file to parse. This is the simplest workflow and usually means you do not need any further disambiguation.

Use a directory path when you keep several OECD files together. In that case MARIO scans the directory and then uses dataset=, year=, and sometimes country= to pick the correct candidate.

Expected local layouts and caveats#

Typical local layouts are:

For dataset="ICIO":

OECD/ICIO/
├── Extended/
│   └── 2022.csv
└── Regular/
    └── 2022_SML.csv

For dataset="IOT":

OECD/IOT/TTL/
└── ITA2022ttl.csv

For dataset="SUT", path is not required; use country= and year= to select the SDMX slice.

Practical caveats:

  • the IOT workflow supports only the OECD accounts labeled as total table;

  • in the extended ICIO release, CN1 and CN2 are aggregated automatically into CHN, and MX1 and MX2 into MEX.

[1]:

import mario

Parse one explicit OECD ICIO csv#

Use this when you want to point MARIO to one exact ICIO file, either regular or extended.

[ ]:

db = mario.parse_oecd(
    path="/path/to/2022_SML.csv",
    dataset="ICIO",
)

db

INFO Parser: reading OECD ICIO file 2019_SML.csv.
INFO Parser: OECD ICIO parsed with 81 sector regions, 50 sectors, 486 final-demand columns and 2 factor rows.
INFO Metadata: initialized.

name = OECD ICIO 2019 regular
table = IOT
scenarios = ['baseline']
Factor of production = 2
Satellite account = 1
Consumption category = 6
Region = 81
Sector = 50

Parse one national OECD total-table IOT#

Use this when you want one national total table such as CZE2014ttl.csv.

[ ]:

db = mario.parse_oecd(
    path="/path/to/ITA2022ttl.csv",
    dataset="IOT",
)

db

Parse one OECD SUT directly from the SDMX API#

In principle, this workflow does not require a local path, but it does require country= and year=.

At the moment we do not show an executable example in the documentation because the official OECD SDMX endpoint can intermittently refuse the request with HTTP errors such as 403 Forbidden. When that happens, the issue is upstream of MARIO’s SUT transformation logic and the parse cannot complete.

For this reason, treat dataset="SUT" as temporarily unstable in the public docs until the OECD SDMX service becomes reliably accessible again.

[ ]:

db = mario.parse_oecd(
    dataset="SUT",
    country="ITA",
    year=2022,
)

db

Environmental extensions#

Warning. The OECD parser currently exposes the economic tables only.

ICIO, national IOT, and SUT tables are therefore not parsed as environmentally extended tables in MARIO at the moment.

Download this notebook