Change reference cost year to 2020

[lisazeyen]

Closes # (if applicable).

Changes proposed in this Pull Request

Change the reference cost year to 2020, closes issue #118

• a new column currency_year is added which represents the currency year of the original input data
• all input data is adjusted for inflation
• for input data where currency year was not clear, year of publication is assumed
• for old pypsa cost assumptions currency year 2015 is assumed

Checklist

• add annual inflation from eurostat

• @JulianGeis one would need to go through all the techs listed here and double check the reference cost year. All the techs with cost year 2020 should be added to the list here

• add reference year for vehicle costs

• Code changes are sufficiently documented; i.e. new functions contain docstrings and further explanations may be given in doc.

• Data source for new technologies is cleary stated.

• Newly introduced dependencies are added to environment.yaml (if applicable).

• A note for the release notes doc/release_notes.rst of the upcoming release is included.

• I consent to the release of this PR's code under the GPLv3 license.

[fneum]

@aleks-g and @koen-vg have used real inflation data instead of snakemake.config['rate_inflation'] from https://ec.europa.eu/eurostat/databrowser/view/teicp000/default/table?lang=en in https://github.com/koen-vg/enabling-agency / https://arxiv.org/abs/2312.11264.

While costs in the above repository are given in 2015 EUR, we have converted all cost data to 2023 EUR for the purposes of this study, using inflation data for the Euro area up to October 2023 [13], amounting to a 24.5% increase compared to technology-data.

Don't know though if that was done afterwards or directly in a fork of technology-data.

[aleks-g]

@aleks-g and @koen-vg have used real inflation data instead of snakemake.config['rate_inflation'] from https://ec.europa.eu/eurostat/databrowser/view/teicp000/default/table?lang=en in https://github.com/koen-vg/enabling-agency / https://arxiv.org/abs/2312.11264.

While costs in the above repository are given in 2015 EUR, we have converted all cost data to 2023 EUR for the purposes of this study, using inflation data for the Euro area up to October 2023 [13], amounting to a 24.5% increase compared to technology-data.

Don't know though if that was done afterwards or directly in a fork of technology-data.

Hi Fabian, we've only included inflation since 2015 in the analysis/post-processing afterwards, without adapting anything in technology-data itself. :)

[lisazeyen]

@aleks-g and @koen-vg have used real inflation data instead of snakemake.config['rate_inflation'] from https://ec.europa.eu/eurostat/databrowser/view/teicp000/default/table?lang=en in https://github.com/koen-vg/enabling-agency / https://arxiv.org/abs/2312.11264.

While costs in the above repository are given in 2015 EUR, we have converted all cost data to 2023 EUR for the purposes of this study, using inflation data for the Euro area up to October 2023 [13], amounting to a 24.5% increase compared to technology-data.

Don't know though if that was done afterwards or directly in a fork of technology-data.

I guess it make sense to take the actual inflation. I downloaded the annual inflation from eurostat and would add this instead of assuming 2% inflation for every year

[lisazeyen]

We haven't checked all technologies yet in DEA, but I think we can already pull it in and open another issue that all single DEA technologies should be double checked for the correct currency year.

[koen-vg]

I have to say I'm a little bit confused by the Eurostat dataset browser sometimes because it sometimes includes and sometimes doesn't include 2023. But here it seems possible to get a dataset in the same format with 2023 included? https://ec.europa.eu/eurostat/databrowser/view/prc_hicp_aind__custom_9900786/default/table?lang=en

I think it would be nice to have as recent as possible data available, and there are now inflation values for 2023 (namely, 6.4%).

Apart from that, thanks a bunch for this work!

[lisazeyen]

I have to say I'm a little bit confused by the Eurostat dataset browser sometimes because it sometimes includes and sometimes doesn't include 2023. But here it seems possible to get a dataset in the same format with 2023 included? https://ec.europa.eu/eurostat/databrowser/view/prc_hicp_aind__custom_9900786/default/table?lang=en

I think it would be nice to have as recent as possible data available, and there are now inflation values for 2023 (namely, 6.4%).

Apart from that, thanks a bunch for this work!

Good idea to include 2023, unfortunately the data you are linking to is only dating back to 2015 and it would be nice to have the earlier years in there as well. I have added 2023 now within the function when preparing the data (not the nicest way, but at least 2023 is included now as well)...

[fneum]

Looks mostly good. Some column headers were removed and there were some local file paths added, and we should link the source for the inflation data.

Did you do a few spot checks whether the adjustments made sense in outputs? If yes, I am happy for this to be merged.

Regarding the range of inflation data available, if the other dataset (dating back pre-2015) is not updated, we can think about merging the two sources.

[fneum]

Let's do the following spot checks:

• H2 (g) pipeline
• HVDC overhead
• nuclear
• methanolisation

[lisazeyen]

Looking at projections for 2020 and comparing data as named in the original source with the outputs of the technology data

                            original            new  currency_year
H2 (g) pipeline investment    363.08     303.684542         2023.0
HVDC overhead   investment     400.0     442.141439         2011.0
nuclear         VOM             4.24       3.546388         2023.0
                fuel             3.0       3.412227         2010.0
                investment   10275.0    8594.135356         2023.0
methanolisation investment  788000.0  819108.478032         2017.0

Let's do some example calculations e.g. for
methanolisation
investment costs 788 000 Eur_2017 -> inflation in 2018 1.8%, 2019 1.4%, 2020 0.7%
788 000 * (1.018 * 1.014 * 1.007) = 788 000 * 1.039478 = 819 108.47 Eur_2020

H2 pipelines
original 363.08 Eur_2023 -> inflation in 2021 2.9%, 2022 9.2% and 2023 6.4%
363.08 / (1.029 * 1.092 * 1.064) = 363.08 / 1.1955 = 303.6845

[euronion]

I have to say I'm a little bit confused by the Eurostat dataset browser sometimes because it sometimes includes and sometimes doesn't include 2023. But here it seems possible to get a dataset in the same format with 2023 included? https://ec.europa.eu/eurostat/databrowser/view/prc_hicp_aind__custom_9900786/default/table?lang=en
I think it would be nice to have as recent as possible data available, and there are now inflation values for 2023 (namely, 6.4%).
Apart from that, thanks a bunch for this work!

Good idea to include 2023, unfortunately the data you are linking to is only dating back to 2015 and it would be nice to have the earlier years in there as well. I have added 2023 now within the function when preparing the data (not the nicest way, but at least 2023 is included now as well)...

You could use the OECD data repository instead, they list HICP since 1991 for the current EURO area and allow for download as CSV or by API access:

https://data-explorer.oecd.org/vis?tm=HICP&pg=0&snb=5&vw=tb&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_PRICES%40DF_PRICES_HICP&df[ag]=OECD.SDD.TPS&df[vs]=1.0&pd=%2C&dq=.A.HICP.CPI.PA._T.N.GY&ly[rw]=REF_AREA&ly[cl]=TIME_PERIOD&to[TIME_PERIOD]=false&lo=100&lom=LASTNPERIODS

[lisazeyen]

I have to say I'm a little bit confused by the Eurostat dataset browser sometimes because it sometimes includes and sometimes doesn't include 2023. But here it seems possible to get a dataset in the same format with 2023 included? https://ec.europa.eu/eurostat/databrowser/view/prc_hicp_aind__custom_9900786/default/table?lang=en
I think it would be nice to have as recent as possible data available, and there are now inflation values for 2023 (namely, 6.4%).
Apart from that, thanks a bunch for this work!

Good idea to include 2023, unfortunately the data you are linking to is only dating back to 2015 and it would be nice to have the earlier years in there as well. I have added 2023 now within the function when preparing the data (not the nicest way, but at least 2023 is included now as well)...

You could use the OECD data repository instead, they list HICP since 1991 for the current EURO area and allow for download as CSV or by API access:

https://data-explorer.oecd.org/vis?tm=HICP&pg=0&snb=5&vw=tb&df[ds]=dsDisseminateFinalDMZ&df[id]=DSD_PRICES%40DF_PRICES_HICP&df[ag]=OECD.SDD.TPS&df[vs]=1.0&pd=%2C&dq=.A.HICP.CPI.PA._T.N.GY&ly[rw]=REF_AREA&ly[cl]=TIME_PERIOD&to[TIME_PERIOD]=false&lo=100&lom=LASTNPERIODS

they are mentioning as source Eurostat (=.
I found a Eurostat table now with data from 2001-2023 and added that one as well as the API link to the source

[lisazeyen]

@fneum I compared now the original data input with the cost adjusted values. Comparing the old and new costs is not very insightful since many of the input data previously was not inflation adjusted.

One thing to double check, I am basically assuming now when converting e.g. Eur_2017 to Eur_2020 the inflation of the three year 2018, 2019, 2020, since Eur_2017 should consider already the inflation for 2017 right?

[fneum]

One thing to double check, I am basically assuming now when converting e.g. Eur_2017 to Eur_2020 the inflation of the three year 2018, 2019, 2020, since Eur_2017 should consider already the inflation for 2017 right?

I think either way is fine, just it cannot be 2017, 2018, 2019, 2020.