Offices vs Stocks: Modeling Vacancy Rates vs Stock Drawdowns

Real estate and the stock market both punish you for assuming the last data point repeats forever. The difference is the type of pain they deliver. Stocks drop fast, usually with a headline attached. Offices empty out slowly, usually with a lease expiry calendar and a landlord’s optimism.

I’ve spent enough time around both asset classes to develop a healthy suspicion of any model that pretends vacancy rates and stock prices behave the same way. They don’t. But you can still steal ideas from stock drawdown modeling and apply them to office vacancy risk. You just have to translate the mechanism, not the math copy-paste style.

Below is how I think about modeling office vacancy rates versus stock drawdowns, why the usual approaches break, and what I’ve found works better in practice for decision-making across offices, warehouses, shops, and even the broader universe of condominiums, landed houses, strata houses, shophouses, and factories.

Why “drawdown” makes sense even when you’re staring at vacancy

A stock drawdown is the peak-to-trough drop in price over a period. Investors care because it’s the part that wipes out risk budgets and forces sales at the worst possible time. Vacancy risk is similar in spirit, even though it’s not a single “price” number.

In property land, the trough is usually not empty units by themselves. The trough is the combination of:

rent softening,
concessions and incentives,
higher tenant downtime or longer re-letting periods,
valuation compression,
and, if you’re unlucky, financing terms that become less friendly.

So when you model vacancy, you’re implicitly modeling a drawdown of income and value, not just physical occupancy.

Here’s the translation I use: an office vacancy rate behaves like a “slow-motion drawdown driver.” It doesn’t usually gap down overnight, but it can create a longer, uglier decline that makes the eventual recovery feel much further away than the raw vacancy percentage suggests.

The core mismatch: discrete events versus continuous drift

Stocks often move because of discrete events. Earnings disappointments, rate surprises, regulatory shocks. Even if the market has been creeping, the big move typically coincides with something that changes expectations immediately.

Vacancy, in contrast, is heavily path-dependent and calendar-driven. Lease expiries are discrete. Renovation cycles are discrete. But the “damage” spreads through time as the market digests those expiries, then pricing adjusts, then demand shifts.

That means a model that works for stock drawdowns, where volatility and correlations are central, needs reinterpretation for vacancy.

If you run a naive office vacancy model that assumes stationarity, you get a rude surprise. Office demand can change structure. A new transit line can improve one submarket and do nothing for another. A policy shift can make certain tenant needs irrelevant. A URA master plan 2025 wave of new supply can raise vacancy in a way that doesn’t revert quickly, even after leasing stabilizes.

In other words, office vacancy behaves like a system with seasons and inflection points, not just random noise.

What vacancy actually is, and what it isn’t

“Vacancy rate” sounds tidy. It isn’t.

Depending on how you collect the data, vacancy could mean:

unoccupied leased space,
available space marketed for lease,
space that is vacant but still under some arrangement,
or an occupancy ratio based on net rentable area, which can be updated after renovations.

If you’re comparing different buildings or cities, these definitions can diverge enough to create fake patterns. I’ve seen vacancy datasets that quietly blend “vacant” with “effectively unavailable,” which can inflate the perceived stress during a redevelopment cycle.

Also, vacancy is not purely supply and demand. It’s operational. Tenant improvements take time. Fit-outs delay re-leasing. Ownership may keep a unit empty longer to avoid locking into weak terms. A landlord can “manage vacancy” like a dial, not just endure it.

So when you borrow from stock drawdown logic, you have to keep one foot in reality: vacancy is an outcome of leasing behavior and timing, not a direct measurement of tenant pain alone.

A practical mental model: vacancy drives cash flow, cash flow drives value

Stock prices are expectations. When expectations fall, you see drawdowns. Property values similarly track expectations, but through cash flow and yield compression rather than instantaneous repricing.

For an office asset, vacancy feeds into cash flows through:

lost base rent,
downtime between leases,
re-leasing costs,
and sometimes changes in achievable rents at rollover.

Then value adjusts through:

cap rate or discount rate movement,
risk perception,
and liquidity expectations (who is buying, and at what terms).

In a simplified way, you can think of office vacancy drawdown risk as coming from a chain:

Vacancy rise → income fall and timing delay → valuation and financing stress → more pressure to lease or refinance under worse assumptions.

The chain matters because it creates feedback loops. When owners believe the market is headed worse, they accept concessions earlier, which might slow the vacancy deterioration but damage income. Or they hold out, pushing vacancy higher for longer. Both behaviors change the trajectory you observe.

That’s why “vacancy is the dependent variable” thinking can get you lost. Vacancy is partly an observed result, partly a behavior.

Where the stock analogy helps: peak-to-trough mindset and nonlinearity

Even if vacancy is slow, it can be nonlinear in impact.

Two examples I’ve seen repeatedly:

Lease rollover timing creates cliffs. If a building has a large share of leases expiring in a short window, the effective vacancy pressure can jump quickly, even if the physical move-outs take months. Investors and lenders react to the expected forward vacancy, not the current one.
Incentives can cascade. Once a landlord starts offering meaningful rent-free periods or tenant improvement packages, the “market clearing” terms can shift. The vacancy rate may not spike dramatically, but rent achieved at new leases can fall. Value drawdown follows.

Stock markets highlight this nonlinearity through drawdowns. In a stock, correlations shift during stress and volatility clusters. In office leasing, correlations show up as customer behavior aligning: tenants reduce space, delay decisions, and renegotiate. You feel it as a broad-based slowdown rather than isolated leasing issues.

So the takeaway is not “vacancy equals stock volatility.” It’s “vacancy risk can cluster and amplify, and you should model the tail, not just the average.”

The biggest trap: confusing long-run averages with crisis drawdowns

In my early spreadsheet days, I treated vacancy like a long-run mean with noise. That gives you something that looks statistically respectable but fails the stress test.

Stocks taught everyone a brutal lesson: the average tells you little about what happens during drawdowns. Drawdowns are about extremes and timing.

Vacancy also has a “timing and extreme” dimension. Suppose an office market is usually fine but experiences periodic oversupply and demand pauses. If you only model average vacancy, you’ll underestimate how long your asset can sit between tenants. And you’ll underestimate the point at which financing becomes a forced seller scenario.

A good vacancy model should therefore track:

Duration of elevated vacancy, not just magnitude.
Forward leasing environment, not just past occupancy.
Sensitivity to supply and absorption changes, which can shift the reversion speed.

If you want the drawdown parallel, you’re modeling “how far and how long” rather than “what’s the typical vacancy.”

How to borrow drawdown tools without pretending you have stock data

Stock drawdown modeling often uses maximum drawdown, rolling drawdowns, stress scenarios, and sometimes regime switching. You can adapt the same philosophy for vacancy outcomes, but you need the right variables.

Instead of “price,” use a cash flow or valuation proxy. For example, you might define an “income drawdown index” like this for illustrative purposes:

baseline: stabilized net operating income at a chosen occupancy level,
current: projected net operating income given expected leasing downtime and rent achievable at next leases,
index: current divided by baseline.

Then your “drawdown” is the peak-to-trough decline in that index.

You can compute a similar thing using valuation estimates from your underwriting model. The key is to keep the horizon consistent. If your vacancy data is quarterly but your underwriting resets annually, you can easily get a mismatch.

Where this works best is when you have enough historical leasing roll data and rent comps to produce credible forward projections, even if they are scenario-based.

If your data is thin, you still can do a regime-style approach, but don’t fake precision. It’s better to be honest about uncertainty and model wider ranges.

Incorporating other property types: the demand story differs by use

Offices don’t trade like warehouses or shops, and they certainly don’t behave like landed houses or condominiums. But the modeling mindset is transferable because the drivers rhyme.

Warehouses can be sensitive to industrial demand, logistics cycles, and e-commerce distribution patterns. Vacancy often responds to employment and industrial throughput, and re-leasing can be faster if the space is more standardized. Still, supply additions matter, especially if large modern blocks hit the market at once.
Factories are typically more idiosyncratic. Tenant needs can be very specific, so vacancy can remain “low” while usage capacity is constrained, or “high” while technically available space exists but isn’t fit for the next buyer. Your vacancy definition and your leasing assumptions matter more than the headline rate.
Shops and shophouses live and die by footfall and tenant mix. A shopping street can look occupied but still be economically distressed if sales-per-square-foot drops. Vacancy can lag that deterioration because small tenants renew for personal reasons, and landlords negotiate around cash flow. Vacancy rate alone can understate commercial stress.
Condominium and strata houses are residential, so the macro drivers matter more: household income, mortgage rates, absorption, and policy. Vacancy is usually less central than price and rental yield, but if you model residential vacancy, you’d better consider owner-occupier behavior and the role of rentals versus lived-in units.
Landed houses behave differently again. Supply dynamics, family preferences, and location specificity can make “vacancy” a quieter statistic, while effective risk shows up through price sensitivity and rental liquidity rather than standard vacancy.

The point isn’t to force one formula across everything. It’s to remember that “drawdown” exists whenever people reprice risk expectations. Different asset classes reprice through different channels, but the stress dynamics can still be similar enough to learn from.

A small checklist for not fooling yourself

If you only remember one thing, remember this: a vacancy model is only as good as its definitions and its forward logic. Before you run with the numbers, I’d ask these questions, in plain language:

does your vacancy definition match what investors care about for cash flow, not just what a dashboard shows?
are you tracking lease-up and downtime timing, or only occupancy at report dates?
how do you treat incentives and re-leasing rent, not just “percent empty”?
what is your assumption about reversion speed when demand improves?
can your scenario framework handle supply additions or removals, not just demand fluctuations?

That list sounds simple, but in practice, these details decide whether your model predicts an office drawdown that actually resembles the market, or one that looks nice on paper and fails when you need it most.

Modeling approaches that tend to work better

I don’t want to pretend there is one best model. There are different tools for different data richness and decision horizons.

In the real world, here are three approaches I’ve seen perform more reliably than the single “average plus noise” method:

1) Scenario-based underwriting with explicit leasing timelines

This is the boring method that wins. You model rent, vacancy, and leasing downtime explicitly in your underwriting cash flow. Then you stress each assumption: lease-up speed, concessions, rent growth or decline, and turnover volume.

It’s closer to how you actually decide. You’re not predicting quarterly vacancy perfectly. You’re deciding what occupancy level you can survive and what market conditions break you.

2) Regime switching, but with humility

Regime switching is useful when you can identify different market phases, like “tight leasing,” “balanced,” and “oversupplied or demand shock.” The model can estimate different volatility and transition probabilities across regimes.

The danger is overfitting. Office markets can be lumpy, and you can end up “finding regimes” that are just artifacts of a small sample. If the historical record is short, you can still use the concept, but you’ll probably need scenario probabilities you set from judgment rather than pure calibration.

3) Tail-risk estimation via historical drawdown analogs

For stock drawdowns, you can look at historical peak-to-trough patterns and translate them into stress levels. For offices, you can do a similar exercise, but you need to define the analog you’re measuring: income drawdown or valuation drawdown.

The value of this approach is psychological and practical. It forces you to ask: if the market behaves like it has in stressful periods, what occupancy and rent assumptions will you need?

How I would stress-test offices specifically

Let’s make this concrete without pretending these numbers are universal. Imagine a mid-size office building. You’re underwriting a forward period, and you’re worried about vacancy.

A reasonable stress test is not “vacancy rate rises by X points.” That’s too shallow. It’s a combined scenario where vacancy rises, but also where leasing becomes slower and concessions deepen.

For example, in an illustrative severe scenario:

lease-up speed slows, increasing downtime between tenants,
rent achievable at rollover declines more than rent declines for renewals,
incentives expand because landlords compete harder,
and you apply a slower reversion to stabilized occupancy.

You then compare outcomes against your financing and operating constraints. Do you breach covenants? Does refinancing become impossible? Does your required yield jump? The drawdown you experience is not only vacancy. It’s everything vacancy triggers.

That’s where the stock drawdown analogy becomes more than metaphor. The “peak” is your underwriting comfort zone. The “trough” is the cash flow level at which the situation becomes dangerous.

The role of supply timing, and why “current vacancy” can mislead

A trap I’ve seen: using current vacancy to infer future risk. Current vacancy is lagging. Forward risk is tied to upcoming lease expiries, new supply deliveries, and the pipeline of renovations.

When new offices come online, the market doesn’t react at delivery day alone. It reacts as leasing signs start to slow and as tenants realize they can wait.

In warehouses, timing also matters, but it often interacts differently with demand. In shops, timing interacts with footfall trends and retailer mix, where occupancy can remain stable while economic health weakens. In shophouses and factories, specificity and tenant fit can dominate over aggregate vacancy.

So the better practice is to build a “forward vacancy pressure” metric, even if it’s rough. Consider:

weighted lease expiries by month or quarter,
anticipated delivery schedules for competing projects,
and realistic absorption speeds, informed by comparable leasing activity.

This approach looks more like a cash flow forecast than a statistical prediction. It’s also more aligned with decision-making.

When the stock market analogy breaks down (and why that’s okay)

Sometimes you’ll try to use stock-style risk quantification and hit a wall. For offices, the wall usually appears when:

vacancy becomes more driven by landlord behavior than market demand,
building-specific factors dominate (floorplate design, floor efficiency, air conditioning reliability),
or the “comparable” set is unstable because the market composition changes.

Also, offices can have a long tail. In stocks, liquidation can happen quickly, forcing drawdowns. In property, owners can choose to wait, refinance later, or accept weaker outcomes temporarily. That option value can reduce forced sales, changing the observed drawdown patterns.

So if your office “drawdown index” does not match stock drawdown patterns, don’t assume your model is wrong. It may be telling you that the constraint is different: not liquidity shock, but leasing friction and financing structure.

A better way to frame the decision: what would you do under stress?

The most useful output of a vacancy-versus-drawdown model is not a single number. It’s a range of outcomes and a set of practical decisions you can test in advance.

If your model says elevated vacancy lasts longer than you can fund comfortably, you change your strategy now, before the market forces the hand. That might mean:

underwriting with more conservative re-leasing timelines,
negotiating lease rollover structures (stagger expiries if possible),
choosing tenant types that re-let faster,
or adjusting your hold versus sell timing.

You can do the same for other property types. For example, shopping assets might require different stress assumptions, focusing on tenant turnover and rent collection rather than physical vacancy. Warehouses might require supply competition and logistics demand sensitivity. Residential strata houses or condominiums might require sensitivity to rental liquidity and buyer affordability, not just vacancy.

Still, the core discipline stays consistent: model the trough, not the average.

Bringing it all together: the shared lesson behind the different markets

Stocks teach you to respect drawdowns. Offices teach you to respect time.

When you combine those lessons, you get a more robust way of thinking about industrial and commercial property vacancy risk:

Treat vacancy as a driver of cash flow and expectations.
Measure risk in “how far and how long,” not just “how high.”
Stress the mechanisms, not the headline rate.
Translate definitions carefully, so your model reflects what investors and lenders actually price.

If you do that, the office vacancy rate stops being a depressing percentage you glance at once a quarter. It becomes a tangible risk narrative, like a stock drawdown chart, except with lease expiry dates in the driver’s seat.

And once you see that, the modeling work feels less like guesswork and more like disciplined preparation. Which is good, because the market rarely rewards confidence when it’s unsupported. It rewards the people who can survive the trough, then buy or refinance when the cycle turns, not when the spreadsheet last looked rosy.