opentaps / open-climate-investing Goto Github PK

Thinking about #28 to find places to buy carbon offsets, maybe we should define what valid carbon offsets are?

In the Tao of Decarbonization I felt that carbon offsets should follow the core carbon principles at the time of activity: additionality, baseline, leakage, permanence, real.

More basically, though, I think some other requirements may be needed. For example,

• Standards should be maintained independently
• Auditors need to be independent organizations than project developers
• Auditors can be part of developing standards, but the standards should not be maintained solely by auditors and definitely not by one auditor.

Try using the Fama-French North America 3 factors from http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html

instead of the ff_factors.csv from the original CARIMA spreadsheet.

See https://investing.opentaps.org/stocks/CVE.TO

When the stock page first loads, or when you change the BMG settings, it takes a while to load, but the page is just blank or not changing. Add a spinner so that while it is loading you know you're waiting for something, instead of thinking nothing is available or the result is the same.

Not an issue with open-climate-investing per se.

The documentation cites https://carima-project.de/en for accessing tools and the BMG factor, but on multiple machines I just get an SSL error trying to access the site. This makes investigation into the factor problematic.

See if ESG funds could be used to construct BMG index, or if the market is not trading according to the fund's ESG criteria:

1. Find a list of popular ESG funds
2. Based on their sector weights, find a matching ETF
3. Calculate a BMG series from the ESG fund and the matching ETF
4. Compare this BMG series to the other funds we've analyzed in the paper

Compare the BMG factor to the historical pricing of the EU Emissions Trading Scheme futures to determine if the EU ETS futures are suitable as a BMG factor.

The EU ETS can be converted into a BMG factor by calculating its price change, with a positive price change being a negative BMG return and negative pric echange being a positive BMG return, since increases (decreases) in emissions prices should lead to worse (better) performance of "brown" stocks vs "green" stocks.

This can be based on correlation of EU ETS with the BMG factor and testing it as a model variable following the steps in #4

Initial implementation of the CARIMA model based on the Fama French factors model from https://www.uni-augsburg.de/de/fakultaet/wiwi/prof/bwl/wilkens/sustainable-finance/downloads/ and https://docplayer.net/156158475-Carbon-risk-management-carima-rollout-conference-2-nd-july-2019-frankfurt-am-main.html

Integrate in the Brown Minus Green returns time series from the spreadsheet and the other return factors from http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html

Integrate stock return data set

Goal is to be able to perform calculations on individual stocks using the CARIMA model

Please follow the file Paper_steps.md to test the results in the 2 google spreadsheets:

• Using data/carbon_risk_factor.csv:
  • "Carbon Loadings Results - CARIMA" - CARIMA Industry, CARIMA Industry EW, CARIMA BMG Significant Stocks, No BMG Industry, No BMG EW
• Using data/bmg_eu_ets.csv
  • "Carbon Loadings Results - Alternates" - EUA Futures, EUA EW
• Using data/bmg_acwi_crbn.csv
  • "Carbon Loadings Results - Alternates" - ACWI-CRBN, ACWI-CRBN EW, ACWI-CRBN Significant Stocks
• Using data/bmg_xop_smog.csv
  • "Carbon Loadings Results - Alternates" - XOP-SMOG, XOP-SMOG EW, XOP-SMOG Significant Stocks
• Using data/bmg_xop_smog_orthogonalized_1.csv
  • "Carbon Loadings Results - Alternates" - XOP-SMOG ORTHO 1, XOP-SMOG ORTHO 1 EW, XOP-SMOG ORTHO 1 Significant Stocks
• Using data/bmg_xop_smog_orthogonalized_2.csv
  • "Carbon Loadings Results - Alternates" - XOP-SMOG ORTHO 2, XOP-SMOG ORTHO 2 EW, XOP-SMOG ORTHO 2 Significant Stocks
• Also for all the data/ files,
  • "Carbon Loadings Results - CARIMA" - Number of Stocks by Sector, Correlations, BMG Regressed vs Other Factors - Except for the bmg_eu_ets.csv, I did not do regressions vs other factors

At the moment, we have a table that is specifically designed for the 4-factor Fama-French factors. It might be a better design to have a long table with factor names and dates that allow for other factors besides FF, and that the user can specify as part of using everything.

Please add in the read me file what all the fields of the stats model output mean, and which ones are important when looking at the results.

Make the loading of daily and monthly data from the Ken French data library consistent:

1. The files in our data/ repository should be the CSV files from the Fama/French 3 Factor, Fama/French 3 Factors [Daily], Momentum Factor (Mom), Momentum Factor (Mom) [Daily]
2. The files should not have the additional FREQUENCY column. Our data loading script should set the frequency based on the file.
3. The risk free rate in both cases should be in the CSV files. There should not be separate risk free rates file.

Use the XOP-SMOG BMG series to calculate the market implied climate risk (BMG factor loading) for energy, airlines, and transportation stocks.

When testing the model against a utility (DUK), it seems that the regression model has little explanatory power. Could we improve it by using a specific market index (such as utility ETF fund) as the market factor and then add in the carbon risk factor? Would that give good explanatory power to the stocks in that industry? Would it help identify stocks with high and low carbon risks when the particular sector is not as correlated with the market as a whole?

This could be expanded to other sectors, such as airlines.

If stocks in these sectors which have high emissions output do not have carbon betas even relative to their own sectors, then the market may be saying that the companies are not vulnerable. For example, it may believe that utilities will simply be allowed to pass through emissions taxes to their customers, and therefore the transition risk is not at the utilities but at the customers of the utilities.

Move all the python scripts to the scripts/ directory so the root directory has data/ scripts/ book/ dao/ ui/

Modify the bulk script to store results for a large (index scale) number of stocks in a postgresql database, so that we can get both the historical carbon betas and their trends (increasing or decreasing) for all the stocks in an index universe.

From this, we can find stocks to construct the BMG index factor and then determine, through studying their disclosures and #4, which ones should be part of the BMG index going forward based on their positive (Brown) and negative (Green) carbon betas and the trends in those betas (declining trend means they are becoming less sensitive and would more likely be excluded.)

This is from pages 184 - 188 of the CARIMA paper, which shows these statistics to validate the BMG factor:

Please build scripts to calculate these statistcs for BMG and other factors in our data so we could see how good different ways of constructing BMG benchmarks are.

See if it's possible to develop model of emissions for airline based on their passenger miles data.

Test using ESG funds as carbon risk factor.

Do ESG funds (using their ETF's price histories) have significant carbon betas?

If so, could their returns minus market general index be used as the BMG factors?

Need to check the data that comes in and ensure that it doesn't fail, removes outliers and provides good output

Modify get_stocks.py to add parameter to get daily data. By default it'll still get monthly data.

Modify get_regressions.py to add parameter to use daily data. By default the regressions will still be run for monthly data. If the daily data option is selected then the -i parameter will be for the number of days. By default the number of days is 730 days (24 months)

Find some places where we could buy carbon offsets for the Decarbonization DAO

We want something like the Gold Standard Marketplace, where we could buy offsets from specific projects rather than generic names. It does not have to be in crypto.

@mattbowler Could you please check the regression results? They're very different than when I ran them initially.

For example, this is the result of factor_regression.py for the stock CVX:

Compare it to the results from #2 (comment)

Also the original bulk script is not working:

We should get results similar to #3 (comment)

See if it's possible to develop emissions calculation of a transportation/freight company based on their total amount of goods and emissions factors.

Check with PACTA Analysis to see if their model of companies following PACTA aligns with our market-based carbon betas.

There are error messages being printed by underlying packages. These are not impacting the results but need to be traced and fixed in case this causes issues down the line.

Compare a few personal emissions calculators, for example from Project Wren Calculator

Let's see what causes the most of our personal emissions footprints, and also how different the results they give are.

Then create a documentation guide on estimating personal emissions footprints and then how to actually measure them using the blockchain-carbon-accounting app

Implement the R correlation script in python

It should calculate correlations and regressions with bond market and other Fama French factors

We should then use the significant regression coefficients to create orthogonalized series of the BMG

Do the companies which set SBTI targets have lower carbon betas?

Does it matter if they are "committed" or "targeted" (ie approved)?

Modify the bulk_script.py to generate results for list of stocks.

It can be a list of stock tickers either in a CSV file or embedded in bulk_script.py as an array for now. (We'll eventually hook it up to a database and backend.)

The output can be the same as now, statistics for each stock separated by a header.

For example, using different sectors of the market. Test these with #4 to validate the quality of the BMG factor created.