Firms' Sustainability Investments

The externality framing (why this chapter exists)

Private vs. social return

What: When a firm spends to reduce its environmental impact, the value it captures is typically smaller than the value to the world.

Why: That gap is the externality — the defining market failure of the environment. Absent carefully designed policy, the firm's incentive to abate is too weak relative to the social optimum.

Remember: A firm can have a private surplus loss while social value is much larger; that wedge is exactly why we need policy. Going green is generally assumed (loosely) to be more expensive, but the framework still works if it isn't.

Five benefit channels (2 direct + 3 indirect)

The channels

Remember:

• DIRECT (1): Reduced fossil-energy spend (switch gas → biomass/geothermal/electrification; Gosnell et al. Virgin Atlantic single-engine-taxi fuel savings).
• DIRECT (2): Reduced regulatory/tax/Emissions-Trading-System (ETS) compliance cost or access to subsidies — e.g. under the EU ETS you need a permit per ton of CO₂; emit less, pay less.
• INDIRECT (3): Better labor recruiting/retention (if workers care about sustainability, your talent pool improves).
• INDIRECT (4): Lower cost of capital (if markets penalize high-CO₂ firms, your financing cost depends on your carbon intensity).
• INDIRECT (5): Increased consumer demand (some consumers pay more for green products).

Why: Whether each channel materializes depends entirely on how consumers, investors, and workers respond to firm sustainability. (Side nugget: domestic electricity / own solar can also insulate a firm from energy-market supply shocks — a supply-chain resilience benefit.)

Three challenges to answering "does sustainability pay?"

The challenges

Remember: (1) Measurement — how do you even measure a green firm? (2) Causal inference — can you predict that cutting Scope 1 by Y% lowers capital costs by Z, free of confounders? (3) Impact — once market responses are netted out, does the investment actually help the environment (or does the externality / unintended-consequences problem swallow it)?

The core confounder: managerial ability

What: Well-managed firms tend to have both higher ESG scores and lower capital costs / lower climate-risk exposure (within the same sector and country).

Why: A naïve correlation ("more sustainable firms pay lower bond yields") can't tell whether ESG caused lower yields or whether a sharp new CEO raised ESG and improved spreads. Correlation ≠ causation.

Remember: Always ask "what is driving the variation in sustainability I observe — and could that same thing affect yields directly?" Other confounders raised in class: sector (oil & gas firms have low ESG but lots of collateral, which independently affects financing), and fixed-asset intensity.

Measuring a green firm: ESG ratings

What ESG is

What: ESG = Environmental, Social, Governance scores, meant to function like credit ratings — an index of a firm's "greenness."

Why: Investors want a clean green-firm signal; reality is messy because sustainability is genuinely hard and multidimensional to measure.

Remember: Example construction — KLD (now MSCI) used binary yes/no indicators for environmental concerns (e.g. hazardous-waste liabilities >$50M; substantial fines for waste violations; among top producers of ozone-depleting chemicals) and strengths (e.g. notably strong pollution-prevention programs), then summed them. Don't dismiss these tools as worthless — measuring sustainability is important and hard.

Why ESG ratings diverge (Berg et al. 2022)

What: Cross-agency correlation is far lower for ESG than for credit ratings, because creditworthiness rests on a century of standardized financial data and reflects probability of default, whereas ESG components are hard to observe and standardize.

Remember (key numbers):

• Credit-rating cross-agency correlation ≈ 0.99.
• ESG cross-agency correlation ≈ 0.54 (as of a 2022 study).
• Same firm can get a 0 from Sustainalytics but ≈ −4 from MSCI and +3 from KLD — positive correlation, but no consensus yet.
• Decomposition of the disagreement: Measurement ~56% (how you measure a given indicator, e.g. worker happiness via turnover vs. lawsuits) is the largest source; Scope ~38% (which indicators count — e.g. Sustainalytics rates animal welfare and forests; others don't); Weights ~6%.
• The environmental dimension generally correlates higher across agencies than social or governance; some governance scores (KLD vs. S&P) are even negatively correlated.

Measuring a green firm: carbon emissions (Scope 1/2/3)

Definitions

What:

• Scope 1: direct emissions from the firm's own production at its sites.
• Scope 2: indirect emissions embodied in purchased electricity, heating, cooling.
• Scope 3: indirect emissions in purchased materials and their transport (a coffee shop's paper cups). One firm's Scope 3 is another firm's Scope 1 and 2.

Why: Emissions look like a more tractable, specific measurement than full ESG.

Remember (cross-provider Pearson correlations):

• Scope 1 ≈ 0.95–0.99 (very high) — partly mechanical, since a disclosed number is shared by all providers; when estimated (not disclosed) it drops to ~0.84–0.87.
• Scope 2 ≈ 0.85–0.95 when reported; falls to ~0.5–0.8 when estimated (it's already indirect).
• Scope 3 is the hardest: correlation as low as ~0.22 (e.g. Trucost vs. CDP). The harder to measure, the lower the agreement.

Total vs. intensity — a surprise

Remember: Even within Scope 1, total emissions vs. emissions per dollar of sales (intensity) correlate only ~0.6; for Scope 2 only ~0.24. A firm can have high absolute emissions but low intensity (or vice versa). So "is emissions a good measure of a green firm?" has no clean answer — and there's no obviously correct choice between level and intensity, which matters enormously for the capital-cost results (see Ch. 6, Bolton & Kacperczyk).

Are emissions even the right impact measure?

Remember: Measuring emissions perfectly still may not capture counterfactual impact. "Green compared to what?" — across sectors (steel vs. software) and even within sectors (specialty vs. galvanized steel making products that aren't substitutes). And general-equilibrium effects: shutting natural-gas production (1 British thermal unit of gas ≈ 40% less CO₂ than the same energy from coal) might just raise coal use and make particulates and CO₂ worse. Law of unintended consequences again.

The green-patent paradox (Cohen et al.)

What: A counterintuitive correlation — the heaviest-emitting firms often hold the most clean-tech patents (even citation-weighted). The top-50 green-patent list includes ExxonMobil, Shell, BP, ConocoPhillips, Chevron.

Why: Several possible explanations, none confirmed: market pressure pushing polluters to prepare for a green future; liquidity/scale (oil majors have huge collateral and research teams, fewer financing constraints); and technology complementarities (carbon capture and storage builds on extraction/pumping expertise; hydrogen on refining). Probably not primarily greenwashing here (these are real, cited patents).

Remember: A 1 standard deviation (SD) higher Sustainalytics environmental score ⇒ ~24% LOWER likelihood of green patenting. Takeaway: high carbon ≠ "not green" — interpret this correlation carefully.

Hedonic / product-level valuation (preview)

Remember: Measuring sustainability at the product level (the regular vs. "bio" version of the same grocery item) is far cleaner than at the firm level. Hedonic valuation methods estimate consumer demand for individual product attributes, including sustainability — an alternative to firm-level measurement. (Developed further in Ch. 6 with the FSC and Dolphin-Safe studies.)

Reading treatment/control graphs (the cross-cutting exam skill)

Recipe

What: The causal effect at any horizon = (actual outcome) − (counterfactual/control outcome) read straight off the graph at that time point — NOT (after − before) for the treated group alone.

Why: Before/after on the treated unit alone confounds the treatment with every other trend.

Remember: Read the gap between actual and counterfactual short-run (just after treatment) and long-run (end of chart) separately — answers come as two numbers, e.g. "(−40, +10)". "Negligible" short-run effect = the two lines sit on top of each other right after the intervention. Trap: a treated series that falls after treatment can still mean zero/positive effect if the counterfactual fell just as much (the whole point of BP and Dolphin-Safe in Ch. 6). This is listed in the Part-1 study guide as a calculation.

One-line takeaways

• A firm's private return to going green is usually below the social return — the externality is the reason this chapter exists.
• Five benefit channels: 2 direct (energy spend, compliance/subsidies) + 3 indirect (labor, capital cost, consumer demand).
• ESG ratings barely agree (0.54 vs. 0.99 for credit); disagreement is mostly measurement (~56%), then scope (~38%), then weights (~6%).
• Scope 1 emissions agree across providers (~0.95–0.99); Scope 3 can fall to ~0.22; total vs. intensity is itself only weakly correlated.
• Managerial ability is the master confounder — well-run firms have both higher ESG and lower capital costs.
• Green-patent paradox: +1 SD environmental score ⇒ ~24% fewer green patents; big polluters hold many clean-tech patents.
• Always read causal effects as actual − counterfactual, never as treated-group before − after.