Workspace Group: A 7% Yield Hiding a Special Situation

The 7% Yield Is Not the Thesis

Yield at publication

7.18%

Base sustainable DPS growth

2.0–5.0%

central 4.0%

Discount to EPRA NTA

47.1%

FY2026 NTA of 687p

Dividend

Reduced

The 7% yield may not survive the next twelve months. That is not the thesis.

Workspace Group owns and runs flexible offices across London. At the reference price of 363.4p its FY2026 dividend of 26.1p is a yield of 7.2%. That dividend has just been cut. Occupancy is near its lowest level in years, earnings are falling, larger London landlords are close to full, £420m of cheap debt matures in 2027 and 2028, and a new management team is asking shareholders to fund a refurbishment-led repair. Management has also said that trading profit after interest will step down substantially in the year to March 2027, and the dividend is now set at 1.2 times earnings cover. Another cut is plausible.

That is the description of a dividend trap, and parts of it may be accurate.

Beneath the dividend sits something else. Workspace owns about £2.1bn of London property, and its shares trade at about 0.53 times reported net tangible assets, the company’s own accounting value for that property after debt. If the repair works, earnings and the dividend recover from a lower base. If it does not, running the whole portfolio at poor returns becomes hard to justify, and selling assets or returning capital becomes the better use of it. Workspace is best understood as an asset-backed turnaround with a dividend attached, which management wants to turn into an income-led dividend growth business.

Four economic paths matter to the thesis, and the asymmetry only makes sense with all four in view. They are not alternatives a reader has to choose between: Workspace could repair part of the estate while selling another part of it.

• A — Turnaround. Refurbishment and product changes work. Occupancy and rent recover, earnings recover, and the dividend grows again after a trough. This is roughly our base case.
• B — Shrink to greatness. Weaker assets are sold, the proceeds repay debt or come back to shareholders, and what remains is a better portfolio. Absolute earnings can fall while value per share rises.
• C — Broader asset realisation. If reinvestment cannot earn adequate returns while the shares stay far below realisable value, a larger programme of sales and capital returns becomes the economically rational path. That is an option, not a prediction.
• D — Value destruction. The turnaround fails and capital keeps being spent at poor returns anyway. Disposals fund capex and dividends rather than per-share value, and rent roll, NAV and the dividend all fall. This is the outcome that matters most, because it uses up the value in B and C before shareholders can capture it.

The dividend sequence to expect is a cut, a trough, a recovery and then growth. Not a 7% yield that starts growing now.

What Workspace Owns and Operates

Workspace has about 4 million square feet across roughly 59 London locations, mostly character buildings let on flexible terms to around 4,000 small and medium-sized businesses, creative firms and young companies. It is landlord and operator at once: it owns most of the property it runs, which separates it from flexible-office operators that lease their buildings.

Customer contracts are short, typically two to three years with frequent break options. That cuts both ways. Rents can reprice quickly, space can be reconfigured, and no single customer matters much. But occupancy can fall quickly too, cash flows are less locked in, switching costs are low, and the product has to stay competitive all the time.

The advantages are owned London real estate, scale, an operating platform built around smaller businesses, and a long record of buying older buildings and re-letting them at higher rents. They are real, but they are not a moat in the infrastructure sense, and they do not protect a product that falls behind. On current evidence this is a potentially durable business under repair, not a demonstrated compounder.

What Went Wrong

Stabilised occupancy, historically around 89–91%, was 81.6% at March 2026. Stabilised rent per square foot fell 2.1% in the year to £46.31. Net rental income fell about 7% to £113.4m, trading profit after interest about 9% to £60.5m, and EPRA net tangible assets per share about 11% to 687p. The dividend went from 28.4p to 26.1p.

The per-share record over the full cycle is weaker still. Between FY2020 and FY2026 three things moved against shareholders together: net rental income fell from about £122m to £113.4m, administrative costs rose from about £15m to around £20m, and finance costs rose from about £23m to £31m. Adjusted underlying earnings per share were about 40.6p in FY2019 and 31.3p in FY2026. Asset values fell as well. NTA per share was 1,086p in FY2019, so the share price has not simply derated against an unchanged portfolio.

Demand at the top of the funnel weakened. Enquiries fell from about 11,000 in FY2022 to about 8,400 in FY2025, roughly 23%. Lettings in FY2026 held close to the prior year because more enquiries converted, but the first quarter of FY2027 did not extend that improvement: conversion from enquiry to letting was 14%, against 15% a year earlier.

The likely causes are inferred rather than measured: historical underinvestment in the product, stronger competition from better-quality flexible and managed space, customer expectations that have moved toward communal space, hospitality, furnished offices and better fit-out, and particular weakness in larger units. The company describes a flexible office market with “ongoing recalibration of occupier requirements”.

Capital allocation did not help. The McKay acquisition in 2022 added property exposure and debt shortly before property yields and interest rates rose. That does not show it was irrational when it was made, but it amplified the fall in NTA per share that followed.

Why the London Office Market Is Not the Whole Explanation

The obvious explanation is a London office downturn, but the evidence suggests that cannot explain Workspace’s underperformance on its own.

For the year to March 2026, British Land reported 96.9% portfolio occupancy, growth of 4.9% in estimated rental value, the rent a valuer judges the space could command rather than the rent currently collected, like-for-like net rental growth of 6%, and Central London office take-up at its highest level in twenty years. Landsec reported 98.0% occupancy, the highest in two decades, 98.6% in offices, estimated rental value up 6.4% and 7.1% in offices, and like-for-like net rental income up 6.0% in its office-led portfolio. In the same year Workspace’s occupancy was 81.6%, its net rental income fell about 7% and its dividend fell about 8%.

These are not like-for-like businesses. British Land and Landsec own larger, more traditional and generally higher-quality offices on longer leases, and their occupancy is not a target for Workspace. The comparison answers a narrower question: does the evidence suggest that London office demand itself is broken? It does not. A meaningful part of Workspace’s problem is its own, and the market is right to apply a company-specific discount.

That is adverse evidence, and it is also where the opportunity comes from. A company whose weakness is specific and partly repairable does not need its market to recover. Workspace needs to lift stabilised occupancy from about 82% into the mid-80s in a city where demand for good space is strong.

The Dividend: Cut, Trough, Recovery

Dividend per share

1. 8.79

   FY2012

2. 9.67

   FY2013

3. 10.63

   FY2014

4. 12.04

   FY2015

5. 15.05

   FY2016

6. 21.07

   FY2017

7. 27.39

   FY2018

8. 32.87

   FY2019

9. 36.16

   FY2020

10. 17.75

    FY2021

11. 21.50

    FY2022

12. 25.80

    FY2023

13. 28.00

    FY2024

14. 28.40

    FY2025

15. 26.10

    FY2026

The ordinary dividend rose from 8.79p in FY2012 to 36.16p in FY2020, was roughly halved to 17.75p in FY2021, recovered to 28.40p by FY2025 and was reduced to 26.10p in FY2026. It has been cut twice in six years, and there is no current run of increases.

Over the full fourteen years available, the dividend compounded at about 8.1% a year. That rate includes the pre-Covid growth, the Covid cut, the recovery and the FY2026 reduction, which is why it is shown over the whole period rather than a shorter one, and it is not a forecast.

Reconstructed results and the dividend they supported

£m, except per share figures in pence	FY2012	FY2013	FY2014	FY2015	FY2016	FY2017	FY2018	FY2019	FY2020	FY2021	FY2022	FY2023	FY2024	FY2025	FY2026
Ordinary DPS (p)	8.79	9.67	10.63	12.04	15.05	21.07	27.39	32.87	36.16	17.75	21.50	25.80	28.00	28.40	26.10
Adjusted underlying EPS (p)	—	—	—	—	—	30.6	36.8	40.6	44	21.3	26.2	29.4	34	34.4	31.3
Dividend cover	—	—	—	—	—	1.4x	1.3x	1.2x	1.2x	1.2x	1.2x	1.1x	1.2x	1.2x	1.2x
Net rental income	—	—	—	—	—	79.2	95.6	111	122	81.5	86.7	116.6	126.2	122.1	113.4
Trading profit after interest	—	—	—	—	—	50.7	60.7	72.4	81	38.7	46.9	60.7	66	66.8	60.5
Loan to value	—	—	—	—	—	13.0%	23.0%	22.0%	21.0%	24.0%	23.0%	33.0%	33.0%	34.0%	35.0%

The dividend has followed earnings rather than being defended against them. When trading profit after interest fell about 52% in FY2021, the dividend fell about 51%. In FY2026 earnings per share fell about 9% and the dividend about 8%. That is bad for anyone who needs a stable income, and good evidence that distributions have not been funded with debt. Nothing in the record suggests past dividend growth came mainly from a rising payout.

In FY2026 adjusted underlying earnings were 31.3p, the dividend 26.1p, cover 1.2 times and the payout about 83%. The policy is now that the total dividend is covered at least 1.2 times by adjusted underlying earnings per share. At that cover, the dividend can only grow as fast as earnings per share.

The headline yield is not the normalised yield. If earnings per share fall in FY2027 and cover stays at 1.2 times, the dividend and its yield at the reference price would be:

We do not know where the trough will be. Our estimate of the normalised starting yield is roughly 5.5–6.6% at the reference price, not 7.2%.

The Turnaround and the KPIs That Decide It

Management’s plan is called Fix, Accelerate, Scale. Its stated aim is “an income-led, dividend growth business”, and the first priority is rebuilding occupancy: low-risk refurbishment, a new Managed product alongside space-only lettings, better systems, pricing and sales, and continued capital recycling. The medium-term ambition is more than £125m a year of trading profit before interest, organically, against about £91.7m in FY2026. That is an increase of about 36%, and our base case does not assume it.

The first quarter of FY2027 is the earliest evidence. Stabilised occupancy rose 0.6 points to 82.3%, rent per square foot 0.9% to £46.87, and the stabilised rent roll 1.9% to £110.2m. One quarter is a signal, not a result.

The dividend is the last link in two chains, one operating and one financial.

1. Enquiries
2. Conversion and retention
3. OccupancyMultiplied by rent per square foot
4. Rent roll
5. Net rental income
6. LessAdministrative costs
7. Trading profit before interest
8. LessFinance costs
9. Trading profit after interestDivided by the share count
10. Adjusted earnings per sharePaid out at 1.2 times cover
11. Dividend per share

1. Asset salesOperating cash flow
2. LessMaintenance and strategic capex · Debt repayment
3. Net debt and refinancing cost
4. Balance-sheet capacity
5. Dividend capacity

Each case is built down those chains, from occupancy and rent to the dividend. The FY2031 column for each case:

Base. Occupancy climbs gradually to about 85.5% by FY2031 and rent reaches £50–51 a square foot. The pilot projects earn around 12–15% and are rolled out selectively. Trading profit before interest reaches about £110m, finance costs about £37m, trading profit after interest about £73m, and adjusted earnings about 38p a share. At 1.2 times cover that supports a dividend of about 31–32p. A plausible path runs 26.1p, then 22–23p, 25p, 27p, 29p and 31–32p: a cut, a trough and a recovery that together compound at about 3.8% a year over five years. Once the dividend has normalised, we estimate sustainable growth of about 2–5%, centred on 4%.

The base case does not require a return to 90% occupancy, management’s £125m target, acquisitions, meaningful buybacks, margin expansion, cheap refinancing or a return to NTA. It does require several moderate things to happen together: a partial recovery in occupancy, modest nominal rent growth, acceptable realised returns on refurbishment, manageable refinancing and no major further fall in property values. It is neither easy nor inevitable.

Bear. Occupancy stays around 80–82% and rent at £46–47. The pilots earn poor returns and rollout is limited. Trading profit before interest is £85–95m, finance costs rise to £38–43m, trading profit after interest is £42–57m and earnings are about 26p. The dividend falls to about 21–22p and does not grow. The balance sheet survives with little flexibility.

Bull. Occupancy reaches 89–90% and rent £54–56. Refurbishment earns 15–20% or more, trading profit before interest reaches £125–140m and finance costs fall to £31–35m. Earnings of about 51p support a dividend of about 42–43p, growth of roughly 10% a year from FY2026. Bull needs strong execution on several variables at once.

Three stress tests sit alongside them, and it is worth separating the three kinds of case in this report. The four paths above describe what may ultimately happen to the business and its capital. Bear, base and bull are the operating outcomes we value. These are tests of particular ways the recovery could stall or fail, not extra valuations:

Delayed Recovery. Occupancy stays at 81–83% until FY2028 or FY2029 before moving toward 85%. Trading profit after interest is £50–62m, earnings 26–32p and the dividend 22–27p: cut or frozen before growth resumes, with little internal capacity for large capex.

No Recovery. Occupancy about 81%, rent £46–47, trading profit before interest about £90m, finance costs about £40m, trading profit after interest about £50m, earnings about 26p and a dividend of about 21–22p, a yield of about 5.9% at the reference price. The dividend growth thesis fails here, and the asset paths become more important.

Structural Damage. Occupancy falls to 75–78% and rent to £43–45. Enquiries keep falling, better conversion fails to offset them, and representative London assets start to sell well below book. Trading profit before interest is £75–80m, finance costs £40–45m, trading profit after interest £30–40m, earnings about 16–21p and the dividend about 13–17p, or 3.6–4.7% on the reference price. The current dividend is clearly unsustainable.

How Safe Is the Dividend?

The business appears capable of supporting a dividend through a normal cycle. The current 26.1p is not assured, and the case against it is stronger than the payout ratio suggests.

At about 193 million shares the dividend costs roughly £50m a year. Operating cash flow after interest was about £63m in FY2026. Cash dividends paid in the year were about £55m, the difference being timing.

The weak point is capital expenditure. Repairs and maintenance charged to the income statement run at about £4–5m a year. Capital spent on the properties has been much larger and volatile, averaging roughly £50–60m a year over the longer record, and the company does not split it between spending that keeps buildings competitive and spending that creates new income. We estimate the capitalised part that is economically maintenance at about £5–10m a year beyond expensed repairs. That is our estimate, not a disclosed figure, and it could be wrong in either direction.

On that estimate the dividend and economic maintenance together need about £55–60m a year, against roughly £63m of operating cash after interest. Headroom before any growth investment is thin. If more of the refurbishment programme turns out to be competitive maintenance, the headroom disappears, the payout ratio was measured against the wrong denominator, and project returns are overstated. That is the strongest version of the dividend-trap argument, and the evidence does not yet settle it.

The thresholds we will use, which are research judgments rather than company guidance: operating cash flow after interest above about £70m is comfortable; £55–70m is a watching brief; persistently below about £55–60m, the current dividend is economically unsafe. For growth the line is adjusted earnings of about 31p a share, because 31p at 1.2 times cover is about 26p of dividend. FY2026 earnings were 31.3p, so there is no near-term room for growth until earnings recover.

At current earnings, Workspace can fund only limited growth investment internally while it pays the dividend. Material refurbishment spending needs retained cash, disposals or debt, which is why asset recycling matters to the income case and not only to the asset case.

The most obvious version of a dividend trap, a high payout on high leverage in a deteriorating business, does not describe Workspace: leverage is moderate, the dividend has been cut rather than borrowed, and the evidence does not point to broken London demand. The subtler one, hidden maintenance capital combined with lasting competitive weakness, remains open.

The Balance Sheet and the 2027–28 Refinancing

At March 2026 net debt was £758m, loan to value 35%, interest cover 3.6 times and net debt about 8.1 times EBITDA. Of the debt, 77% was fixed, at an average cost of 3.6%. By the first quarter net debt was £740m, with £260m of cash and undrawn facilities. Covenants leave substantial room: loan-to-value limits are generally 60%, 65% for the green bond, and interest cover limits generally about 1.75–2 times.

Solvency is not the central issue. Earnings are.

Two cheap instruments mature close together: £120m of private placement notes at about 3.19% in August 2027, and a £300m green bond at about 2.25% in March 2028. Together they cost about £10.6m a year. Refinanced in full at around 5% they would cost about £21m, roughly £10m a year more, against £60.5m of trading profit after interest. The base case assumes some debt is repaid first and finance costs settle at £35–39m. Moderate loan to value does not mean low earnings risk.

Disposals reduce the refinancing need and remove rental income at the same time. Workspace sold £125.7m of property in FY2026 at 7.2% below the most recent book value, and £12.6m in the first quarter of FY2027 at 22.3% below. More than £200m is being marketed. Whether those sales add to value per share depends on the prices achieved and on where the money goes.

Management and the Capital Allocation Test

Chief executive Charlie Green and chief financial officer Tom Edwards-Moss joined in February 2026. Green co-founded and led The Office Group, now Fora; Edwards-Moss held senior finance and chief executive roles at Hibernia Real Estate Group. Their record at Workspace is months long, which is too short to establish skill.

Their early decisions are consistent with discipline: acknowledging product shortcomings, testing refurbishment on four buildings before any rollout, pausing or reviewing some projects, selling assets, reducing debt, and cutting the dividend to restore 1.2 times cover. None of that is yet evidence of returns.

The company’s longer record is mixed. Its asset management has generally been better than its corporate capital allocation. A 2018 placing of about 16.3 million shares raised roughly £176m; the McKay acquisition in 2022 issued about 10.5 million shares and added substantial debt; capital expenditure was heavy throughout; and there was no significant buyback before the current disposal and deleveraging cycle.

Targets are not returns. The four pilot buildings and management’s targets for them:

Together they cover about 102,500 square feet, around 2.7% of the portfolio, for £16–20m of capital. To earn the targeted yields they need roughly £2.7–4.35m of additional annual rent. These are management’s targets. Completion is expected within 12 to 18 months of the first quarter update, so no realised returns exist yet.

The measure that matters is realised incremental yield: the actual increase in net rental income divided by the total capital invested, including rent lost during the works. A valuation uplift or a rise in gross rent does not answer the question. The framework we will apply, which is ours and not a company hurdle: realised returns above about 15% make a strong case for rolling the programme out; 10–15% supports selective reinvestment; 7–10% has to be weighed against repaying debt or buying back shares at a discount; below about 7–8% the rollout should probably stop.

The real test of this management team is what it does if the answer is poor.

Why Half of NTA Is Not Automatically Cheap

At the reference price the shares trade at about 0.53 times FY2026 EPRA NTA of 687p, a reported discount of about 47%. That is not the same as buying the buildings at half price.

Reported NTA is a valuation of a portfolio that is still being operated. It is not what the buildings would fetch if sold, and it is not what would reach shareholders after debt, costs and time. It has also been falling: 1,086p in FY2019, 800p in FY2024, 774p in FY2025 and 687p in FY2026. Recent disposals have been below book. Allowing for that, our estimate of economic NAV, meaning what the property is worth per share after debt and other claims on our figures rather than the accounts’, is about 550–650p a share, with a central working area around 600–620p, held with medium confidence.

A simple sensitivity shows how quickly the gap narrows. It starts from £2,133m of investment property, deducts £758m of net debt, about £47m of other net liabilities and an assumed £50m of realisation costs, and divides by about 193.3 million shares:

These are calculations on assumptions, not forecasts of what a sale would achieve. Against the reference price of 363.4p, the simplified asset backing is used up at a portfolio discount of roughly 27% to book.

Time matters as much as the discount. A realisation takes years, and value reaches shareholders only after debt and costs are paid. Achieved within about three years at 10–15% below book, the gap to the share price would be a meaningful return. Stretched to five or seven years, or achieved at 20% below book, time consumes most of it. At 25–30% below book little or nothing is left, however quickly it happens. Rent collected during a sale programme would help; values falling while it runs would not.

The NAV discount alone is not enough. What gives it value is the price achieved, the speed, and what happens to the cash along the way.

The Special Situation Under the Dividend

The asset paths are not extra upside to add to the dividend case. They are alternative uses of the same capital. The base case assumes the portfolio keeps operating and partly repairs itself, and its value comes from dividends and from the price the market pays for the recovered income. Broad realisation matters most in the world where that repair fails. Adding the two together would count the same buildings twice.

Path B, as an illustration. Suppose Workspace sold £500m of property at 10% below book. Of the £450m received, £300m repays debt and £150m goes back to shareholders, about 78p a share. Net debt falls from £758m to about £458m, and simplified loan to value from 35% to about 28%. But those buildings were earning rent. At a 6% yield about £30m of income goes, against about £15m of interest saved at 5%, so earnings fall. Whether value per share rises depends on the price achieved and on whether capital is returned below what the shares are worth. Size is not the objective. Value per share is.

Path C becomes rational if the pilots and the wider programme cannot earn adequate returns while the shares stay far below realisable value. At that point reinvesting in the whole portfolio stops being the best use of shareholders’ capital. Nothing requires it to happen, and it would not be quick or cheap. In rejecting Saba’s proposals the Board argued that they assumed disposals at narrower discounts than recent sales suggest are achievable, and described the investment market for its buildings, “non-conventional character buildings, largely in secondary locations and with short-term income”, as slow by nature.

Path D is the one to watch for: poor realised returns, continued heavy spending, disposals that fund capex and the dividend rather than per-share value, and a portfolio that loses value while it is being repaired. That combination damages the income case and the asset case at the same time, and it is the most important way this thesis fails.

Saba and the Pressure on Capital Allocation

Saba Capital built a stake of 29.11% of voting rights by July 2026, just below the 30% level at which UK takeover rules require a mandatory offer. It argued publicly for an accelerated disposal programme, which the Board described as a wind-down, and requisitioned resolutions to replace non-executive directors. At the AGM in July shareholders re-elected the Board, and none of Saba’s six nominees received more than 13.3% support.

Saba does not guarantee liquidation, a takeover, buybacks or sales at NAV, and its first attempt to change the Board failed decisively. What its position does create is a large shareholder focused explicitly on the gap between the share price and asset value. The inference is limited: it makes it less likely that failed reinvestment could continue indefinitely without challenge. Saba is pressure on capital allocation, not the thesis.

Valuation & Scenarios

Workspace is two things at once, and each is worth something on its own. It is a business that earns money from letting offices, and it is the owner of the buildings it lets. Valuing the earnings alone ignores the property behind them; valuing the property alone ignores whether the business can earn an adequate return on it. So each case is valued twice, once on what the business earns and once on what its property is worth, and the two carry the same weight. Neither lens is the senior one.

The useful part of the result is that the two land close together in every case: within about 13p of each other in the bear case, 3p in the base case and 34p in the bull case. The valuation is therefore not particularly sensitive to the 50/50 split between them, which would matter much more if earnings and assets were telling opposite stories. It still rests on the two lenses themselves, on the multiples and on the asset values behind them.

Asset realisation is in none of them. It is the alternative to running the portfolio, not an addition to it.

Base, in full. Occupancy reaching 85.5% and rent of £50–51 a square foot produce trading profit after interest of about £73m, or 38.0p a share of earnings. At 13 times earnings, each 1p of annual earnings is valued at 13p, which gives 494p a share. That multiple is roughly a 7.7% earnings yield: what a repaired but capital-intensive flexible-office landlord earns, rather than what a compounder earns.

The same path leaves economic NAV at about 686p a share by FY2031: still our estimate rather than the accounts’, and still not a liquidation value. We value it at 0.725 times, meaning the shares are valued at only 72.5% of our estimated economic NAV, a 27.5% discount. Something like that discount survives in every case, for the reasons the market already applies one: short leases, a business that has to keep spending on its buildings, and a capital-allocation record still to be proved. That gives 497p. The base case therefore does not assume the discount to asset value disappears; it assumes it narrows from today’s level and stays wide.

Where that 686p comes from. It is not a forecast of reported NTA, and it is not built property by property: we have no rent forecast for each building. It starts from the external valuation of the portfolio at March 2026, £2,133m, and moves that figure with the two things that set any property valuation, the rent a building can earn and the yield a buyer demands on that rent, before taking off what is owed.

• What gets sold comes out first. Recycling continues, so about £250m of March-2026 book value is assumed to have left the portfolio by FY2031, removed at book. The discount those sales achieve, about 12% in the base case, changes the cash that comes back and therefore the debt, not the value of what is still owned.
• Rent moves the valuation up. The FY2026 valuation rested on an estimated rental value of £48.94 a square foot against £46.31 actually achieved, and that relationship is held constant. Base-case rent of £50.5 becomes an estimated rental value of £53.37, about 9% above FY2026. Occupancy is not a separate line here: it is what has to happen for that rent to be earned, and it is already in the case.
• The property market is assumed to give almost nothing back. The yield a buyer demands, the equivalent yield, is a divisor: at a higher yield the same rent buys a lower valuation. It moves from 6.66% at FY2026 to 6.70%, marginally worse. This is a rent recovery, not a repricing of London property.
• Capital spending is not added to value. Refurbishment counts through the rent and the yield it produces, which is where the first two steps already capture it. Adding the spending on top would count the same money twice.

That leaves about £2,041m of property. Net debt is assumed to fall from £758m to about £675m, paid down by disposal proceeds and retained cash and consistent with the £37m of interest this case already pays, and about £40m of other net liabilities come off with it. What remains is £1,326m of equity across 193.3 million shares: 686p.

The 686p is numerically almost identical to the 687p of EPRA NTA Workspace reported for FY2026, and the resemblance is worth naming only to set it aside. One is our estimate of economic value five years out; the other is the company’s accounting measure today. What Workspace will report as NTA in FY2031 is not something we have forecast, and nothing here should be read as saying it stays flat.

The comparison that does work is against economic NAV on the same definition today, which we put at 550–650p. On that basis the base case is a real improvement, and a partial one: higher rent lifts the portfolio, while disposals take part of the asset base away, financing costs more and property yields give almost no help, so an operating recovery does not turn into a large expansion of asset value. A repaired rather than restored portfolio.

Bear is the failed version of the same calculation: yields out to 7.25%, more selling, and debt that barely moves. Bull needs £55 a square foot and a modest tightening to 6.50%.

None of those figures is what the shares would be worth. They are the value of the property equity; the price-to-NAV multiple above is what turns that into a market value.

Averaging the two lenses gives about 496p a share in FY2031, and the five years of dividends along the way add about 136p.

What that means for a buyer at 363.4p. The base case is to collect about 136p of dividends over the five years and still hold a share worth about 496p at the end: about 631p of value for each 363.4p invested. Under our return convention that is an annualized total return of about 11.7%. The dividends are counted once, as received during the five years and not reinvested; they are not part of the 496p; and the annualized figure spreads the whole five-year gain evenly across the five years rather than tracking the date each dividend arrived. Nothing about that path is smooth: it starts with a cut and a trough.

Bear is a failed recovery rather than a slower base. Occupancy stays near 81%, earnings settle at 26.0p, and 11 times gives 286p; economic NAV falls to about 471p and 0.58 times gives 273p. The average is about 280p, with about 107p of dividends. Bull needs 89–90% occupancy and repeatable refurbishment returns: 51.5p at 14 times is 721p, economic NAV of about 888p at 0.85 times is 755p, and the average is about 738p with about 166p of dividends.

The reference price is 363.4p. These are values per case, not price targets, and the returns are calculated from that price.

Why the dividend is a cross-check rather than the valuation. Capitalising each case’s FY2031 dividend gives 287p in the bear case, 485p in the base case and 708p in the bull case: close enough to the two lenses above to be a useful consistency check. The bull case is the loosest fit, about 30p below, which is what happens when the asset side does more of the work than the dividend does. It is not a third opinion, and it carries no weight in the valuation, because the dividend is set from those same earnings at 1.2 times cover. Counting it again would value one stream of earnings twice. A persistent 7% terminal yield instead of 6.5% would put the base case at 450p, which is why the case does not depend on a low terminal yield. The dividends in the table are a different thing again: cash received between FY2027 and FY2031, added once to the terminal value and not reinvested.

What NAV does and does not do here. Every case keeps a discount to economic NAV, so none of them requires the market to close the gap. The gap narrowing further is possible upside that the valuation does not claim.

The strongest bear case. Occupancy of 89–91% is no longer normal. Customers have moved permanently toward better offices, older buildings need far more recurring capital to stay relevant, and refurbishment returns overstate growth because much of the spending is maintenance. Short leases make the cash flows less secure, refinancing raises costs while earnings are depressed, and the dividend and NAV keep falling. Even with NAV above the share price, a large permanent discount is deserved for mediocre returns on equity, high capital needs, volatile earnings, an unreliable dividend and unproven management. On that reading today’s price and yield are rational, and much of the supporting evidence is already visible.

What the price appears to assume. A dividend of 21–22p, our No Recovery outcome, is a yield of about 5.9% at the reference price. A dividend discount model is too sensitive to value this company, but as a cross-check, a 22p dividend at the reference price implies perpetual growth of about 2.9% at a 9% required return and about 3.9% at 10%, so the price does not assume zero growth under every reasonable input. Our inference, held with medium confidence, is that the price is broadly compatible with permanently depressed dividend capacity and does not require recovery.

1. 01

   The weakness looks partly company-specific, in a market that is not broken.

   British Land and Landsec are close to full with rents rising, while Workspace sits near 82%. That gap is a reason for the discount, and it is also the reason recovery does not require a better London office market. Workspace needs to fix its own product, and the first quarter of FY2027 moved occupancy, rent and rent roll in the right direction.

2. 02

   The base case needs 85.5% occupancy, not a return to 90%.

   Occupancy was 89–91% for most of the company’s history. The base case assumes about 85.5% by FY2031, rent of £50–51 a square foot and trading profit before interest of about £110m, well below management’s own target. A price consistent with a 21–22p dividend does not give credit for that partial repair.

3. 03

   The discount gives capital allocation a second route if repair fails.

   If refurbishment cannot earn adequate returns, selling assets, repaying debt and returning capital become the rational alternative to reinvesting, and a 29% shareholder is pressing for exactly that. The value of that route depends on sale prices and time, and it disappears if capital is spent at poor returns first.

Dividend Outlook

	FY2031 dividend per share	Yield on cost	Sustainable growth once normalised
Bear	21p–22p (central 21.5p)	5.8–6.1% (central 5.9%)	No growth; the dividend is cut
Base	31p–32p (central 31.5p)	8.5–8.8% (central 8.7%)	2.0–5.0% (central 4.0%)
Bull	42p–43p (central 42.5p)	11.6–11.8% (central 11.7%)	High single digits

The dividend in the first column is the one that case reaches in FY2031, written by the research as the endpoint of a scenario rather than obtained by compounding today’s 26.1p forward at a steady rate. The base case cuts the dividend first, troughs, and only then recovers to 31–32p. Yield on cost divides that dividend by the reference price of 363.4p: what the shares would pay each year on the price paid today, if the case holds. It is not an expected total return, which the valuation section sets out.

The last column is a different claim again: how fast the dividend could grow once the business has normalised, after the trough and the recovery, which is not the FY2026 to FY2031 path. Where the research put no number on it, its own words are shown instead. Workspace reports, trades and pays its dividend in sterling and its shares are quoted in pence, so these yields need no conversion; they are gross, and what an investor receives depends on residence and account type.

Key Risks

Ordered by how much each would change the conclusion.

1. Capital intensity and hidden maintenance. The largest analytical uncertainty. If much of the refurbishment spending is what the buildings need simply to stay competitive, distributable cash flow and project returns are both overstated, and the dividend is less covered than it looks.

2. Execution and competitive position. Peers show strong London demand while Workspace underperforms. The repair depends on Workspace fixing its own product, and one quarter of better numbers does not show that it can.

3. Capital allocation after failed pilots. The worst outcome is continued investment after the evidence shows inadequate returns. It damages the dividend and the asset backing together.

4. Refinancing. £420m of low-cost debt matures in 2027 and 2028. Materially higher interest costs, or less debt repaid beforehand, would reduce dividend capacity directly.

5. Property values. NTA has already fallen by more than a third since FY2019. A further large fall would reduce or remove the asset protection, especially without matching debt reduction.

6. Disposals diluting earnings. Sales can lower leverage while shrinking the rent roll, and recent sales have been below book.

7. Flight to quality. Customer expectations may permanently require more recurring capital than the business has historically spent.

8. Short leases. Cash flow can deteriorate faster than it would under long office leases.

9. The activist. Saba creates pressure, not realised value, and the Board has so far prevailed.

Another dividend cut is a plausible part of the path rather than a separate risk. Direct technological disruption is not a material thesis risk on current evidence, though demand from young and small companies could be affected indirectly.

What Would Prove Us Wrong?

Concrete enough to check against reported figures.

On the dividend:

• Operating cash flow after interest persistently below about £55–60m without the dividend being adjusted.
• The dividend maintained through debt or underinvestment rather than internally generated cash.
• Stabilised occupancy still around 81–82% after FY2028 despite the refurbishment spending.
• Rent per square foot falling while occupancy also stays weak.
• Realised returns on refurbishment persistently below about 8%.
• Refinancing leaving trading profit after interest per share structurally below about 31p.
• The payout raised materially while cash generation is weak.

On the mispricing and the asset case:

• Representative stabilised London assets repeatedly selling 15–25% below book.
• Enquiries still declining after the product improvements and marketing changes.
• Peers staying 95–98% occupied with rising rents while Workspace stays around 81–82%.
• Refurbished buildings failing to outperform comparable unrefurbished ones.
• Weakness in large units spreading into the core base of smaller units.
• Property values falling about 30% from FY2026 levels without sufficient debt reduction.
• Heavy reinvestment continuing after poor realised returns.

A further dividend cut on its own would not break the thesis. Several of the points above happening together would.

Conclusion

Workspace is not a high-quality dividend compounder today. It is an asset-backed turnaround whose dividend may fall before it grows.

It has underperformed its London peers by a wide margin, it has not compounded value per share over the last cycle, its dividend has been cut twice in six years, and the true capital cost of keeping its buildings competitive is unknown. None of that should be read past.

What makes it interesting is that its outcomes run along several economically distinct paths. A partial operating repair can support meaningful income and moderate dividend growth while the shares still trade at a discount to their own asset value. A successful refurbishment programme can do considerably more. A failed one can make realising assets the rational use of capital, at a price well below reported NTA. And a failed programme combined with continued spending can destroy the asset protection that makes the other paths valuable.

So the central question is not whether the 7.2% yield is safe. It is whether Workspace can earn adequate returns on the capital required to repair its portfolio, and, if it cannot, whether that capital will be returned or realised instead.

We hold business quality at medium confidence, dividend safety at medium, dividend growth at medium-low, the mispricing diagnosis at medium and the valuation at medium. Those are separate judgments and should not be averaged into one.

What we will be watching, in the order the evidence usually arrives: stabilised occupancy, enquiries and conversion, retention in units below about 3,000 square feet, rent per square foot, and the realised returns on the four pilot buildings. Those move first. Trading profit before and after interest per share, the 2027 and 2028 refinancing, disposal prices against book, NTA and the dividend declaration confirm later what they implied. Future updates will ask which of the four paths the evidence is moving toward.

Research Status & Updates

No updates since publication.

Sources

Disclosure

Author

Javier AnguloIndependent Researcher

Research completed

17 Sept 2026, 15:40 UTC+02:00

Published

18 Sept 2026, 03:52 UTC+02:00

Reference price

GBX 363.4017 Sept 2026, 16:35 UTC+01:00 · Closing price, London Stock Exchange

Methodology

2026-01

Valuation basis

Each case is valued on two lenses and blended in equal weight: normalized FY2031 earnings at a terminal multiple, and FY2031 economic NAV at a terminal price-to-NAV that stays below one. Dividend capitalisation is a cross-check carrying no weight, because the dividend is set from the same earnings at 1.2 times cover. The cumulative dividends received over the five years are added to the terminal share value, not reinvested. Asset realisation is an alternative path and is never added to a case. Values are presented per case, not as price targets.

Author position

No positionDirection only. Position size is not recorded.

Issuer compensation

None received from the issuer

Shown to the issuer

Not shown before publication

Other material conflicts

None disclosed

Prior research on this issuer

None in the preceding twelve months.

This is general research, not personalised investment advice. It does not take account of any individual reader’s objectives, financial situation or needs. Every price shown is the price observed at the time of the analysis and is never updated afterwards.

Unpriced receives no payment from the companies it writes about. The author’s dealing policy and the timing rules that apply to it are published in the repository as TRADING_POLICY.md.