The response of sub-sector REIT returns to shocks in fundamental state variables

Abstract

Using monthly data from 1994:01 to 2005:03, the results from vector autoregressive models and generalized impulse response analysis indicate that unexpected shocks in industrial production, inflation, term structure, default risk, and the federal funds rate have virtually no statistically significant impact on sub-sector REIT returns.

Notes

¹ Chan et al. (1990), Li and Wang (1995), Chen et al. (1997, 1998), Naranjo and Ling (1997), Peterson and Hsieh (1997), Karolyi and Sanders (1998), Chandrashekaran (1999), Payne (2003) and Ewing and Payne (2005).

² The use of the generalized impulse response approach to derive shocks parallels Ewing (2002), Payne (2003), Ewing et al. (2003), Payne and Mohammadi (2004) and Ewing and Payne (2005).

³ This period corresponds to the hypothesized structural shift in the REIT markets in 1991 due to the increased domination of large institutional investors, market liquidity, media coverage and transparency; and flexibility in purchasing property due to the introduction of umbrella partnership REIT organization. See Payne and Waters (2005) and citations therein.

⁴ Unit root tests included both a constant and linear trend and are available upon request.

⁵ Cointegration tests assumed a linear deterministic trend and are available upon request. The lag lengths for the respective Johansen–Juselius cointegration tests were based on Akaike's information criterion. Moreover, given the relatively short time horizon, the power of the cointegration tests are questionable (see Hakkio and Rush, 1991 and Kremers et al., 1992).

⁶ Full results of the respective VAR models are available upon request. Dummy variable for the ‘January effect’ was statistically insignificant and subsequently omitted from the VAR results.

⁷ The discussion of the generalized impulse response analysis draws heavily from Payne (2003).

⁸ Orthogonalized and generalized impulse responses will be identical only in the case when the covariance matrix is diagonal (see Koop et al., 1996 and Pesaran and Shin, 1998).

⁹ LR statistics for the respective sub-sector REIT VARs are as follows: apartments 43.12, industrial 48.35, lodging 39.69, manufactured homes 45.81, office 45.95 and regional malls 43.23.

¹⁰ Generalized impulse responses beyond 1 month are statistically insignificant across all sub-sectors.