Hermès (RMS): Entry Assessment

Assessment date: 8 August 2026

The moat was judged in Layer 1 and found intact. This document takes that as given and asks the second question only: whether the current price is an attractive entry, and on what terms. It never reopens the moat, and Layer 1 never looked at price.

Snapshot

Field	Reading
Trigger	Drawdown through the Layer 1 alert threshold of 40%
Drawdown	~43% from the February 2025 high
Price	~€1,678 against €2,957 peak
Trailing P/E	~38.9x on FY2025 EPS of €43.1 (peak: ~68x)
Market cap / EV	~€177bn / ~€164bn (net cash €12.9bn)
Moat	Established in Layer 1, 8 August 2026. Exceptional strength, intact condition, clean on every condition
Implied growth at current price	~11 to 12% earnings compounding sustained for a decade
Recommendation	Deploy two thirds of target weight; one third reserved for the pre-2018 tail
Reserve trigger	~27x trailing ≈ €1,165 (−31%), conditional on a fresh Layer 1 revision at that level
Next checkpoint	Next Layer 1 revision 22 October 2026 on Q3 revenue, or sooner on any named event

Decision in one paragraph. The shares have fallen roughly 43% from the February 2025 high on flat earnings, so the decline is almost entirely multiple compression, from a bubble peak near 68 times trailing to about 39 times, which is the floor of the regime that has governed since 2018. Layer 1 reads the moat clean on every condition, including a resale premium measured directly, which places the rationed formats at 2.47 to 3.12 times retail. At €1,678 the market prices roughly 11 to 12 percent earnings compounding sustained for a decade, which is continuation of the disclosed model rather than acceleration, and the reserve tranche price embeds roughly 7 percent. The recovery mechanism requires only that the moat holds and earnings compound; multiple reversion shortens the wait rather than creating the return. Two thirds deployed at the modern floor, one third reserved for a regime break that would itself demand a fresh Layer 1 revision before deployment.

The spine of the case

The recovery thesis does not lean on the market being irrational, on any hidden fact, or on out-guessing the sell side. It runs on one mechanism: if the moat is intact and stays intact, earnings keep compounding, and a compounding franchise recovers its price even if the multiple only holds, and recovers it faster if the multiple also reverts. Layer 1 establishes that the franchise is intact now. The valuation section below establishes that the price paid does not require the multiple to expand for the thesis to work. Whether the moat stays intact over the years ahead is judged in Layer 1, which is re-read on every disclosure and on any named event. This document does not monitor the moat. It decides the entry and nothing more.

One qualification is carried from the start. The spine has three links and only the first is verified. The moat is intact. Whether earnings compound is a separate question, and in the half just reported they did not: basic earnings per share printed €21.36 against €21.43, on a currency headwind above €360m. The cause is translation rather than operations, but the spine runs on reported earnings and those stood still. Currency is carried below as an explicit assumption instead of a footnote.

Data sources: Hermès H1 2026 half-year report and FY2025 key figures, market price and multiple history, an independent year-end P/E series (2001 to 2024), the resale premium measurement series established 7 August 2026 across three vendors in three regions, Bain Altagamma spring 2026 and Bain China reports, recent sell-side notes (Jefferies, Bernstein SocGen, Kepler Cheuvreux, Barclays, J.P. Morgan), ECB policy statements and June 2026 projections, and Global Blue tourism data. Each is tagged by evidence tier as in Layer 1, and systematic series are weighted above spot observations wherever both exist.

Section one: the trigger

The move. The shares trade at approximately €1,678 against a February 2025 high of roughly €2,957 (tier one, price data): a drawdown of approximately 43%, through the Layer 1 alert threshold of 40% and into the zone where the deep work begins. The half-year report of 29 July took the shares to roughly €1,500 in the days that followed, from which they have recovered to the current level. The market marked the print down and then took most of that mark-down back within two weeks, which tells us something about the seller, and Section Three returns to it. The decline came as a grind rather than an event: a first leg through mid-2025, a partial recovery, then a second, steeper leg into 2026 culminating in a fall of as much as 14% intraday on the Q1 print of 15 April.

The relative dislocation. This is not an idiosyncratic Hermès collapse. Kering and LVMH fell in sympathy, and the press frames the move as the market questioning whether luxury as a whole has peaked (tier two). Subject to the retest below, a large part of this is sector-wide de-rating rather than company-specific impairment, the profile Layer 1 flagged as the most common source of a Hermès dislocation. One qualification is carried honestly from the outset: the Q1 print itself was a genuine miss (5.6% organic against roughly 7.1% consensus, down from 9.8% the prior quarter), so part of the fall is a repricing of the never-disappoints premium, not pure sentiment. Section Three weighs this.

The multiple compression. At roughly €1,678 against FY2025 earnings of €43.1 per share, the trailing multiple is approximately 38.9 times (tier one on both inputs). The peak was far higher than a casual read suggests: the 14 February 2025 all-time high of roughly €2,957, divided by trailing FY2024 earnings of roughly €43.6 per share, was approximately 68 times, essentially the same nosebleed level as the 2021 spike. So the drawdown ran from about 68 times to about 39 times, a near-halving of the multiple. A point that must be stated honestly rather than softened: across this specific window earnings were roughly flat (FY2024 €43.6, FY2025 €43.1), so the roughly 43% price fall was almost entirely multiple compression, not a case of the multiple falling while earnings grew underneath. That cuts against one easy comfort and reinforces a more important reading: an unchanged business de-rating from a genuine bubble peak back toward its post-2018 floor is valuation normalisation, not fundamental impairment. Whether roughly 39 trailing is that modern floor or a mid-range level is the central valuation question of Section Four.

The trigger tells us only that the work is warranted. Nothing in this section is a view on the moat, and the verdict comes from Section Two.

Section two: the moat, established elsewhere

The moat is not judged here. That was done in the Layer 1 assessment of 8 August 2026, which worked through every condition against the H1 2026 half-year report and found all of them clean, reading exceptional on strength and intact on condition. That revision has to come first: no Layer 2 is written for a company whose Layer 1 has not been revised, and the revision itself is written without any reference to price.

The readings that bear most directly on what follows are stated once, without argument, and can be checked in that document.

Core leather, the métier carrying the mechanism, grew 9.8% at constant currency in the half and accelerated into the second quarter from 9.4% to 10.2% while the group decelerated. Supply cadence is unchanged at the disclosed six to seven percent and is now corroborated by two figures the company cannot easily flatter: workforce up 5.5% year on year and capital expenditure up 8.9%. Gross margin rose to 71.1% of revenue from 70.7%, with no discounting, wholesale or outlet channel anywhere in the disclosure, which is the opposite of what a house defending volume would show. The apex resale premium is now measured directly across 173 observations at three vendors in three regions, placing the rationed formats at 2.47 to 3.12 times retail against a trigger at 1.5 times. Governance is unchanged.

Two of the risks Layer 1 tracks cannot be read on the timescale a quarterly report runs on: the slow drift of apex signalling away from the leather object, and a gradual miscalibration of the cadence. Layer 1 keeps both as standing uncertainties rather than measured conditions, because neither has an actor, a date or anything observable to read. Neither shows a near-term signal, though the multi-year migration indicators worsened slightly. They are carried the same way here, as things to watch rather than things currently readable.

The regional picture is carried forward here because the valuation section depends on it. Group revenue grew 6.1% at constant currency and 1.6% as reported on a currency headwind above €360m. The Americas grew 15.3%, Japan 11.0%, Europe excluding France 8.8%, Asia excluding Japan 2.4%, and the region containing the Middle East fell 4.2%. Both regions where an external cause was identified recovered sequentially within the half: the Middle East from minus 5.9% to minus 2.4%, and France from minus 2.8% to plus 6.2% on returning tourist flows. That is the cleanest available corroboration of the attribution, and it is the reason Section Three treats the decline as predominantly exogenous.

On that basis the assessment proceeds to price. Everything below assumes the moat holds and asks only what the market is charging for it.

Section three: the anatomy of the decline

With the franchise intact, the second question is why the shares fell roughly 43%, because understanding the seller tells us whether the price is likely to be wrong. Four forces, none of them the moat, plus one that partially is.

Currency. A translation drag above €360m in the half turned 6.1% constant currency growth into 1.6% as reported, and turned mid-single-digit underlying growth into flat earnings per share. At this magnitude it is more than an optics problem feeding a nervous tape. It is a live input to the return, because the recovery mechanism runs on reported earnings and those did not compound in the half (tier one).

Sector de-rating. The luxury complex fell together; Hermès is sold indiscriminately through sector baskets in such episodes, as Layer 1 warned (tier two).

China proxy selling. A meaningful holder cohort trades Hermès as a liquid China proxy; the soft Asia-excluding-Japan print gave them the reason.

Multiple normalisation from an extreme. The peak was roughly 68 times trailing (€2,957 against FY2024 EPS of about €43.6), a bubble level matching the 2021 spike and requiring near-perfection. Because earnings were roughly flat across the window, essentially the entire fall is this extreme unwinding, from about 68 to about 39 trailing. A drawdown that is almost pure de-rating of an unchanged franchise, rather than a response to falling earnings, points to a seller repricing the multiple and not the business.

The fifth force, owned honestly: a real miss. Q1 came in at 5.6% organic against roughly 7.1% consensus, the first genuine disappointment in years, and the half at 6.1% against an internal expectation of seven to eight. Part of the seller is rationally repricing the never-misses premium. That part is a fundamental repricing rather than a mispricing, and the valuation section does not treat the entire fall as sentiment. One further piece of evidence arrived after the print and cuts in the thesis's favour. The market marked the shares down to roughly €1,500 on the half-year report and has since taken most of that back. A seller repricing a franchise does not reverse course within two weeks, whereas a seller reacting to a headline does, which suggests the marginal seller is still mostly mechanical and sentiment-driven without being entirely so. The sizing reflects that the fall is not one hundred percent noise.

Section four: the financial position

A deployment decision needs the balance sheet and the cash generation stated plainly, because they determine what the tail scenarios actually cost. A franchise that can be waited out is a different asset from one that cannot.

The balance sheet removes financing risk from the question. Net cash stands at approximately €12.9bn against no meaningful debt, on shareholders' equity of roughly €19bn. There is no refinancing wall, no covenant, and no scenario in which a weak year forces a decision. That matters more than it appears in a document about a drawdown: the lower tail below is dead money on a compounding franchise rather than impairment, and the balance sheet is why.

Margins are at the top of the disclosed range and did not contract. Recurring operating margin printed 41.0% for the half, level with the full year 2025 and 40 basis points below the prior half, which the company attributes to currency. Gross margin rose to 71.1% from 70.7%. A sell-side projection of roughly 100 basis points of contraction from fixed-cost underabsorption, as the capacity build ran ahead of decelerating revenue, did not materialise. That risk is closed and the earnings denominator is firmer than that scenario assumed.

Cash generation requires decomposition, and the headline overstates it. Adjusted free cash flow of €2.18bn in the half was up 18%. But operating cash flow before working capital was €2,694m against €2,733m, a decline of 1.4%, and the change in working capital swung from minus €403m to plus €6m, a movement of €409m that is larger than the entire €335m increase in free cash flow. The company attributes the swing to inventory management and sell-through, which is a genuine and favourable operational signal and direct corroboration that allocation is clearing. But it is a base effect against a weak comparison and should not be extrapolated. Underlying cash generation was flat, and the improvement sits on top of it as timing.

Returns on capital remain in the band that justifies the multiple. The combination of 41% operating margins, minimal debt, and capital expenditure of €344m against revenue of €8.2bn produces returns comfortably above the twenty percent floor Layer 1 uses as a quality threshold. The capital intensity of the workshop programme is real but small relative to the cash it generates.

Capital allocation is the one open question, and it reads as a question rather than a concern. Dividends paid in the half were €1.9bn, roughly €18.1 per share, down from €26.1 in the prior half as the exceptional element of the previous distribution was not repeated. Buybacks are immaterial: 94,846 shares for €160m, or 0.09% of shares outstanding. Cash therefore accumulates. For a business earning high returns on the capital it does deploy, holding €12.9bn in net cash is a drag on returns, and the family's preference for balance sheet strength over distribution is a known feature rather than a discovery. It is noted because it caps the return in a scenario where the multiple does not move, and because it is one of the few things about this company that could change without the moat changing.

Section five: valuation

The current multiple. Roughly €1,678 against €43.1 of FY2025 earnings is about 38.9 times trailing. The business behind it: 41.0% recurring operating margin, gross margin expanding to 71.1%, €12.9bn net cash, and a Layer 1 that reads clean on every condition.

Reverse DCF, what each price assumes. This is the primary instrument in the section, because it states what the market is charging instead of forecasting what the business will do. Illustrative, on an 8.5% discount rate and 2.5% terminal growth; all inputs disputable and stated for that purpose.

At €1,678 the price implies roughly 11 to 12 percent earnings compounding sustained for ten years. For calibration, the €2,957 peak implied high-teens compounding, which is acceleration the model cannot deliver at a six to seven percent capacity cadence. The reserve tranche price near €1,165 implies roughly 7 percent, below the group's demonstrated rate in almost every year of the modern record.

The entry therefore poses a narrow and answerable question: is 11 to 12 percent something this business can deliver? Volume cadence is disclosed at six to seven percent. Price increases run at six to nine percent annually and were absorbed in January 2026 with gross margin rising rather than falling. Mix has historically added. Eleven to twelve percent is therefore inside what the disclosed model produces, without requiring anything the company has not already done. That is a far weaker claim than the peak required.

At 68 times the price required acceleration, at 39 times it requires persistence, and at 27 times it would require no more than adequacy.

The floor scenario. The most useful way to size the downside is a single conservative construction rather than a probability-weighted table: assume the multiple never recovers, and assume earnings grow only at the disclosed volume cadence with no contribution from price or mix at all.

At six percent earnings growth and an unchanged multiple, the return is six percent plus roughly one percent of dividend. That doubles the position in approximately ten years. An index compounding at ten percent doubles in roughly seven.

Two features of that construction matter.

It is conservative on two counts, and deliberately so. Six percent is the volume cadence alone, and earnings growth is volume plus price plus mix, which historically ran higher. And an unchanged multiple assumes the market never reprices a franchise it has repeatedly refused to value on a mechanical growth multiple: on a pure six percent growth assumption a discounted cash flow would support something closer to the mid-twenties, yet none of the three modern bottoms went below 35 times, in periods when growth was also weak. The market has consistently paid a premium to the mechanical number for the durability behind it.

The honest cost here is time rather than capital. The floor scenario is not a loss at all but a seven percent compounding return on a business that cannot be forced into a decision by its balance sheet, set against an index that may do better. Three years of underperformance is the price of being wrong here. That is the risk being taken, and it should be read as such rather than as a downside argued away.

The opportunity cost is real and current. Indices sit at record highs while this franchise sits 43 percent below its own. If the index continues from here and Hermès holds the modern floor, the position lags for years without anything having gone wrong.

What the entry is actually betting on is speed rather than size of return. At an unchanged multiple the doubling takes ten years. At the top of the range the multiple has held outside spikes, the expansion alone contributes materially and the doubling comes years earlier. The asymmetry lives in duration rather than in magnitude, and it sits on top of a downside bounded by three historical bottoms rather than by an argument.

The entry does not depend on multiple expansion, which is the spine of the case restated in DCF terms.

EPS bridge (analyst assumptions, pending full model). FY2026E EPS around €44 to 45 on revenue up 7 to 8% at constant currency, with the H1 margin dipping toward 39.5 to 40% on under-absorption before recovering as the workshops season. FY2027 to 2030: revenue up 8 to 9%, being capacity of roughly 6 to 7% plus price and mix, margin normalising around 41%, and EPS compounding at 9 to 10% to reach about €65 by 2030.

The drawdown record, with trough multiples:

Episode	Peak-to-trough drawdown	Approx. trough trailing P/E
2001, dotcom	~47%	~24
2002	~50%	~19
2008, financial crisis, first leg	~45%	~23
2009, financial crisis, second leg	~50%	~24
November 2010	~28%	~34
2015 to 2016	~22%	~31
2018	~27%	~35
COVID, 2020	~30%	~35
2021 to 2022	~43%	~38
Current, 2025 to 2026	~43%	~39

The two-regime framing, stated without the statistical overclaim. The record divides into a pre-2018 regime (troughs mid-twenties to low-thirties, centring about 27) and a post-2018 regime (troughs about 35 to 38, centring about 36), the re-rating underpinned by margin rising from about 30% to 41%, fifteen further years of compounding proof, and, critically, a low-rate environment. Three post-2018 bottoms (2018, 2020, 2021 to 2022) each found support at 35 to 38 and none revisited the old lows. This is a pattern and a plausibility rather than a probability, since three observations do not make a base rate. What the pattern does establish is the cost of the wrong benchmark. Anchoring on pre-2018 cheapness would have refused all three modern bottoms and forfeited the large runs that followed each. Demanding pre-2018 cheapness in a post-2018 world has been a systematically punished error.

The live threat to the regime floor. One pillar of the post-2018 regime, low rates, has reversed in direction if not yet in level. The ECB cut through the early months of 2026, then reversed after the Middle East conflict drove an energy shock, raising all three key rates by 25 basis points in June, its first increase since 2023, taking the deposit facility rate to 2.25%. The Governing Council has since described the pause at that level as conditional on an energy shock whose full inflationary impact has yet to play out, and its June projections revised 2026 headline inflation up to 3.0% while cutting euro-area GDP growth to 0.8%. High-growth multiples are compressing and sell-side targets have been cut across the board, €2,000 at Jefferies and €1,700 at Barclays.

The direction and the cause matter more than the level. A 2.25% deposit rate is low by any historical standard, so this is not a high-rate environment in the sense that phrase usually carries. What has changed is that the cutting cycle stopped and reversed. That is worth weighing carefully, because the record does not support the easy worry that a tightening backdrop breaks the floor. Two of the three modern bottoms formed against exactly that. Hermès fell about 27% in 2018, in the middle of the Fed's 2016 to 2019 hiking cycle, and found support around 35 times. It fell about 43% into the summer of 2022, in the middle of the 2022 to 2023 hiking cycle, and found support around 38 times. In both cases the multiple held through the tightening and then re-rated. The modern floor has already survived two rate-rising cycles rather than none.

What is genuinely different this time is the cause rather than the direction. The 2018 and 2022 cycles tightened into growth. This one tightened into a supply-side energy shock, and the ECB paired the increase with a downgrade to euro-area growth. That combination, weak growth alongside rising rates, is the stagflationary case, and stagflation has historically been harder on luxury than either strong growth or clean disinflation. So the rate cycle is a real risk to the floor, but through its stagflationary character, not through the simple fact of tightening, which the franchise has weathered twice before.

Separately, Kepler Cheuvreux projects roughly 100 basis points of H1 margin contraction from fixed-cost underabsorption as the capacity build runs ahead of decelerating revenue, so the earnings denominator itself is softer than in the prior episodes. The stagflationary rate backdrop and the softer denominator together raise the probability of the lower tail, a re-rating toward the pre-2018 trough of about 27, implying a price near €1,165, roughly 31% below current, relative to a naive read of the three prior bottoms. The weight on that tail is smaller than an earlier draft assumed, because the tightening itself is not the threat the two prior hiking-cycle bottoms already answered.

The transmission mechanism is worth naming, because it is not the one that usually worries a luxury investor at all. A rising discount rate does not reach Hermès through its customers, who are not rate-sensitive in any way that shows up in a Birkin waitlist. It reaches the shares through the multiple. A franchise trading at 39 times trailing is a long-duration asset whose value sits predominantly in terminal cash flows, and terminal cash flows are arithmetically sensitive to the rate used to discount them regardless of how the business itself is performing. That is why the regime floor is under test while every operating gauge in Section Two passes, and why the tail scenario is written as dead money on an intact franchise rather than as impairment.

Why the entry does not depend on winning the regime argument. This is what makes the spine hold. The recovery does not require the multiple to expand, or even to be defended at 39. If the moat holds and earnings compound at high single digits, the price recovers on earnings growth alone at an unchanged multiple, and recovers far faster if the multiple reverts toward the modern range. The regime question therefore governs the pace and shape of the return rather than whether there is one. The single scenario that breaks the recovery is the moat failing, not a lower multiple, because only then do the earnings stop compounding. This is why the lower tail is sized as a reserve conditional on the moat still passing, and why a fall through the pre-2018 trough is read as a possible moat signal rather than a simple discount.

Why deployment still follows, the asymmetry of errors. Since the regime question cannot be resolved in advance, and every bear case at every prior bottom also sounded structural at the time, the decision rests not on predicting the regime but on comparing the two ways of being wrong. Error one, buy and the regime breaks: the position marks down roughly 31% to the pre-2018 trough on a franchise still compounding earnings, with the position sold only if a fresh Layer 1 revision finds the moat broken, a bounded and recoverable error. Error two, abstain and the moat holds: the recovery that has followed every prior verified moat drawdown is missed entirely, historically an unbounded and unrecoverable error of omission. The expected cost of error two dominates the expected cost of error one across any reasonable weighting, and that asymmetry, not a claimed base rate, is what justifies deploying. The live rate and margin threats do not change whether to deploy; they set how much is committed now and how much is reserved for the tail.

On probability-weighted scenarios. This section does not carry a probability-weighted table of outcomes. Assigning probabilities to a regime holding or breaking would present a judgment as an input, and the weighted figure it produced would do argumentative work its construction could not support. Computing an internal rate of return would additionally require an exit multiple, which is a forecast of what the market will pay in 2031 and precisely the kind of claim this file otherwise refuses to make.

What stands in its place is narrower and easier to defend. The reverse DCF states what the market is charging today, which is a fact about the price rather than a forecast. The floor scenario states what happens if nothing goes right except the moat holding, which is bounded by the disclosed cadence rather than by an assumption. And the drawdown record states where the multiple has found support before, which is history rather than probability. Those three together answer the deployment question without requiring a distribution nobody can estimate.

Valuation verdict. Against the regime that has governed since 2018, roughly 39 times trailing is at the modern floor; against the risk that the rate cycle breaks that regime, it is a floor under live test. The quality benchmarks behind that multiple are intact: 41.0% margin, gross margin expanding, returns on capital above the twenty percent threshold, net cash, and a Layer 1 that reads clean on every condition. There is no fundamental erosion justifying a discount to the modern regime, though a turning rate cycle could impose one anyway. The honest reading: attractive, with a fatter lower tail than a naive drawdown read would assign, which is a sizing instruction, not a rejection.

Section six: the plan, two tranches

The plan is a framework fixed in the calm of the analysis, not a live order. It rests on the spine: the moat is intact, so the entry follows from the dislocation. Whether the moat stays intact is judged in Layer 1, not here. Deployment is split into two tranches.

The two tranches

Tranche one, the majority, now. Justified directly by the spine: the moat is intact on the H1 data, earnings are compounding at 9.8% core leather, and the price sits at the post-2018 regime floor during the largest drawdown since the dotcom crash. The dislocation is the opportunity and it is open only while the mispricing is live, so waiting for confirmation means waiting for the discount to close. Roughly two thirds of the intended position is deployed here, across the −40% to −50% band rather than at a single tick, since the alert threshold at −40% and the modern floor around −45% define a zone rather than a point.

Tranche two, the reserve, for the pre-2018 tail. Roughly one third is reserved for the single lower-probability scenario that would offer a materially better entry: a re-rating toward the pre-2018 trough of about 27 times, a price near €1,165, roughly 31% below current and roughly 61% below the 2025 peak. This tranche is deployed at that level only after a fresh Layer 1 revision, written at that price and without reference to it, still comes back clean on every condition. A fall through the pre-2018 trough would more plausibly signal the moat breaking than a simple de-rating, so the condition is a full re-reading rather than a confirmation, never mechanical on price. It is the smaller share precisely because it is the lower-probability branch.

Exit This document sets no selling price. A target multiple years out is a forecast of what the market will pay, and nothing here supports that kind of claim. The position is held while the moat holds, and whether the moat holds is judged in Layer 1.

One exit is defined, and it is the only one. If a Layer 1 revision, whether scheduled or triggered by an event, moves the moat from intact to broken, the position is sold regardless of price, because the earnings compounding the whole recovery depends on has stopped. Nothing short of that is a reason to sell. A disappointing quarter that leaves Layer 1 clean is not a reason to sell, because a soft print with the mechanism intact is exactly the cyclical noise the two-layer split exists to separate from real damage.

A note on what this document is and is not. This is written for the author's own investment process and published in that form. It reflects the analysis, assumptions and judgment of the author at the date of writing and nothing more. Nothing here is investment advice, a recommendation tailored to any reader, or an offer to buy or sell any security. Valuation figures, scenarios and price levels are illustrative analyst assumptions unless explicitly sourced to company reports or other primary data. Investing involves risk, including loss of capital, and past performance is no guarantee of future results. Any reader considering an investment should do their own work and consult a qualified adviser who knows their situation. The author may hold, or may come to hold, positions in the securities discussed.