Document/unify handling of PPP and MER GDP projections

[khaeru]

@r-aneeque and I reached out to @setupelz @Milkally (w/ help from @yiyi1991) to understand different practices for producing and using GDP projections in PPP and MER. This was initially to clarify what we should do in MESSAGEix-Transport, but the information is more broadly useful.

The “Details” section below collects most of what we received. This issue is to:

1. Collect, as comments, any further details, including:
   a. Specific code or locations of code.
   b. Methods implemented by that code.
   c. Configuration used.
   d. Other practices for handling such data.
2. Discuss options for shared code that can meet the various requirements identified in (1). For instance, we already have classes for reading the SSP 2017 and SSP 2024 data (first link below under “Details”) as provided from the respective official sources. These could be improved to (optionally) implement or invoke the methods identified in (1a–d).

Details

In message-ix-models we have .project.ssp.data

• See the link for documentation; it's not repeated here.

Elsewhere, predating the SSP 2024 project:

• Conversion of monetary units IAMconsortium/units#25 (no action since 2023-07)
• The above links to https://github.com/gidden/salamanca (no action since 2021-03).

For the SSP 2024 project, @setupelz:

• Created the R code in SharedSocioEconomicPathways2023 > WP - Drivers > GDP > 20240311_v1 (private; IIASA SharePoint) which does the following:
  • Uses GDP [$PPP2017] “given by OECD colleagues”.
  • Selects GDP|PPP from (model=“OECD ENV-Growth 2023”, scenario=“SSPX – Final Review”) in new SSP database.
  • Convert to $PPP2010 & $PPP2005, constant conversion factors.
  • Convert to $PPP2005 to $MER2005, using time-varying conversion factors (Tieju Ma), only goes back to 2010. SSP1 factors used for SSP5, SSP3->SSP4.
  • Convert $MER2005 to $MER2010 & $MER2017, constant conversion factors.

For GLOBIOM, @Milkally:

• Used a GDP processing workflow documented in a slide deck (currently not attached)
• As shown in the slides, derived / download different economic converters (for PPPvsMER conversion, or constant price conversion between different MER prices) from WB WDI database (Apr 2024 version).
• Discussed with Setu and we “realised that we are using similar/identical methods to convert GDP|PPP between different price years, but just different methods (i.e. MESSAGE applying dynamic while we applying static converters) when converting GDP|PPP to GDP|MER (for a same constant price year).”
• Use [in GLOBIOM] the converted “GDP|MER (2000 $)” as model input, to match other prices in GLOBIOM (which are all in 2000$ unit). This was converted from the original GDP|PPP (2017 international $) value from SSPV3 database, following the workflow documented in the slides.
• Take this conversion to GDP|MER (2000 $) for example:
  • When I say I am using “fixed PPP to MER converter”, I am first using the PPPtoMER converter for year 2017, to convert the original GDP|PPP (2017 international $) value from SSPV3 database into “GDP|MER (2017$)” value.
  • Then I used the “constant price converter” derived from WB data, to convert “GDP|MER (2017$)” into “GDP|MER (2000$)”.
  • Similar workflow is applied when we convert the GDP data into different constant years’ prices and different metrics (i.e. PPP or MER-based).