Due to the private nature of real estate, investors are often faced with making decisions based on broad market-level information or data relating to a rarified class of hypothetical, top-quality, perfectly located, “prime” benchmark properties, which, much like Peter Pan, never age. But determining underwriting assumptions for individual properties requires greater specificity. For a more realistic view, one should consider data relating to a grouping of real buildings with a same-store sample1 applied. This can help ensure depreciation is captured properly and that appropriate dispersion data can be calculated to help appropriately scale asset-specific adjustments away from a market average.
DRAFTING A NEW BLUEPRINT
Valuing a property requires analyzing rental value and net operating income growth to establish a view on cash-flow growth over its holding period (usually five to 10 years). Rent data widely available for “prime” benchmark properties are useful in understanding general market dynamics, but they can be misleading for all but the best quality, brand new buildings. And while year-over-year growth based on broad market data is also appropriate for monitoring market movements, their use in underwriting a particular asset over a multi-year holding period can be biased by year-to-year sample changes.
We used MSCI Global Intel PLUS to select more appropriate samples of buildings and measurement periods to refine the market data and help investors make more informed decisions at the asset level. Specifically, we were able to control for the bias of sample changes with a same-store sample approach. This allows an investor to understand what holding period growth for a “typical” asset has looked like at various points in the cycle.
But what if the subject asset is not typical? Our filtering approach also allowed us to calculate percentiles defined over the holding period to account for the fact that, throughout the cycle, there is often a large dispersion between the best and worst performers. This provided quantitative evidence to help scale asset-specific adjustments away from the average. The exhibit below examines rental growth in the City of London office market to illustrate.
A quantitative contribution to asset-level underwriting in the City of London
|MSCI VS. "PRIME"||MSCI||"PRIME"||Difference|
|SAME-STORE 5-YR FORWARD||2.6|
5-year forward data ends in 2012
|MSCI SAME-STORE 5-YR FORWARD||AVERAGE||MIN||MAX|
Source: MSCI Global Intel PLUS.
Through our analysis, we were able to see that, over the long term, average rental growth has underperformed the “prime proxy”2 by 90 basis points on a year-over-year basis. Additionally, although year-over-year growth averaged 2.8% over the entire time series, the average for the same-store rolling five-year average — a more relevant metric to support underwriting assumptions — is lower, at 2.6%.
Looking historically, we see that even this measure varied between a low of -16.6% in the five years from the market peak in 1988 and a high of 17% over the five years from 1984 running up to that peak. When we dig deeper, we find that, throughout the cycle, there has been wide dispersion between the best and worst performers averaging 11.8% over the time series but widening to nearly 15% in 1984 as the market boomed.
Understanding such parameters could help investors scale their adjustments from market forecasts to asset-specific, point-in-time underwriting assumptions. This approach can be applied consistently to various metrics across global markets and bring previously absent quantitative evidence to bear on a complex real estate investment problem.
1 Same-store sample ensures the range of buildings is constrained to those consistently held over the measurement period (in this case, five years).
2 Prime proxy calculated as the growth between 90th percentile rental values in each year.
3 Dispersion measured by the 10th-90th percentile range on the 5-year growth measure.