The manuscript describes a framework for evaluating long-term damages from climate change to a large number of different specific U.S. economic sectors. The damage functions have generally been previously published, and have been extended to 2300 (from an original 21st century time horizon for most of them). The common FaIR climate model is used to simulate and take into account uncertainty in global climate, specifically temperature change, which is used as input to the various damage functions. Probabilistic projections for socioeconomic conditions (e.g., GDP, population change) from a recent study are used, and combined with different emissions scenarios. The work decomposes climate-driven damages by geographic region within the U.S. and by different vulnerable populations.

The manuscript is well-written and well-organized. As near as I can tell from the methodological descriptions of FrEDI provided in the manuscript, the conclusions and results follow nicely from the experiments that are conducted. However, my comments below revolve mostly around the theme of a need for further details about the structural assumptions baked into the damage function underlying FrEDI. You understandably defer a great deal to other recently published works that are the primary sources for these parameterizations, and I understand that this is not a model description paper. However, this makes it a bit more difficult for the reader to understand the context for the results presented, particularly for interpreting sector-specific damages. Most of these and my other comments I anticipate can be cleared up through adding some modeling details and/or caveats. Thus I recommend for minor revisions.

General comments:

L65-67 - Perhaps any of these FAIR warming scenarios *can* be paired with any socioeconomic projection, but *should* they? There are some SSP-RCP combinations (e.g.) that modelers use but shouldn’t. For example, see O’Neill et al 2020, https://doi.org/10.1038/s41558-020-00952-0, particularly their Figure 2. I can imagine that your probabilistic framework may account for this in some way, but I’d be interested to know more about how you ensuring that unlikely/implausible combinations of scenarios aren’t unduly biasing results?

L74, 159 - What assumptions about coastal adaptation (and other forms of adaptation/mitigation, and development) are baked into FrEDI? Can you comment on how realistic these are on such long time scales?

L186 - Related to above, looking specifically at marginal damages means that the assumed adaptation strategies could be masking the real magnitude of these climate risks. For example, if it is assumed that you make least-cost decisions, then these marginal damages will be more of a lower bound. While the adaptation assumptions are provided in an appendix, it may be worth discussing these in main text more, because (my impression is that) the results rely a great deal on these assumptions.

L210-212 - It isn’t totally clear here what the default adaptation assumptions are for each sector (maybe cite Table A2 here?), or what each of these adaptation assumptions represents with respect to the individual sectors. For example, what does “reasonably anticipated adaptation” mean for coastal properties? And how reasonable are these assumptions on such long time scales, for each sector?

Specific comments:

L23-27 - Is this by the year 2090? Or when?

L50 and 52 - “Climate Change Impacts and Risk Analysis (CIRA) project”  and “Climate Impact Lab” - I’m aware of these projects/groups, but I’m not sure readers in general will know the context for these. Could be worth specifying where they’re housed.

L70 - In what way(s) is FrEDI distinguishable from a traditional IAM?

L114-116 - Are there or should there be correlations and/or feedbacks represented between the RFF-SP emission scenarios and the FaIR scenarios?

L116-118 - How sensitive are the results to this moderate assumption? And no uncertainty sampled in these other gases/aerosols.

L175 - should this be g_t? 

Table 2 - should “2,1” be “2,100” in the 95% CI column? 

L270-272 - I’m intrigued by this, and would love to see more work on this prominently featured in contemporary research. But I’m also curious about how this modeling is done within FrEDI and what scale data is being used, and what assumptions there are when disaggregating damages for socially vulnerable populations.

L424-430 - You cite two studies that used a power law relationship, finding that it fit the temperature-damage relationship better than linear. Why choose to use linear instead in FrEDI?

Figure A2 - There is a lot to unpack here, including a few of what appear to be some tipping points or other kinds of jumps in some of these sectoral damages around 2060 or so. E.g., wind damage, agriculture, or the sudden jump in coastal property damages after about 2080 - how do you make sense of these results, in light of the adaptation assumptions embedded in FrEDI?

L469 - perfect foresight over what time frame?

Figure A5 - Are some of these damage distributions in later years so heavily skewed that the means are outside of the 95% CI? It might be tidier to display the median, and perhaps more consistent with the use of percentiles for the 95% CI too.

Please see attached file.