The apartment industry recovered significantly in 2010. Demand for apartments jumped, even while job gains remained subdued, as the advantages of renting became more obvious with the bursting of the housing bubble. With the improvement in the market, investment interest in apartment properties rose as well. So where do prices of apartment properties now stand relative to their prior peak? By one measure, prices are still 30 percent below the previous peak. Other measures indicate they are within 10 percent.
Clearly, this is a difficult question because there is no single standard measure for property prices. Instead, there are a variety of indexes using different methodologies and different data sets, and each provides very different answers. Let’s look at the various price indexes and what they suggest about the market.
Comparing the purchase price of a specific property gives a clear appreciation rate for that property. But aggregating data from different properties, with different characteristics, bought and sold at different times, has inherent problems.
There are two common data problems. If only those properties with transaction prices (that is, only properties actually sold) are included, then there is a “compositional” bias (commonly known as “comparing apples and oranges”)—properties sold in different time periods may be too different to compare usefully.
The raw transaction data from Real Capital Analytics (RCA) is an example of this. The property prices are calculated on whatever transactions take place in a given time frame. An alternative approach is to include all properties, not just those that have sold. The problem here is that most property prices will have to be estimates of market value (e.g., appraisals), which tend to lag true market values. The widely used data from the National Council of Real Estate Investment Fiduciaries (NCREIF) exhibits this problem.
Despite these problems, both methods are worth a look. The RCA line, which uses raw transaction data, shows how variable such data can be. By contrast, the NCREIF data, which use appraisals, are very smooth but are also clearly slower to capture market changes. Comparing the most recent data with the 2007 peak, RCA shows apartment properties selling at only 8 percent less, while NCREIF shows a gap of 23 percent.
Is there a way to produce commercial property price indexes that avoid compositional bias without having to rely on appraisal prices? MIT’s Center for Real Estate has partnered with both RCA and NCREIF and employed two different statistical techniques (one with each data set) to address these problems.
MIT’s Transactions-Based Index (TBI) strips the appraisal prices out of the NCREIF data and then uses “hedonic” estimation to come up with a “representative property” whose characteristics match the average of those in the NCREIF data set. The TBI is based on changes in the price of this representative property.
With RCA data, MIT (in conjunction with RCA, Moody’s and Real Estate Analytics) employs the “repeat sales” technique. This only measures price changes for each individual property. With a robust data set, that is sufficient to calculate an industry-wide index. (This is the same technique used in creating the Case-Shiller Home Price Index.)
The resulting Moody’s/REAL Commercial Property Price Index (CPPI) is designed not just to provide better information for researchers and investors, but also to support trading of commercial property price derivatives.
These two data series look much more similar. Even so, the different underlying data sets, coupled with the different methodologies, give somewhat different results. The Moody’s/REAL CPPI fell by 40 percent from peak (2007 Q1) to trough (2009 Q3) and is still 30 percent below the peak. MIT’s TBI showed a price drop of 33 percent from peak (2007 Q3) to trough (2009 Q4) and is now 21 percent below the peak level.
Applying these techniques to raw data sets solves the bias problems. But they also raise some questions. Are we really interested in the price of a “representative property?” Or should, for example, larger and/or more expensive properties have a greater weight? Is it better to have different indexes for Class A, B and C properties instead of mixing them together in a single index?
There is no single correct answer to these questions. It depends largely on the reason for using the price indexes. For our purposes, there is yet another index to consider—the Green Street Advisors Commercial Property Price Index (GSA CPPI). The GSA CPPI is based on properties bought and sold by public REITs, and it places more weight on large deals. It also includes data from deals in process, not just on closed transactions. This may make the index more timely, but it also means it is not completely objective—the information on deals in process comes from expert observers—limiting its use for some purposes.
By the GSA CPPI measure, apartment prices in February 2011 were only 7 percent below the peak. (The December 2010 figure was 8 percent below, while the 2010 Q4 figure—perhaps the best “apples-to-apples” comparison with the other indexes—was 9 percent below the peak.)
No single price index can convey all the information we really want to know. But the good news is that all these price series agree on one thing: prices for apartment properties today are closer to their previous peak than are prices of other real estate asset types.
Mark Obrinsky is vice president of research and chief economist at the National Multi Housing Council in Washington, D.C.