On Work From Home
Painful but not fatal
Not to overly focus on the office sector, but I thought as a follow-up to my prior office posts it would be good to do a deep dive into work from home (or WFH). This is the billion dollar question in the entire office sector, so I figured I’d share some thoughts & analysis.
The first thing I want to say is - by no means do I think I have a definitive answer here. My view on this, as a I view all things, is to try and assess a range of potential outcomes and their relative probability of occurring. So the following represents essentially what I view as the base case - but I also believe a wider range of outcomes is quite possible. Do not be lulled into a false sense of precision by the figures here!
A Quick Summary
Clearly WFH has had a huge negative impact on the office market, but the key question for office owners is how negative is it? I have seen very little quality analysis on the subject & so decided to share a bit of my own thinking.
In analyzing the impact of WFH on the office market, I believe vacancy (including space available for sublease) is the key metric. It is important to remember that utilization measures are indirect, and what ultimately matters is occupancy of space by office users. By combining vacancy data with job growth and supply data we can begin to estimate WFH damage that has already occurred, and what may be yet to come. I share this analysis on 5 markets (Austin, Atlanta, Seattle, NYC & Columbus), and the resulting WFH damage across the markets range from ~8% to 32% reduction in demand. There is some uncertainty here that may significantly overstate the worst of these numbers, which I detail later on.
8 to 32% is a huge range, however this large variance is supported by our separate study of company WFH policy across the top 200 Fortune 500 companies, as well as the most prominent fully or heavily remote companies. This study revealed significant disparity in WFH adoption rates by industry, namely that significant WFH/remote work adoption appears to be heavily concentrated in low margin or, in the case of tech, low cash flow businesses. In other words - firm’s that go heavy remote seem to do so primarily based on its impact on the bottom line. On its face this makes a lot of sense - the firms shedding office space are those which generate the greatest relative increase in profits or cash flow by doing so.1 The industries going heavy WFH are insurance, & retail (both low margin), as well as government, and smaller tech firms (non profit and low free cash flow respectively). Now it is worth point out that this margin relationship could be coincidental, and it might be that something else like job function is really driving these differentials.2 Regardless, as the distribution of industries across markets is highly variable, this means the relative impact on various office markets is also going to have a large range, which gels with our market by market analysis.
All that said - I would be remiss to not discuss utilization data at all, which I am going to do in a second article after this one. There I will review the various office utilization (eg attendance) data providers which have sprung up over the last few years, providing some general commentary on each and putting them together as another check on potential demand impact.
Analyzing WFH’s Impact
It is important to remember though that from the perspective of an office investor, WFH is an indirect measure. What matters to an office investor is demand (and of course supply) for office space. WFH obviously has some impact on office demand - however I believe it is very much unclear still what this exact impact is. In other words - I don’t think we can confidently correlate a given level of WFH with a given reduction in office demand, at least yet.3 However I do think we can put some rough brackets around the impact, as I attempt to do here.
This uncertainty is why I believe the most important metric is vacancy4. Vacancy captures all of the space office users have given back, and what has been put on the market as available for sublease as well. Remember, we are now 3.5 years into this brave new WFH world - a significant portion of the impact has already been felt. This is critical - the passage of time can give us a relatively high confidence level in terms of bracketing overall impact of WFH. Many office leases are 5 years (more like 10 in NYC), so for most markets a majority of the damage from WFH has already been done. Further it is important to realize that firm’s which have decided on a significant reduction in office space use are able to list their space available for sublease.
Now clearly not every firm that is shrinking is going to sublease their space - smaller firms for example may not be able to easily do so (eg if you only have 1 floor or a portion of 1 floor it is hard to lease out 25% of that without some significant investment). And if the firm’s target reduction is slight, then it also logistically may be difficult to do (eg a 5% reduction in space).
But for larger users with meaningful space reductions, they have very much been able to sublease space (or at least try to!).
Take an example made up corporation Megacorp that leases 100k sf in New York, across 5 floors of 20k each. If Megacorp wanted to reduce their square footage by 20% or greater due to WFH, they would quite easily be able to list an entire floor for sublease and consolidate their workers on the remaining floors. Not every firm is going to do this (we aren’t all perfectly rational economic actors), but a large portion will - any sublease is essentially free money.
Therefore - it is reasonable to assume that we have already felt much of the WFH pain, as firms that have gone remote are mostly already reflected in the current vacancy data. We can use this principle combined with market level vacancy, office construction and job growth data to make some inferences about the extent of WFH impact, and how much more damage is yet to come.
Market Level WFH Analysis
To measure the current WFH impact, we can look at total office using job growth for a given market, total office stock growth (eg expanded supply), and the change in vacancy rates.5 We can use total office jobs growth as a proxy for total demand in office space use. Subtract out supply growth and we get what in theory should be the overall change in office vacancy rates.6 We can compare this to actual change to impute WFH related damage.
I do so in the table below for 5 markets. As you can see – the impact felt thus far from WFH is highly variable by market, and tracks fairly well in terms of what we are seeing from corporations announced WFH policies (as I shall detail later). Austin and Seattle, as tech heavy markets, seem to be the most impacted. It is also worth noting that these numbers may be overstated – as I assume here that 100% of vacancy increase is due to WFH & all job growth is local7. Given office using jobs are up, this isn’t a crazy assumption, however if firms had taken on excess space in anticipation of future growth (something tech companies are known to do & have done in the past), then some of the vacancy could be related to firms giving back excess space due to the tech downturn.
Sources: Census data for office jobs, Costar for vacancy & supply
Most relevant for my NYC office thesis – NYC has the least WFH impact by a good margin. However given NYC’s longer lease terms (WALT is weighted average lease term) one could assume that a fair amount of damage is still to come.
Critically for our analysis it looks like a significant portion of WFH damage is frontloaded. As I noted above it is reasonable to assume firms that have gone all or mostly remote have already put their office space up available for sublease, & to confirm this we surveyed the top 200 Fortune 500 companies, as well as many of the remote first firms. Based on this analysis, it appears that potentially up to 70%+ of WFH demand destruction comes from firms who have gone primarily or mostly remote. In other words – firms that are heavy remote have significantly greater rates of space reduction than those which are hybrid, and collectively these remote heavy firms appear to constitute most of the increased vacancy. At this point these firms should by and large already be reflected in the vacancy statistics.
To be conservative, I have run an analysis below based on 50% of destruction being frontloaded for most markets, and only 40% for NYC (on the chance longer leases have led some firms to delay decisions).
This gets a little complicated so please bear with me. I will break down the below piece by piece.
The first 3 lines are hopefully fairly clear. Total destruction already experienced, plus our assumptions on the share of loss which is front loaded, and the share which would be experienced linearly as leases roll over.
Now – given 3.5 years have passed, we have already experienced a good portion of the linear demand destruction. In most markets this is likely a majority (3.5 out of 6 is my assumption here), in NYC it would be more like 35% (based on a 10 year average lease term). Multiplying this percentage by the linear share gets us the linear portion of total destruction already experienced.
We can add this to the front loaded component to calculate the share of WFH demand destruction a given market has already experienced. As you can see – in most markets this is probably about 80% (the 50% front loaded + 29%, which is the portion of linear already experienced), and in NYC it’s a good bit lower at 61%.
Based on this share of destruction experienced + the quantified losses we have seen, we can then calculate the implied total destruction from WFH – the “Imputed total destruction” line item. As you can see, in NYC this is around 8.2%, but ranges much much higher in the tech markets, which may be overstated somewhat for Austin as I detail in the footnotes.
These numbers on the face of it makes sense based on known corporate WFH announcements as it relates to NYC. As far as I’m aware almost none of NYC’s major employers are mostly remote. None of the big banks, none of the big PE funds, few if any law firms (Quinn is the only Biglaw firm I know of that is heavy remote), and none of the big tech firms are remote (although they have had some layoffs & Meta has given back a good bit of space). Bio/Pharma, another major NYC industry, is also mostly in person. Obviously not that there has been no impact – a good number of smaller and midsize NYC tech firms are heavy remote - Yelp as an example is one of NYC’s larger tenants to go remote.
So – based on these numbers we can see that there is only a relatively small amount of WFH destruction yet to come for NYC. Given the extended rollover period, this works out to about a ~.5% drag on office demand per year for the next 6.5 years. NYC’s pre-covid office using job growth averaged 1.64% - reducing this by 8.18% to account for WFH, we can see that NYC should still average a solid ~1% / year in office using job growth barring a recession, even after factoring in WFH.8
Given the last legs of pre-covid new supply are about to deliver (which would be a further drag on the market), based on this analysis NYC will probably begin to see vacancy bottom next year, and within 3-4 years the market ought to be in a much healthier position (barring a recession, which may cause some short term pain & would delay these numbers a year or two). The REITs may well bottom before the overall market - FWIW the NYC office REITs had a slight pickup in occupancy from Q1 to Q2, with only Empire posting a continued decline (Empire has a lower quality portfolio on average).
A similar story plays out in the other markets analyzed. The real problem markets in my view would be slow growth markets. While the WFH impact in these markets appears to be lower on average, slow growth means it would be very difficult to grow out of the WFH hole. Austin and other tech markets also look to still have some pain in store over the next year or two (especially as it looks like tech office job growth will be limited this year & next), but if they can return to their rapid pre-covid growth within 3-4 years (admittedly a big if!) they should recover. A big caveat - the above analysis is just focused on WFH, any kind of recession does not feature into the numbers. This is because recession is a ‘vanilla’ office risk & isn’t something which is novel or unique, & so is something most investors ought to be familiar with. Also given the corporate job losses we have already seen I suspect by excluding any recessionary impact I am actually being conservative (as it may mostly have already happened, but I’m counting all vacancy increase as WFH driven here).
Also a very important note on all of the above - this kind of analysis is very high level and ultimately imprecise due to the messy complexity of reality. Do not let the precision of the charts above fool you - this analysis is ultimately all very broad and should be viewed as having a large margin of error! It is meant to provide a starting point for thinking about the problem, & should not be viewed as the gospel truth.
Company by company WFH policy analysis
As part of Warden’s deep dive into WFH and the office sector (and as I noted earlier), we analyzed the Fortune 200’s WFH policies and found some interesting data. Our finding is that high levels of WFH appeared to be highly correlated with low margin or low free cash flow businesses. In particular, insurance companies and retailers have gone very heavy into remote work, and many of these firms have put up large swathes of their office space up for sublease.
On the face of it, this makes sense. In a low margin business, given the extremely high operating leverage involved, relatively small reductions in expenses can have a huge impact on profitability and net cash flow.
Interestingly, the F200 tech sector does not feature all that highly in having significant amounts of WFH, despite other anecdotal evidence to the contrary (Salesforce is a notable exception to this although a recently leaked memo seems to imply they may be coming back to the office soon). I believe this is explained by the fact that the F200 tech firms are all pretty healthy generators of cash. It appears that smaller tech companies, many of whom which have little to no cash flow, are significantly more likely to be remote or office optional. To confirm this we also ran a separate analysis of companies that have moved to significant remote work, and sure enough found that many (most) companies that are significantly remote are smaller tech firms, presumably with poor cash flow generation.
Now we haven’t run a full analysis comparing tech firms cash flow and profitability against their WFH policies, but it would be an interesting exercise to go through. At the face though it would make sense that a company which is burning cash would look to minimize all costs possible, especially in an environment where VC funding has become scarce.
So – while many firms talk about WFH as an employee benefit or perk, it looks like their actual behavior is much more driven by considerations of the bottom line.
In a low margin business, even if WFH is less productive (and for certain jobs it may not be), it may still make sense to cut office space usage from a financial perspective (especially if you have limited control over many of your inputs as is often the case). And certainly in the short run it is going to look great for the bottom line, especially if productivity loss is not apparent for awhile.
If true this means that markets with the greatest exposure to low margin businesses may suffer the most from WFH. Critically for NYC (an area of focus for me) – most low margin/low profitability office businesses & jobs have already left the city. NYC is an expensive place to do business! It is very difficult to operate a low margin business in the city, and if you are worried about the cost of office space you are certainly not leasing space on Park Ave or in Hudson Yards. This gels with our WFH impact data above, and could explain why NYC appears to have some of the lowest (if not the lowest) exposure to WFH demand destruction. This would be especially true of NYC’s highest end submarkets, which have some of the least cost sensitive users in the country.
So - based on our survey of actual users, and also analysis of overall market data, it appears that the WFH impact to NYC is manageable (albeit painful!). In other markets the impact is significantly higher, and could prove to be particularly problematic for markets with low growth and/or a large concentration of low margin / low cash flow businesses.
Conclusions
Pulling it all together - based on my analysis the impact of WFH is highly variable across industries and therefore also geographies. This is supported by our survey of F200 corporate users + firms that have gone fully remote, and is further reinforced by the variability in our metro level analysis.
Based on our analysis of high level metropolitan office data, 60-80% of the damage from WFH has already been done. Most markets will have fully worked through WFH within a few years, and even the longer average lease term markets like NYC should bottom in the next 12-18 months or so.9
I believe this analysis based on total market statistics is the best way to tackle the WFH question, as all the measures of office utilization are indirect. What really matters is how much firms reduce office space use, if at all.
That said it would be foolish to completely ignore the utilization data out there (several good updates came out recently), which I will do so in the next article. On utilization we must remember that data sources need to be viewed in appropriate context of office sizing for maximum employee utilization, rather than average utilization. Think about a bridge - engineers design weight limits around the maximum car load. If they used the average the bridge would collapse! Yet average is what all the office data providers focus on.
Overall - all this analysis of the data would point towards the impact of WFH being significant. End results will vary tremendously by city, with some slow growth metros potentially being impacted very negatively. Even at the low end destruction appears to be at least 8%, and over 30% at the high end. However despite the magnitude of the impact, overall it appears to be manageable (albeit very painful!) for many top office markets, including NYC. Office job growth in the top metros has averaged 1.5 - 4% pre-covid, and once major corporations return to growth vacancy figures should begin to decline. By my estimates many markets should return to their pre-covid vacancy levels by 2028 - however for some slow growth or high damage markets it may take until 2033 or later to fully normalize.
5 years is a long time, especially for public markets investors who are so focused on quarter to quarter results. But this is actually a typical hold period for a real estate investment, and given this 5 year time frame for recovery I believe many office real estate investments are very attractively priced. However this is not true of all markets - with a potential 10 year+ recovery period the weaker markets are borderline un-investable in my view, and I think rents & prices may have a good bit further to fall in some of these markets.
Of course there are a broad range of outcomes, and some of my assumptions here could be incorrect. On the optimistic side, WFH could continue to trend down and in 2 years may be fairly limited, which would potentially significantly reduce the total WFH demand destruction to office space. I could easily see this being reduced by a factor of 50%, so the total demand destruction could be more like 4-16%. In this scenario the market would potentially stabilize much sooner and you could well see an office bull market in 5 years given limited supply and significantly higher construction costs. All else equal given the massive bout of inflation we have seen office prices really need to be a good 20-30% higher than pre-covid values to rationalize new construction in most markets.
On the pessimistic side, return to office rates could level out and WFH could become embedded at current levels. Rates could conceivably even increase over time as smaller firms which seem to have higher WFH rates grow & potentially do not adopt further in office policies. Further I may have overestimated how much of WFH demand destruction is front loaded, and therefore how much latent demand loss is still outstanding.10 Depending on how far off I am, this may mean another few years of recovery.
On the more extreme sides, given what has happened thus far I think it is very unlikely the impact of WFH goes to zero, or alternatively ramps up significantly from where we are today. A large number of firms are committed to remote work, and the financial/retention benefits are so large for some firms it is hard to imagine they all go back to fully in the office. Many lower complexity jobs also just don’t seem to need as much in office interaction (call centers are a great example, I would be shocked if that portion of demand doesn’t stay nearly fully remote). On the flip side literally every corporate announcement I have seen is for more time in office, not less (even Zoom is going back to the office!), so it is difficult to see the WFH tide reversing too much and damage ramping up more from what we have already experienced.
Fin
The cloud of potential outcomes is still fairly broad, and I want to reiterate that this analysis should be viewed as having a good sized margin for error. But I do think at this point WFH is more quantifiable than many people assume which creates some potentially interesting opportunities. No matter what happens it will be interesting to see how the next few years play out.
An example may make this clearer. A retailer operating on a 7% overall profit margin is going to generate a relatively large increase in profits for a given dollar of expense reduction when compared to a much more profitable firm with say a 40% profit margin. A reduction in operating expenses of 1% (against revenues), would increase the retailer’s profits by 14% (1/7), but the more profitable firm’s profits by only 2.5%. Another key variable here is employee ‘value’, or cash flow per employee. Low cash flow per employee means a relatively higher share of employment burden is office space, and so cuts thereto generate much larger returns. So this dynamic could leads to some scenarios where lower margin business with high value employees wouldn’t be as incentivized to reduce space usage (like say an investment bank).
This potential job function explanation is not as clean. EG if we use how rote a job is it works well in the case of say insurers, but less so for say corporate retail jobs. Another potential explanation is how well a given firm or industry is doing. It is important to acknowledge the uncertainty here though & so while I’m confident the industry impact is highly variable, I am less so as to why exactly that is.
Obviously at the extremes you could make some fairly confident predictions, 100% in office or 100% remote. Although I would bet many 100% remote firms actually still have stub offices for meetings etc.
Which I define as actual vacant space + space available for sublease. Some measures only call vacant space ‘vacancy’, and add in sublease space in an ‘availability’ figure, but it is simpler here to just use ‘vacancy’ to cover both.
I am using 4/5 star vacancy data here from Costar, rather than total market data. I believe this allows a cleaner comparison, as it appears that total market vacancy is skewed in certain smaller markets by large owner users. This is most clear in the case of Columbus – JPM and L Brands both have massive owned campuses in Columbus. Since neither has shrunk these campuses (seeing as they own them and both appear to be in office), they make the market vacancy look lower than it is from the perspective of a third party landlord. This is a problem with many smaller markets where a handful of employers can represent a huge share of office jobs. This mixing and matching of data means the analysis is subject to some potential skew should job growth & office construction not run pro-rata with these large employers vs the rest of the market.
It is worth noting that the overall vacancy stats are lower than 4/5 star for every market by a good bit, and if I used those data it would imply less WFH destruction & a generally rosier picture. So I am trying to be conservative by using the 4/5 star data only.
It is worth noting two potential sources of bias in this analysis are shrinking office space usage per employee (aka increasing density) and firms warehousing space to accommodate future growth. Employee density has been increasing fairly steadily over the past few decades, but seems to have been leveling out before covid. You can see this trend in the rise of the bullpen open office concept that many large corporates adopted over the past 20 years. Warehousing was fairly common recently but really only in the tech markets. There is unfortunately not great data on either item & the first in particular has had a fairly large effect in the last decade. However the impact here on our analysis should be offset by the fact that if density was continuing to increase significantly this would actually account for much of the increase in vacancy, as opposed to WFH. So the two factors should cancel out to some degree here. The second probably does overstate WFH damage in certain tech markets a bit, as some increase in vacancy is just firms giving back space they were warehousing and no longer need given a lack of hiring, same as happened in 2001.
Austin’s total job growth here could be overstated if Austin was a net WFH gainer. That would mean a portion of Austin’s job growth was not associated with office occupancy, and therefore is overstated, which would in turn mean my WFH damage calculations for Austin would be too high as well. This could also impact other markets as well, eg if a market was a net WFH loser the WFH damages would be understated. I suspect this overstatement is likely the case for the Texas markets, as DFW also has a rather large gain in office using jobs far out of proportion to office construction. So Austin probably is not in quite as bad straits as this analysis would imply, but it is difficult to disentangle in person job growth vs remote additions.
It is also worth noting that if my 40% frontloaded figure turns out to be conservative, and 70% is more the right number (which it appears to be based on my analysis), the drag on growth gets pretty small with growth at ~1.35% vs 1.64% precovid.
Setting aside the short term potential impact of a recession or further job cuts.
I’m relatively confident this can’t be too far off, especially given we are already 3.5 years through the typical 5 year lease term for most markets. But for longer lease term markets like NYC this could be more inaccurate.