Asset Management: A Systematic Approach to Factor Investing (Financial Management Association Survey and Synthesis Series) by Andrew Ang

Author:Andrew Ang
Language: eng
Format: mobi
Publisher: Oxford University Press
Published: 2014-07-06T16:00:00+00:00

Fiduciaries (NCREIF), which collects property data from its members. NCREIF

computes a value-weighted index based on appraisals on a quarterly basis. The

indirect property returns are the FTSE NAREIT All Equity returns, also at the

quarterly frequency. I compare these property returns with S&P 500 equity returns.

Table 11.11 shows that NCREIF returns are very smooth (the quarterly

autocorrelation is 78% and volatility is just 2.2%) compared to REIT and equity

returns, which have autocorrelations close to zero and volatilities around 8% to

9%. Notably, the correlation between REIT and equity returns is high, at 63%,

while the correlation between NCREIF and REIT returns is low, at 15%. Thus

REITs resemble equities more than they resemble direct real estate. Hartzell and

Mengden, who were industry practitioners at Salomon Brothers, were the first to

document this phenomenon in 1986, and they generated substantial discussion about

the underlying nature of real estate returns. Was real estate indeed a separate asset

class, or was it just a different version of equity (and debt)?

This debate continues in real estate economics today. My takeaways for asset

owners from this long literature are:

Direct Real Estate Returns Are Not Returns

The NCREIF series (and the corresponding widely used Investment Property

Databank series in Europe) have many data biases because the values are not based

on market transactions. REIT returns, in contrast, are investable returns. There are

two important and large biases:33

1. Smoothing bias

Using appraisals (done at most once or twice a year) artificially induces

smoothness,34 which is why the autocorrelation of the NCREIF series is so much

439

higher than REITs or equities in Table 11.11.

2. Selection bias

The properties that you see sold are not representative of the entire stock. For

example, perhaps only the best properties are sold because those are the ones a

developer is sprucing up, while the foreclosed crummy ones are not being sold. 35

In addition, REITs are levered while NCREIF returns are reported on an

unlevered basis. Researchers have developed many methods to move from

appraisals to transaction-level (and repeat-sales) indices, to desmooth real estate

returns, to take into account selection bias, and to remove other illiquidity biases.

The effects of these biases are enormous. Lin and Vandell (2012), for example,

report that for a one-year holding period, the variance of direct real estate returns

should be three times higher than the raw variance reported in Table 11.11.

Pagliari, Scherer, and Monopoli (2005) argue that there is no difference in direct

and indirect real estate returns after adjusting for leverage and sector composition

and also adjusting direct real estate returns for appraisal smoothing.

Smoothing and selection biases are shared by all illiquid assets—including

private equity—and I discuss methods of dealing with them in chapter 13. I use

NCREIF returns for now, but you’ve been warned: these are not actual returns.

In the Long Run, Direct and Indirect Real Estate Returns Move Together

Direct and indirect real estate returns are linked in the long run. Figure 11.12

graphs the correlation of long-horizon log NCREIF returns with long-horizon log

REIT returns. At the one-quarter horizon, the correlation is around 0.22, which

reflects the low correlation between NCREIF and REIT returns in Table 11.11.

(Quarterly returns in Table 11.11 are arithmetic returns but I use log returns in

Figure 11.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.