Q5.2) Critically compare and contrast the alternative methods of adjusting for smoothing.
1. Static Correction Methods
1.1. Full-Information Models --Blundell & Ward (1987) --Firstenberg, Ross & Zisler (1988) --Fisher, Geltner & Webb (1994)
1.1.1. Blundell & Ward (1987)
1.1.1.1. Assume that property values follow a first-order autoregressive process
1.1.1.1.1. A relatively reasonable assumption as generally only the first-order autocorrelation coefficient is statistically significant
1.1.1.2. As the model views returns as following a first order autoregressive process we can estimate a as b in the following model
1.1.1.2.1. FORMULA
1.1.1.3. As you use a lag you¡¯ll loose an observation
1.1.1.3.1. FORMULA
1.1.1.4. As you’ll see the average return hasn’t changed that much at all, just the standard deviation
1.1.2. Firstenberg, Ross & Zisler (1988)
1.1.2.1. Similar to Blundell & Ward (1987) except extend the AR1 model to a AR4 model
1.1.2.2. The authors do this to take account of the seasonal nature of the NCREIF Index, in that most constituent properties are only valued annually
1.1.2.2.1. FORMULA
1.1.2.3. The estimates for a can be estimated from the following AR4 model
1.1.2.3.1. FORMULA
1.1.3. Fisher, Geltner & Webb (1994)
1.1.3.1. A similar model to those discussed previously
1.1.3.2. As with FRZ they include a seasonal element
1.1.3.2.1. –Except only use the 1st and 4th order lags
1.1.3.3. In addition the authors make assumptions concerning the 'true' volatility of the asset
1.1.3.3.1. A number of survey papers have argued that the true volatility of property is approximately equal to 50% that of equities
1.1.3.3.2. –Can be problematic in some markets and time periods
1.1.3.4. FGW impose this condition in their model
1.1.3.5. They estimate the de-smoothed return using the following equation
1.1.3.5.1. FORMULA
1.2. Partial-Information Models --Geltner (1993)
1.2.1. Geltner (1993) proposed an alternative means of estimating the unsmoothed return without having to implicitly assume market efficiency
1.2.1.1. FORMULA
1.2.2. The problem is that the parameter (a) can¡¯t be estimated statistically
1.2.3. Researchers applying this technique have to rely on an a priori assumption regarding the true volatility of real estate
1.2.3.1. Most papers have assumed one of the following –The volatility of real estate is half that of equities –The volatility of real estate is two-thirds that of equities –Real estate’s risk-return trade-off is equal to that of stocks