series. The results by Geltner and Kluger (1998), Seiler et al. (1999), Pagliari et al. (2005), Li
et al. (2009), and Yavas and Yildirim (2011) imply that REIT returns lead direct returns even
after making adjustments for the property-type mix. Geltner and Kluger base their analysis
simply on contemporaneous and lagged correlation coefficients. Pagliari et al., in turn,
mention their finding in a footnote without reporting the analysis in more detail, while Seiler
et al., Yavas and Yildirim, and Li et al. use data that exhibit appraisal smoothing as a proxy
for direct real estate returns. Also Chau et al. (2001) include sector level direct real estate
data in their analysis. However, the securitized real estate data are at the aggregate level and
Chau et al. do not study the lead-lag relations or dynamics between the securitized and direct
markets. Moreover, none of the above mentioned studies attempts to derive impulse
responses to study the response speeds of real estate returns to shocks in the fundamentals.
To our knowledge, this is the first study that compares the reaction patterns of securitized and
direct real estate returns to shocks based on impulse responses derived from an econometric
model.
The recent findings of Yavas and Yildirim (2011) imply that there are differences
between property types and between firms within a property sector. We acknowledge the
possibility of different reaction speeds among REITs. However, this study concentrates on
examining the price discovery at the market (sector) level, since our aim is to investigate
whether the REIT market generally reacts faster to shocks than the direct market or whether
the previous findings can be attributed to the property-type complications.
3 Data Description
We include four real estate sectors in the analysis: apartments, offices, industrial property, and
retail property.2 For securitized real estate, the FTSE/NAREIT Equity REIT sector level
indices are used and for direct real estate we use the sector level transaction-based NCREIF