Smoothed...

What does “smoothed” mean in the context of one of those real estate indices?

returns are lower than what they might have been. as a result they all appear closer to the average value - so std. deviation is lower… (less volatile).

cpk123 Wrote: ------------------------------------------------------- > returns are lower than what they might have been. > as a result they all appear closer to the average > value - so std. deviation is lower… (less > volatile). So unsmoothed reflects the actual market? What is the smoothing process? wiping out high and low returns?

What is the difference between “without correction” and “unsmoothed” ?

real estate tends to be more based on appraisals, which are few and far in between. actual market per se is not observable, since the real estate investments are traded much less frequently. So it like an extrapolation of a few and far between appraisals to indicate what is happening in the entire market. Since these are not a true representation - the returns themselves tend to be lower - so volatility is lower. I think - may be wrong – that “unsmoothed” and “without correction” mean the same thing.

The reason I ask, is that Page 123, book 5 fr. CFAI, the question appears to differentiate between “unsmoothed” and “uncorrected” and that the UNSMOOTHED doesn’t suffer from the bias, implying "smoothed’ does…appearing to make smoothed and uncorrected mean the same thing, (according to the proper answer to the Q on page 125, no.26) Its very confusing - what does it look like to you? BTW CP thanks in advance, you have been a big help for years… I find myself getting nostalgic about being on here so long, the “friends” I’ve made, and the journey we have all suffered together…

NCREIF index are based on appraisals - so they would have lower returns and underestimate return volatility. Stale valuations cause the volatility dampening. (Pg. 18 just above the table). This is the uncorrected version - because that is all is available in the marketplace. Table on Pg 18: NCREIF Index (Market place index) --> 6.14% Return, 3.37% STDEV NCREIF (Unsmoothed) -> 7.27% return, 8.95% STDEV… Once you correct it… for smoothing the volatility more than doubles.

I think Smoothed means less volatile because of a desync with real prices in the market . Appraisal data may be calculated using formulae or heuristics which may not correspond to actual transactions ( not sure which ones if there is not much trading going on , maybe comparable transactions in nearby towns or cities). So dispersion numbers may be low or return/risk may be overstated

On, so unsmoothed is actual market, I get that, and I get that smoothing lowers volatility, makes sense, But what is the difference in Question 26-31 reading p 123, where one is called unsmoothed and one is called uncorrected - your statement in your last post equates unsmoothed and uncorrected (saying they are synonyms), but the exact statement below on page 123 implies that “without correction” means smoothed and that “unsmoothed” will be the one with the wilder volatility: “He recommends the NCREIF (unsmoothed) because the NCREIF (without correction) historically suffers from a bias” Then paraphrasing, the answer to the question (27) “What is that bias” is “understated volatility” Whats the deal?

Mentally I just simplify it for myself . Smoothed means mistaken , not-in-sync with real prices , needs correction or volatility injection. Unsmoothed means corrected , with more representative prices , hence returns. Corrects appraisal data returns and dispersion numbers.

janakisri Wrote: ------------------------------------------------------- > Mentally I just simplify it for myself . > Smoothed means mistaken , not-in-sync with real > prices , needs correction or volatility > injection. > Unsmoothed means corrected , with more > representative prices , hence returns. Corrects > appraisal data returns and dispersion numbers. I get that, Totally. I get it… But why are they calling one “unsmoothed” and the other “uncorrected” and the “uncorrected” which IS DIFFERENT FROM UNSMOOTHED, is the one with the bias…implying uncorrected=smoothed, when to me, uncorrected would mean unsmoothed… My question is no longer what is smoothing, it is why the question is phrased that way…is it errata and they should say “corrected”?

uncorrected = smoothed artificially or by following a process/formula resulting in smooth data corrected = UN-Smoothed , resolving for insufficient range or dispersion , correcting returns Note the contrariness of the wording , smoothed = uncorrected, unsmoothed = corrected

Ok, makes sense…thanks for the patience guys

Why if they are the same index and one is “Smoothed” and Un-Smoothed" would they have different returns? I understand that if you look at an index 4 times a year vs daily the Std Dev will change but at the end of the year the same index would give the same return. Anyone have an idea why has the unsmoothed index has shown higher returns?

Because the NCREIF index is not based on real transaction but on (mostly) owner’s estimate on a quarterly basis, they suffer (at least) two problems - Not based on real-time data - Autocorrelated: R(t) = alpha + beta*R(t-1) To fix them, there are several methods to correct for them - Use more frequent market-based/transaction-based data using econometric modelling (like the MIT- CRE index) or - First order autoregressive process to remove the autocorrelation: r corrected(t) =r(t)/(1-beta) - beta/(1-beta)*r(t-1) Those methods are complex and based on statistical modelling, thus are NOT ‘the same’ as the original ‘unsmoothed’ NCREIF longer. Their stddev is much higher. The expected mean is also (slightly) different from the unsmoothed mean. Don’t think you need to know/understand more than this for the exam.

Underestimated the true volatility.

Its really helpful.

Thanks.

Hye guys, An example I’ve always found very helpful to grasp the concept of smoothed data is the following: Imagine a real estate entity (a huge building) priced only once a year via an appraisal, but a valuation is communicated to investors every month. In this case every month the valuation report will referred the last appraised value (which is computed only once a year). It means that the value each month will be exactly the same --> no monthly volatility = smoothed data. Now imagine the exact same real estate entity ( same huge building) but securitized via a listed Real Estate Investment Trust representing 100% of the ownership of this building and which shares price is determined by the market (Bid/Ask pressures) on a daily basis --> each month the valuation communicated to investors could change --> monthly volatility = un-smoothed data Very logic tough

1 Like