Zoom Logo

Virtual Seminar on Climate Economics - Shared screen
Glenn Rudebusch (SF Fed)
25:00
To ask a question, type it into this chat feature and select “Raise hand.” If you want me to read the question, type “Please ask” as well.
Emily McGlynn
32:58
Not a macroeconomist, but it looked from that figure like real interest rates are mean reverting - does it make sense to use some estimated mean discount rate rather than continuously updating it?
Michael Bauer
37:19
@Emily, there is a lot of evidence for a long-run trend component in real interest rates, often called r*. This trend has declined over the last three decades. In other words, the long-run mean is lower now than it used to be. In my most recent paper with Glenn, we show that r* is the anchor for all social discount rates. See here: https://www.frbsf.org/economic-research/publications/working-papers/2020/25/
Emily McGlynn
38:06
Thank you!
Juan
50:05
Perhaps not in the costs detailed on the slides, but beyond mortality, would one want to consider impact on morbidity and more generally on quality of life?
Ishan Nath
51:36
@Yichun, in the paper on labor supply we use the effects of heat on worker hours to infer the disutility they experience during the time they spend working in adverse weather conditions. So that’s a partial version of the idea you asked about, though we don’t have results on outdoor leisure and related amenities that you mention.
John Morehouse
53:32
Super interesting work. For adaptation -- do you have anything to say about short-run vs. long-run adaption? For example, in the short-run people use more energy to adapt to hotter/colder realizations of weather. In long run, those that can afford it move -- as their understanding of the underlying climate distribution updates, or something like that?
Ishan Nath
54:01
@Luis the details of the method can be found in this working paper: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3224365 and a recording of this presentation (including slides) will be posted here later https://www.frbsf.org/economic-research/events/virtual-seminar-on-climate-economics/
Ishan Nath
55:37
(I was deputized by Tamma to try to handle clarifying questions in the chat - leaving the longer ones for the Q&A after)
Wojciech Szadurski
56:59
When you make projections, do you allow warming impacts to feed back into average temperature and average GDP per capita?
Ishan Nath
58:42
@Wojciech no, we take temperature and growth pathways as exogenously given, though we do account extensively for uncertainty in both
Emily McGlynn
01:00:24
One thing I haven't been able to wrap my head around with this method is a "dosage" issue - by exploiting weather random shocks are you able to account for the fact that under climate change, you will ostensibly be facing a larger dose of those shocks which may marginal effects?
Emily McGlynn
01:00:42
*may increase marginal effects
Brian Prest
01:02:11
Is the VSL a constant, or does it vary with projections like GDP? (Sorry, resending this question to all attendees)
Ishan Nath
01:02:40
@Brian, we have projections both with a region-varying VSL based on local income and one in which we value all global deaths at a constant average VSL. In both, we allow the VSL to evolve with economic growth (either locally or globally).
Ishan Nath
01:05:53
@Emily, this is probably best left for the Q&A, but a quick answer (also relevant to John’s question from earlier) is that by conditioning marginal effects on local climate, I think we do account for whether repeated dosage (e.g. of hot days) differs from occasional dosage. But my intuition and our findings is that repeated dosage reduces marginal effects because populations are more adapted (e.g. hot day in Texas vs. Seattle). If there are nonlinear convexities of clustered shocks (e.g. droughts in multiple consecutive years that reduce groundwater stocks), which I think is more along the lines of what you were thinking, then I think we would not capture those. But perhaps it’s best to let Tamma answer this more completely in the Q&A.
Emily McGlynn
01:07:16
Yes the multiple droughts example is a good one
Ishan Nath
01:08:35
@Toan, the damage uncertainty comes from Monte Carlo draws from the confidence intervals of our empirical estimates. For calculating risk premia etc. (in the paper referenced on this slide) we use CRRA preferences, though we’re happy to show robustness for other preference specifications that people want to see.
Brian Prest
01:09:32
I'm curious about the effective climate beta of your damage estimates. In DICE, the beta is ~1. Is it similar in your results? Relatedly, how it interacts with uncertainty about future economic growth and the choice of your endogenous discounting parameters in SSPs 2-4 (i.e., how you choose your Ramsey parameters)?
Ishan Nath
01:10:21
@Emily one more thought a colleague just texted me is that it’s apparently challenging for climate science to make projections about such clustered shocks (e.g. consecutive droughts) so the challenge of accounting for them would go beyond just impact estimation econometrics
Ishan Nath
01:10:37
@Kristina, yes that’s right!
Ishan Nath
01:14:55
@Brian we will have a lot to say about the effective climate beta and accounting for it in the write-up of this paper! In a nutshell, we are finding that climate mitigation pays off quite disproportionately in bad future states of the world (across the range of uncertainty represented in our projections), so the discount rate we calculate that accounts for that insurance value is substantially below the risk-free rate.
Ishan Nath
01:16:29
@Chris, that’s a deep question that’s probably best to let Tamma handle at more length, but I think the short answer is that we try to account carefully for the double-counting aspect but do not account for the general equilibrium cross-sector spillover aspect that you raise.
Brian Prest
01:21:51
I'm curious to hear more about the mechanism that's generating a negative climate beta. E.g., Mortality and labor impacts both sound like they'd be bigger when growth is bigger (since the VSL evolves with growth), but maybe adaptation overwhelms this?
Ishan Nath
01:23:25
@Bryan yes we do extend the emissions profiles to 2300, though I’d have to look it up to know exactly where they come from or what they look like
Glenn Rudebusch (SF Fed)
01:23:50
Type in your questions to “all panelists and attendees” using this chat feature and select “raise your hand.”
Brian Prest
01:31:52
I've already talked too much, so just putting this in the chat. Sorry I missed the distinction between "Mean over uncertainty" and "Certainty equivalent". What's the difference?
Ishan Nath
01:34:05
@Brian “mean over uncertainty” is just the mean over all draws relative to the value with the single draw with the median FAIR parameters, but doesn’t actually incorporate a risk premium. “Certainty equivalent” uses the CRRA utility function to value the welfare consequences of the uncertainty.