Office REITs Sentiment Still Lower Than GFC
The stock market has taken a very dim view of future of office space and office REITs for a while now, and a sentiment analysis of the filings of office REITs indicates they feel the similarly. CoStar’s value-weighted repeat sales index indicates that office pricing has taken a ~15% haircut, year over year through 22Q4, but the public markets think pricing is worse than that. The DJUSIO, a REIT Index, declined by ~59% over the same period, while an equal-weighted sample of office REIT share prices, chosen based on data availability, declined by ~21% over the same period. The public markets are punishing office REITs, and the sentiment in office REIT filings indicates justification for that.
Office REIT Sentiment Lower Than During The Great Recession
A sentiment score uses the words in a text document to assign an average value to the total document, in this case assigning positive values to a word like “gain” (+2) and negative value to a word like “loss” (-3). The word scoring used here is referred to as “AFInn” and is cited below. Of course many words get removed or have a zero score, like “and” or “but”. Some of the most common scorable words from a recent filing might be “debt”, “loss”, and “assets”. These can all be averaged out to give a score for each quarterly filing for each office REIT with data back to 2003.
The average of each of those REITs with data back to 2003 reveals that aggregate office REIT sentiment is lower than it was in the aftermath of the great recession, even if it is a little bit higher than its absolute lows in mid-2020.
Theoretically we could use these sentiment scores to front-run REIT share prices. If the REIT is giving negative sentiment, or if we’re able to score their filings without regard to our own biases that should give us the opportunity to buy or short a particular stock. In fact, office REIT sentiment and office REIT stock prices correlate very well. Unfortunately, they only correlate well if you significantly lag office REIT sentiment, with a six quarter lag as the best match.
That relationship is most obvious in the GFC, when stock prices started declining in ’07, but sentiment waited until late ’08 to start declining. Inversely, stock prices bottomed out in ’09, but sentiment hit its bottom at the end of ’11. In the pandemic they moved in tandem, at least, but given what stocks are doing now, history suggests sentiment will start declining meaningfully soon.
On an individual basis the relationship between office REITS filings and their stock price varies greatly. Boston Properties (BXP) has perhaps the weakest relationship between filings sentiment and stock price. Recently price and sentiment for BXP have moved together, but looking back in time there are some meaningful disconnects. Kilroy Realty Corp (KRC), on the other hand actually has a negative relationship between its sentiment scores, as sentiment in their filings has been in a slow decline since ’13, but up until ‘20 their stock price generally moved (modestly) higher even as the sentiment in their filings eroded.
The most interesting disconnect comes from Corporate Office Properties Trust (OFC). When the pandemic hit and their tenants packed it in and sent their people home, OFC immediately responded by weakening the sentiment in their filings. The 22Q4 filing decreased by a score of about .2 compared to the 19Q4 filing, and OFC’s sentiment has hardly moved since 21Q4 after falling throughout ’20. This is the biggest negative change in sentiment of any of the office REITs studied. Meanwhile, the stock price has hardly budged at all, indicating the market must not believe the sentiment in OFC’s filings.
Presumably, the market has a lot of faith in OFC’s tenants, who are mainly the U.S. Government. Presumably there are some fortunes that have been made by betting on the government spending frivolously. Additionally, the transcript from their latest quarterly earnings call read as fairly positive.
Most REITs Are Writing More Negative Filings, All REITs Have Lost Value
领英推荐
It gives me no pleasure to point out the other outlier on this chart, Franklin Street Properties (FSP). FSP’s troubles predate the pandemic. The stock has been in decline since 2005 and their sentiment scores have been in freefall since 2016. This saddens me because if I were to seek employment at FSP I would have a very short commute, the kind of commute where I wouldn’t even mind coming into the office 5 days a week. If everyone had that kind of commute, I’d bet office REITs would be in fine shape. Alas.
Finally, it’s worth throwing some context into the mix. Industrial REIT sentiment scores, while not at al-time highs, are nowhere near all-time lows. An industrial supply wave, Fed action, and general economic outlook are surely weighing on industrial REIT sentiment; a local peak in 21Q1 crashed pretty hard as it became apparent that the Fed was going to have to take action on inflation. But industrial sentiment is still in a reasonable place, as you might expect if you follow that part of the market. It’s worth noting that comparing these sentiments on a one-for-one basis, office vs industrial, is probably ill-advised. There may be aspects of writing about office buildings versus industrial buildings that naturally push the sentiment score of office higher. Notably, from ’15 to ’20, which included some good times for industrial, industrial has the lower sentiment score. If these filings were written in such a way that the only differentiation was performance, we would probably expect industrial and office sentiments to be closer together, at a minimum. Nevertheless, it is still noteworthy that industrial sentiment has had a sustained run above office since ‘20 for the first time that I’m able to measure. Given where things stand today, it’s hard not to view this as a harbinger.
Industrial Sentiment Still In Relatively Good Shape
One last note on the sentiment analysis used here; the “AFinn” word scoring I used and cited below is not specifically optimized for public filings, let alone REIT filings. A more specialized word score list might return more significant results, but this generalized sentiment score returns believable results.
Sources:
Finn ?rup Nielsen, "A new ANEW: evaluation of a word list for sentiment analysis in microblogs", Proceedings of the ESWC2011 Workshop on 'Making Sense of Microposts': Big things come in small packages. Volume 718 in CEUR Workshop Proceedings: 93-98. 2011 May. Matthew Rowe, Milan Stankovic, Aba-Sah Dadzie, Mariann Hardey (editors)
CoStar’s Commercial Repeat Sales Index – Value Weighted Office As of April ’23 - https://www.costargroup.com/costar-news/ccrsi