CFA Level II Exam · Overview of Types of Real Estate Investment

Real Estate Indexes and Performance Measurement for CFA Level II

Q: What is appraisal smoothing in real estate?

It is the tendency of appraisals to lag true market value because appraisers rely on past values and comparable sales. Reported returns look steadier than real returns. As a result, standard deviation and correlations with other assets are understated.

Q: How do you unsmooth real estate returns?

Use r*(t) = [r(t) − (1 − λ) × r(t−1)] ÷ λ, where λ is the smoothing weight. This removes the carry-over from the previous period. The unsmoothed series has a higher standard deviation, with a similar mean.

Q: What is the NCREIF Property Index?

It is an appraisal-based US index of unleveraged properties held by institutional investors. Because it uses appraisals, it is subject to smoothing and lagged turning points. For the exam, treat it as the standard example of an appraisal-based index.

Q: What is the difference between repeat-sales and hedonic indexes?

A repeat-sales index measures price changes of the same properties sold more than once. A hedonic index uses a regression on property characteristics to estimate values, so it can use properties that sold only once. Both are transaction-based and avoid appraisal smoothing.

Updated 7 October 2026 · Fact-checked

Real estate indexes track property returns in three main ways: appraisal-based (valuations), transaction-based (actual sales) and REIT indexes (traded share prices). Appraisals smooth returns, so measured volatility is too low and correlations with stocks too low. To solve questions, identify the index type, name its bias, then judge the effect on risk and diversification.

Understand Real Estate Indexes and Performance Measurement

Direct property does not trade on an exchange every day. Each property is unique and sells rarely. So there is no continuous market price, and you need a method to estimate how property values move. The three index families are three answers to that problem.

Appraisal-based indexes use professional valuations of properties held in a portfolio. An example is the NCREIF Property Index (NPI) in the US, which reports returns on unleveraged properties held by institutional investors. Appraisals rely on past comparable sales and are updated infrequently. Appraisers anchor on the previous value and adjust slowly. This creates appraisal smoothing (also called stale or lagged appraisals). Turning points show up late, and reported returns are less volatile than true returns.

Transaction-based indexes use actual sale prices. There are two common methods. The repeat-sales method tracks price changes of the same property sold at least twice. The hedonic method uses regression to estimate the value contributions of property features such as size, location and age, so it can use properties that sold only once. Transaction-based indexes avoid appraisal smoothing, but they have their own issues: sales are infrequent, only a subset of properties trades, and sold properties may not be representative. Repeat-sales indexes also exclude properties that never resell, and property changes between sales (renovation, ageing) can bias results.

REIT indexes are built from the share prices of publicly traded real estate investment trusts. Prices are observed daily, so there is no smoothing, and the data are timely and liquid. But REIT returns reflect equity market sentiment and leverage as well as the underlying property. Short-term, REITs correlate more with stocks. Over longer horizons, REIT returns converge toward direct real estate returns.

Why it matters: smoothing biases risk measures. Standard deviation is understated, correlation with other asset classes is understated, and the Sharpe ratio is overstated. Real estate then looks like a better diversifier than it really is, and mean-variance optimisation will tend to overallocate to it. Analysts can unsmooth the series to correct this.

Key formulas to remember

Unsmoothing appraisal returns: r*(t) = [r(t) − (1 − λ) × r(t−1)] ÷ λ; r(t) is the observed appraisal return, r*(t) the unsmoothed estimate, and λ the smoothing weight on current true return, between 0 and 1. Lower λ means more smoothing. Use the form given in the vignette.
Smoothing relationship (model behind it): r(t) = λ × r*(t) + (1 − λ) × r(t−1); Observed return is a weighted average of the current true return and last period's observed return.
Effect on volatility: σ(unsmoothed) > σ(smoothed) when 0 < λ < 1; Unsmoothing raises standard deviation. The mean return is roughly unchanged.
Sharpe ratio: Sharpe ratio = (Rp − Rf) ÷ σp; Understated σ from smoothing gives an overstated Sharpe ratio. Unsmoothing lowers it.

How to solve Real Estate Indexes and Performance Measurement questions

Use this sequence for any question on real estate indexes in an item set.

1Read the exhibit and identify the index type: appraisal-based (valuations, e.g. NPI), transaction-based (repeat-sales or hedonic) or REIT (share prices).
2Name the main bias of that type: smoothing and lag for appraisal; infrequent, unrepresentative sales for transaction-based; equity market and leverage influence for REIT.
3Check the data horizon. Quarterly appraisal data lags turning points. Daily REIT data reflects the stock market in the short run.
4If numbers are asked, apply the unsmoothing formula with the given λ, using the observed return for the current and prior period.
5Judge the effect on risk measures: volatility and correlation understated, Sharpe ratio overstated, diversification benefit exaggerated.
6State the conclusion in terms the question asks for (which index, which direction of bias, or the corrected value) and check it against the direction logic.

Quickest way: Index type, bias, direction

When to use it: For conceptual three-option questions where you must pick the index or the bias without calculation.

Appraisal-based: think smoothing, so low volatility, low correlation, high Sharpe, late turning points.
Transaction-based: think actual prices but sparse and unrepresentative data. Repeat-sales needs two sales; hedonic uses regression.
REIT: think daily pricing, stock-like short term, leverage, and convergence to property over the long term.
Unsmoothing: volatility goes up, mean stays about the same, Sharpe goes down.

Common mistakes in Real Estate Indexes and Performance Measurement

Saying appraisal smoothing raises volatility.
Students link 'bias' with 'riskier' without thinking about averaging.
Fix: Smoothing averages across periods, which damps swings. Reported volatility is too low; unsmoothing raises it.
Treating REIT indexes as smoothed.
REITs hold real property, so students assume property-like behaviour.
Fix: REITs trade on exchanges with observable prices, so there is no appraisal smoothing. Their issue is equity market and leverage effects.
Confusing repeat-sales and hedonic methods.
Both are called transaction-based and both use sales data.
Fix: Repeat-sales uses the same property sold more than once. Hedonic uses a regression on property characteristics, so it can include single sales.
Rearranging the unsmoothing formula wrongly or mixing periods.
The formula has λ, (1 − λ) and a lagged term, and students use r(t+1) or swap the two returns.
Fix: Subtract (1 − λ) times the prior observed return from the current observed return, then divide by λ.
Claiming unsmoothing changes the average return a lot.
Students assume any correction shifts every statistic.
Fix: Unsmoothing mainly raises standard deviation. Over a long series the mean is about the same.
Concluding real estate is a strong diversifier from NPI correlation data alone.
Low reported correlation with equities looks like proof.
Fix: Reported correlations are biased down by smoothing. Say the diversification benefit is overstated.

Worked examples

Example 1

An analyst studies an appraisal-based quarterly index of unleveraged commercial properties. The index shows an annualised standard deviation of 6% and a correlation with equities of 0.20. A REIT index over the same period shows a standard deviation of 18% and a correlation of 0.75. The analyst assumes the smoothing parameter λ is 0.40. Q1: What explains the gap in volatility? Q2: What is the likely effect on the appraisal index's Sharpe ratio? Q3: Is the 0.20 correlation a reliable measure of diversification?

Show the solution

Q1: The appraisal index is based on valuations that adjust slowly to past values, so returns are smoothed. The REIT index uses observed share prices with leverage and equity sentiment. Part of the gap is smoothing, and part is leverage and market pricing, so the difference is not all smoothing.
Q2: Smoothing understates standard deviation. Sharpe = (Rp − Rf) ÷ σ, so a smaller denominator gives a higher ratio. The appraisal Sharpe ratio is overstated.
Q3: Smoothing also biases correlations with other assets downward, because lagged returns do not move with same-period equity returns. So 0.20 understates the true correlation.

Answer: Q1: Appraisal smoothing (plus REIT leverage and equity pricing). Q2: Overstated. Q3: No, the true correlation is likely higher, so diversification is overstated.

Example 2

An appraisal-based index reports quarterly returns of 2.0% for the prior quarter and 3.0% for the current quarter. The analyst models smoothing with λ = 0.50 using r(t) = λ × r*(t) + (1 − λ) × r(t−1). Q1: Estimate the unsmoothed current-quarter return. Q2: Would unsmoothing the whole series raise or lower its standard deviation? Q3: Which transaction-based method could be used to avoid smoothing but needs a property to sell twice?

Show the solution

Q1: Use r*(t) = [r(t) − (1 − λ) × r(t−1)] ÷ λ.
Substitute: r(t) = 3.0%, r(t−1) = 2.0%, λ = 0.50.
Numerator: 3.0% − 0.50 × 2.0% = 3.0% − 1.0% = 2.0%.
Divide by λ: 2.0% ÷ 0.50 = 4.0%.
Check: 0.50 × 4.0% + 0.50 × 2.0% = 2.0% + 1.0% = 3.0%, which matches the observed return.
Q2: Unsmoothing removes the averaging, so it raises the standard deviation.
Q3: The repeat-sales method needs the same property to sell at least twice.

Answer: Q1: 4.0%. Q2: Raise. Q3: Repeat-sales method.

Exam tips

Always name the bias and its direction: smoothing understates volatility and correlation and overstates the Sharpe ratio.
When a vignette gives λ, use the formula from the vignette and verify by substituting back into the smoothing equation.
Distinguish what is wrong with each index type. Appraisal: smoothing. Transaction-based: sparse, unrepresentative sales. REIT: equity and leverage effects.
For asset allocation questions, remember that smoothed real estate data makes mean-variance optimisation overallocate to real estate.

Real Estate Indexes and Performance Measurement in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

Real Estate Indexes and Performance Measurement: frequently asked questions

What is appraisal smoothing in real estate?

It is the tendency of appraisals to lag true market value because appraisers rely on past values and comparable sales. Reported returns look steadier than real returns. As a result, standard deviation and correlations with other assets are understated.

How do you unsmooth real estate returns?

Use r*(t) = [r(t) − (1 − λ) × r(t−1)] ÷ λ, where λ is the smoothing weight. This removes the carry-over from the previous period. The unsmoothed series has a higher standard deviation, with a similar mean.

What is the NCREIF Property Index?

It is an appraisal-based US index of unleveraged properties held by institutional investors. Because it uses appraisals, it is subject to smoothing and lagged turning points. For the exam, treat it as the standard example of an appraisal-based index.

What is the difference between repeat-sales and hedonic indexes?

A repeat-sales index measures price changes of the same properties sold more than once. A hedonic index uses a regression on property characteristics to estimate values, so it can use properties that sold only once. Both are transaction-based and avoid appraisal smoothing.