Follow-Up Question on Damage Coefficients

[austinrwg]

Hello @kemccusker, thank you for your response from the previous issue. I have a follow-up question and, unsure if a closed issue would notify you, I'm opening a new issue for this.

I thought that the DSCIM model has the functions to compute the damage coefficients, if the damage files are not provided. So, in code, I was wondering how to compute the coefficients under an alternative scenario, say 3%, 3.5%, or other discount rates?

[austinrwg]

I took a closer look at the input folder, and there seems to be damage files for 3.0% discount rate. Just to confirm: is this correct?

[kemccusker]

hi @austinrwg, can you clarify where in the inputs you saw the 3.0% discount rate? I don't recall there being anything but 1.5, 2, 2.5%.

To answer your broader question, yes, the code to estimate damage functions is part of DSCIM. However, as mentioned in your other Issue #27, the data on which the damage function coefficients are fit is many terabytes in size and not provided. The starting point is the already estimated damage function coefficients.

The coefficients do not technically vary by discount rate. However, they do vary based on the eta value, which defines the curvature of the utility function used in the certainty equivalent calculation over econometric uncertainty in the spatial step and in the Ramsey discounting formula. For discounting, the eta value is paired with a rho to match starting discount rates of 1.5%, 2%, and 2.5%. It is not possible to vary these. The eta value is kept consistent across the spatial step and the discounting step.

The Damages Module and Discounting Module (Sections 4 and 5) in the DSCIM User Manual, describes the process of projecting future climate damages and fitting damage functions.

[austinrwg]

Thank you for your answer. It is very helpful.

There are damage function coefficient files that refer to the 3% discount rate based on eta and rho values. The file name looks like this: risk_aversion_euler_ramsey_eta1.568_rho0.008_dfc.nc4

Note: I didn't mean to close this conversation. I don't know how to revert it back to open.