class: center, middle, inverse, title-slide # Portfolios with climate risk measures ### Guillaume Coqueret .font90[.grey[(EMLYON Business School)]] ### .font90[Financial Risks International Forum] 2022-03-22 --- <!-- xaringan::inf_mr() --> class: inverse, center, middle <style type="text/css"> .remark-slide-content { font-size: 32px; padding: 1em 4em 1em 4em; } </style> # Introduction --- # Context <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .font90[The demand for .green[green assets] has **soared**. .font80[.grey[(Morningstar data below)] ]] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .bib[ <img src="morning.png" width="1050" style="display: block; margin: auto 0 auto auto;" /> ] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> ... .font80[which may have driven green **returns**: see [van der Beck (2021)](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3929359) - and [Gabaix & Koijen (2021)](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3686935) more generally.] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> `\(\rightarrow\)` .font80[**issues remain**: cost of sustainability, implementation leeway, greenwashing?] --- # The problem(s) .font90[**Portfolio choice** seeks to determine optimal allocation between assets and requires "DOC" inputs from the investor:] - .font80[**data**: returns, usual characteristics (for factors), ESG features: .font90[.grey[used for estimation purposes (e.g., expected returns, covariance matrix,] .green[greenness].grey[, etc.)]]] - .font80[the **objective** (utility function): preference with respect to risk, greenness] - .font80[some **constraints**] .font70[.grey[(liquidity, factor/geographical/sector exposure, leverage, etc.)]] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> -- .font80[In the end:] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> `$$\small w^*=f(D,O,C)$$` .font80[In practice, portfolios can be (very) sensitive to these three groups of choices.] --- class: inverse, center, middle <style type="text/css"> .remark-slide-content { font-size: 32px; padding: 1em 4em 1em 4em; } </style> # First issue: data --- # ESG disagreement <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .font80[Raters have different **sources** & **methodologies**, which] .font80[generate potentially large discrepancies in their assessments. This is widely documented:] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> -- .font80[ [Dimson, Marsh, Staunton (2020)](https://jpm.pm-research.com/content/47/1/75), [Avramov, Cheng, Lioui & Tarelli (2021)](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3711218) [Gibson, Krueger & Schmidt (2021)](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3433728) & [Berg, Kolbel, Rigobon (2022)](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3438533): ] -- <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> <img src="disagreement.png" width="890" style="display: block; margin: auto 0 auto auto;" /> <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> -- `\(\rightarrow\)` .font80[**confusing** for asset managers: which to use, how to average?] --- # Which metrics to use? <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .font80[[Le Guenedal & Roncalli (2021)](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3999971): *Portfolio Construction with Climate Risk Measures*] .font90[What's the best **proxy** for greenness?] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> -- .pull-left[ - .font80[current] `\(\small \text{CO}_2\)` .font80[/ GHG emissions] `\(\rightarrow\)` .font80[**size bias**] - .font80[emission **intensities** (emissions divided by revenue or mkt. cap.)] `\(\rightarrow\)` .font80[hard to average] - .font80[**pathways**] `\(\rightarrow\)` .font80[require models & estimations for firms-specific trends:] ] -- .pull-right[ <img src="pathway.png" width="690" /> ] `\(\rightarrow\)` .font80[reliable **Scope 3** time-series are short! Extrapolation can be hard/risky.] --- # The cross-section of extrapolations <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .font90[Very much trending upwards! Hard to sustain...] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> <img src="paths.png" width="660" style="display: block; margin: auto;" /> --- # Other risk metrics... & approaches - sensitivity to **carbon taxes** (default à la Merton) - exposure to a long-short *carbon* portfolio ( `\(\beta\)`) - vulnerability to climate events / **disasters** Other facets: - .green[**opportunities**] (avoided .brown[emissions], .green[green] revenues from patents) - aggregation (scoring, weighted averages, etc.) --- class: inverse, center, middle <style type="text/css"> .remark-slide-content { font-size: 32px; padding: 1em 4em 1em 4em; } </style> # Second issue: objectives & constraints --- # "Simple" benchmarks (1/2) <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .font80[[Jondeau, Mojon & Pereira da Silva (2021)](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3987186): *Building Benchmarks Portfolios with Decreasing Carbon Footprints*] .pull-left[ .font90[**Starting point**: emission intensities are not uniformly distributed across firms:] .font90[**Issue**: sector & geographical tilts. But **benchmarks** require traditional exposures!] ] -- .pull-right[ <img src="intensity.png" width="650" /> ] --- # "Simple" benchmarks (2/2) <div style="width: 100%;float: bottom;margin-top:-30px;margin-bottom: 50px;"> .font80[**Targeted Reinvestment**] `\(\rightarrow\)` .font80[smart approach, **BUT**:] <div style="width: 100%;float: bottom;margin-top:-35px;margin-bottom: 50px;"> - .font80[does the world need to stick to **business as usual** (!) benchmarks?] - .font80[is it desirable to maintain sector tilts? Why maintain .brown[**inertia**]?] .font80[If anchoring is so important, then maybe we need **new** .green[benchmarks].] - .font70[Passive investing made the cap-weighted indices the anchors, but ETFs can change that] `\(\rightarrow\)` .font70[new **framing** for practitioners & academics. Why not be more ambitious: ] <div style="width: 100%;float: bottom;margin-top:-25px;margin-bottom: 50px;"> <img src="benchmarks.png" width="750" style="display: block; margin: auto;" /> <div style="width: 100%;float: bottom;margin-top:-25px;margin-bottom: 50px;"> --- # What's the limit? <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .font90[=> How far can we go without hurting performance .grey[(significantly)]?] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .pull-left[ [.font80[Pedersen, Fitzgibbons & Pomorski JFE 2020]](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3466417): .font80[the **ESG** efficient frontier] <img src="frontier.png" width="600" style="display: block; margin: auto;" /> ] .pull-right[ <div style="width: 100%;float: bottom;margin-top:150px;margin-bottom: 50px;"> .font90[There seems to be a .bluen[sweet] spot.] But here: overall .bluen[**ESG**] & ESG `\(\neq\)` Emissions <div style="width: 100%;float: bottom;margin-top:60px;margin-bottom: 50px;"> It's likely this remains true for .brown[**emissions**]... ] --- # Back to optimization <div style="width: 100%;float: bottom;margin-top:-30px;margin-bottom: 50px;"> .font80[[Le Guenedal & Roncalli (2021)](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3999971) propose a penalized quadratic optimization scheme with particular **constraints**. The agent is endowed with a .brown[**brown**] footprint that she wants to reduce by a factor] `\(\small \mathcal{R}\)`: <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .font80[ - focused on turnover, volatility / tracking error (.brown[**benchmarks**] again!) - accomodates for any footprint metric & many constraints: ] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> <img src="eq.png" width="690" style="display: block; margin: auto;" /> --- # But, in short <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .font70[With this machinery, you can control & fine tune everything... except **uncertainty**! **Estimation** versus .bluen[realization], **in-sample** versus .bluen[out-of-sample] performance.] <div style="width: 100%;float: bottom;margin-top:10px;margin-bottom: 50px;"> .font70[ | Indicator (firm) | Short term prediction (1Y-5Y) | Long term prediction (10Y+) | | --- |:---:|:---:| | Return | Hard | Hard | | Volatility / correlations | So so | So so | | Emissions / intensities | Easy | So so | ] <div style="width: 100%;float: bottom;margin-top:-20px;margin-bottom: 50px;"> .font80[**Question**: how can we build reliable long-term granular emission forecasts?] .font70[Pathways, trajectories: not enough chronological depth? NLP, scrapping annual reports?] .font80[Moreover, meta-meta analyses suggest that, on the long run, greenness is not related to average realized returns. Hence, it should suffice to optimize on .bluen[risk] and .green[sustainability]. (whereas in MPT, it's risk vs return) ] --- # Open research questions Preferences drive demand; demand drives .bluen[**AUM**]... .bib[ <img src="conclusion.svg" width="890" style="display: block; margin: auto;" /> ] .font90[**Question**: what's the .grey[(actual/aggregate?)] **appetite** for greenness and how can we calibrate these trade-offs? ] --- # Thank you! **Going further**: - [Perspectives in sustainable equity investing](http://www.esgperspectives.com) - [Thierry Roncalli's course on sustainable finance](http://www.thierry-roncalli.com/SustainableFinance.html) <div style="width: 100%;float: bottom;margin-top:100px;margin-bottom: 50px;"> .center[ .font180[Time for **Q&A**!]] --- # Bibliography .bib[.font60[ - [Avramov, Cheng, Lioui & Tarelli 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3711218): *Sustainable investing with ESG rating uncertainty* - [Berg, Kolbel, Rigobon 2022](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3438533): *Aggregate Confusion: The Divergence of ESG Ratings* - [Berk & van Binsbergen 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3909166): *The impact of impact investors* - [Coqueret & Vu 2022](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4002705): *ESG news spillovers to (and from) the supply chain* - [Derrien, Krüger, Landier & Yao 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3903274): *ESG news, future cash flows, and firm value* - [Dimson, Marsh, Staunton 2020](https://jpm.pm-research.com/content/47/1/75): *Divergent ESG Ratings* - [Gabaix & Koijen 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3686935): *In search of the origins of financial fluctuations: The inelastic markets hypothesis* - [Gibson, Krueger & Schmidt 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3433728) : *ESG Rating Disagreement and Stock Returns* - [Jondeau, Mojon & Pereira da Silva 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3987186): *Building Benchmarks Portfolios with Decreasing Carbon Footprints* - [Koijen & Yogo 2020](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2537559): *A Demand System Approach to Asset Pricing* - [Le Guenedal & Roncalli 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3999971): *Portfolio Construction with Climate Risk Measures* - [Lioui & Tarelli 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3878314): *Chasing the ESG factor* - [Pastor, Stambaugh & Taylor 2020](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3498354): *Sustainable investing in equilibrium* - [Pastor, Stambaugh & Taylor 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3864502): *Dissecting green returns* - [Pedersen, Fitzgibbons & Pomorski JFE 2020](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3466417): *Responsible Investing: The ESG-Efficient Frontier* - [Sautner, van Lent, Vilkov & Zhang 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3642508): *Firm-level Climate Change Exposure* - [van der Beck 2021](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3929359): *Flow-Driven ESG Returns* ] ]