CBRE Econometric Advisors Develops Office Taking Rent Series

Executive Summary

The CBRE Econometric Advisors (EA) Taking Rent series leverages proprietary CBRE transaction data to model the spread between asking and taking rents across markets over time. By joining this pool of transaction data with suite-level asking rents at the time of transactions (with, on average, a match rate over 70%), we’re able to track this spread with unprecedented precision. Some key takeaways from the data include:

• While asking rents fell just 0.5% from Q1 2020 to Q2 2022 on the Sum of Markets level, taking rents declined 4.2% during that time. This indicates that the rise in vacancy during the pandemic did, in fact, hit market fundamentals despite the impact not being fully apparent in asking rents.
• The divergence between asking and taking rents was most pronounced in markets that saw greater changes to vacancy, indicating that taking rents are valuable in times of market volatility.
• Only 11 of 62 major U.S. office markets saw asking rent declines from Q1 2020 to Q2 2022. This would generally be an indication of market resilience, but our EA Taking Rent series dives deeper, showing that 34 markets saw taking rents decline during that time—far more in line with the macro backdrop impacting real-time investment decisions.

Introduction

In 2021, EA introduced the EA Asking Rent series for Tier I office, industrial and retail markets across the U.S. This series—a repeat rent methodology, which is widely considered the gold standard in the industry—leverages a more advanced statistical model than any rent series on the market. After incorporating building-level interpolation and a chain-linked aggregation, the EA Asking Rent series runs back to 1988 for major office and industrial markets and to 2005 for retail and is projected 10 years into the future. 

While the EA Asking Rent series captures market movements that are absent from nearly any other competing series, the pandemic has shown us that any series based solely on asking rents fails to detect nuances of real-time market movements in actual negotiated transactions. For this reason, the EA team developed the EA Taking Rent series. Building off the EA Asking Rent series, the Taking Rent series takes things a step further by mapping out the spread between asking and taking rents, how it changes over time, and what drives this spread.

The EA Taking Rent series will be available to EA clients for 62 U.S. office markets and will be rolled out to cover other property types as well. This vital step toward an effective rent series provides invaluable insights to EA clients around real-time market movements giving our clients a much-needed edge in understanding idiosyncrasies across investment opportunities.

How the EA Taking Rent Series was Constructed

As the largest commercial real estate services firm in the country (and the world), CBRE maintains a market share of leasing representation of roughly 20-30% across major U.S. markets. This creates the unique opportunity to leverage a large, relatively random, consistent, and standardized dataset across markets and time periods.

With hundreds of thousands of data points going back to 2010, we were able to join each of these transactions with a suite-level asking rent at the time of transaction. With an average matching rate of over 70%, this step was vital in making the leap from asking rents to taking rents. Obviously, a building level asking rent is not sufficient for drawing conclusions about an asking-to-taking discount given the dramatic variance between asking rents on different floors of a building.

At this point, EA undertakes a relatively standard process for outlier detection and filtering out deals that are not representative of the ‘market.’ Some of these filters leverage a standard deviation outlier detection method and others set a ceiling/floor for certain deal-level characteristics. Once this is complete, we calculate an asking-to-taking rent discount for each transaction that passes our data cleaning and scrubbing process.

Modeling the Discount

With deal-level ask-to-take discounts for a vast transaction database, we model the drivers of that spread across markets and time using the following model:



In this process, it is vital to include both deal-level asking rent as well as market-level asking rent. Including both allows us to isolate the impact of changes to the market asking rent, accounting for variation in the sample of what transacted in each period.

As is often the case in commercial real estate, there are certain markets and time periods in which the data is too thin to draw appropriate conclusions. To bolster the data completeness in these cases, the EA team leverages a panel model methodology in which each market is given a market-fixed effect. This allows the model to pick up nuances across markets while smoothing out the noise that comes with limited observations in a particular period. We then single out those markets where data is lacking and use the panel model to predict those discounts.

Below are some examples of our model vs. actual discounts across key markets. The model does a great job of predicting the actual discount in markets with ample observations, such as Manhattan and Washington, D.C. In markets with fewer observations, such as San Jose and Orlando, the model is able to pick up the underlying trend in the data while smoothing out the volatility that comes with a limited data set.

Figure 1: Manhattan

Source: CBRE Econometric Advisors

Figure 2: Washington D.C.

Source: CBRE Econometric Advisors

Figure 3: Orlando

Source: CBRE Econometric Advisors

Figure 4: San Jose

Source: CBRE Econometric Advisors

Weighting each deal equally (since each transaction is one data point representative of ‘market,’ regardless of size), we then predict the discount across time and markets. Predictions are aggregated by year and then interpolated quarterly using a cubic spine interpolation method, a common practice to smooth the data series.

Finally, we apply the modeled time series of the asking-to-taking ratio to the EA Asking Rent series, resulting in the one-of-a-kind EA Taking Rent series. 

The EA Taking Rent Series in Action

Below are some examples of how the EA Taking Rent series captures market movement that is hidden from any analysis of asking rents. In each of the below markets, asking rents remained relatively unchanged over the course of the pandemic despite vacancy increases exceeding 400 basis points (bps) in some markets.

For example, in Washington, D.C., where the vacancy rate increased from 14.5% in Q1 2020 to 17.5% in Q2 2022, asking rent remained unchanged. But after we layer in taking rent, we see that taking rent decreased 2.6%. In Philadelphia, for example, asking rents increased 2% despite vacancy increases of 430 bps*. Taking rents, on the other hand, actually decreased 4.2%, far more in line with expectations and accurately reflecting what local market participants are experiencing in real time. The story is similar in Manhattan, where the decline in taking rents of 12.7% since 2020 is more than double the drop in asking rents during that time.

Conversely, in markets such as Miami and West Palm Beach, vacancy fundamentals have bounced back quickly. As such, it is not surprising that asking rents and taking rents are trending in the same direction.

Figure 5: Miami

Source: CBRE Econometric Advisors

Figure 6: Manhattan

Source: CBRE Econometric Advisors

Figure 7: Washington D.C.

Source: CBRE Econometric Advisors

Figure 8: Philadelphia

Source: CBRE Econometric Advisors

The National Picture