Private Markets Pathway · Private Real Estate Investments

Appraisals, Indexes and Performance Measurement in Real Estate

Q: What is appraisal smoothing in real estate?

It is the tendency of appraisal-based values to change gradually because appraisers rely on past data and anchor to earlier values. This makes reported returns less volatile and less correlated with other assets than true market returns.

Q: What is the difference between an appraisal-based and a transaction-based index?

An appraisal-based index uses estimated values of properties, so it lags and smooths. A transaction-based index uses actual sale prices, so it reflects the market more quickly, but it depends on how many sales occur.

Q: What is the difference between a hedonic and a repeat sales index?

A repeat sales index follows price changes of the same property sold more than once. A hedonic index uses regression on property characteristics across many sales to isolate the price change over time.

Q: Does unsmoothing raise or lower volatility?

It raises it, provided the smoothing weight is below 1. The unsmoothed series shows larger swings and higher correlation with other assets, giving a more realistic risk estimate.

Updated 9 October 2026 · Fact-checked

Private real estate rarely trades, so its values often come from appraisals. Appraisals lag the market and smooth returns, which understates volatility and correlation with other assets. Transaction-based indexes (repeat sales, hedonic) use actual sales. You must know how each index works, its bias, and how to adjust for it.

Understand Appraisals, Indexes and Performance Measurement

Listed stocks trade every second, so prices are observable. Private real estate trades rarely, and no two properties are identical. So most return series for direct property rely on appraisals: an expert's estimate of value, often once a quarter or once a year.

Appraisals create two problems. First, they use past data such as earlier comparable sales, so they lag the market. Second, appraisers anchor to the previous value and adjust only partly. This is appraisal smoothing. The result is a return series that looks calmer than the true market. Measured standard deviation is too low, correlations with stocks and bonds are too low, and the Sharpe ratio looks too good. Peaks and troughs are also delayed and flattened. A portfolio optimizer fed these numbers will over-allocate to real estate.

There are two index families. Appraisal-based indexes (for example the NCREIF Property Index in the US) aggregate appraised values of properties held by institutional investors, plus income. They suffer smoothing and lag. They also cover only properties held by contributing members. The NCREIF Property Index measures unlevered property-level returns, so it does not reflect the returns of leveraged investors. Transaction-based indexes use actual sale prices. A repeat sales index tracks price changes of the same property sold more than once. A hedonic index uses regression to estimate how features (size, location, age, quality) drive price, then isolates the effect of time.

Repeat sales uses real prices of identical assets, but it needs many properties that sell more than once, so it discards single-sale data. It can be biased toward properties that trade often, and a property may change in quality between sales (renovation, aging). Hedonic uses all sales, but the result depends on choosing the right variables, and omitted features cause error. Both are less smoothed than appraisals but still depend on sales volume, which is thin in downturns.

Performance is measured as total return: income return plus capital (appreciation) return. For a single period, return equals net operating income plus the change in value, divided by beginning value. For fund-level or multi-period evaluation, the choice between time-weighted returns and IRR depends on who controls the cash flows. Time-weighted returns remove the effect of the timing of cash flows, so they suit comparing against an index. IRR reflects the timing of the investor's actual cash flows, so it suits assessing the investor's own result.

Separately, appraisal smoothing understates risk, and you can address this by unsmoothing the series to estimate true risk.

Key rules to remember

Total return (single period): Total return = (NOI + (Ending value − Beginning value)) ÷ Beginning value; Equals income return plus capital return. Use consistent NOI and values for the same period. Ignores capital expenditures unless you adjust for them.
Income and capital return split: Income return = NOI ÷ Beginning value; Capital return = (Ending value − Beginning value) ÷ Beginning value; The two parts add to total return.
Appraisal smoothing model: Reported return(t) = α × True return(t) + (1 − α) × Reported return(t−1); α is between 0 and 1. A smaller α means more smoothing. Unsmoothing with this model is a possible exam application, but it is not confirmed whether a calculation will be tested, so learn the mechanics and understand what the model implies.
Unsmoothing (de-smoothing) a return series: True return(t) = [Reported return(t) − (1 − α) × Reported return(t−1)] ÷ α; Rearranged from the smoothing model. Unsmoothed volatility is higher than reported volatility when α < 1.
Effect of smoothing on standard deviation: σ(true) ≈ σ(reported) × √((2 − α) ÷ α), assuming serially uncorrelated true returns under the first-order smoothing model; Derived from σ²(reported) = α × σ²(true) ÷ (2 − α). Only an approximation. Equivalently, σ(reported) ÷ σ(true) = √(α ÷ (2 − α)), also approximate. The key point is the direction: unsmoothing raises volatility.

How to solve Appraisals, Indexes and Performance Measurement questions

Use this method for any question on real estate appraisals, indexes or performance.

1Identify what the data source is: appraisal, repeat sales, hedonic, or an actual portfolio's cash flows.
2State the main weakness of that source: lag and smoothing for appraisals, limited sample for repeat sales, specification risk for hedonic.
3Link the weakness to the statistic affected: volatility understated, correlation understated, Sharpe ratio overstated, turning points delayed.
4If a calculation is asked, write the formula, substitute values, and show each step. Check that NOI and value are for the same period.
5If an adjustment is asked, unsmooth the series with the stated α and say what happens to risk and to the optimal allocation.
6Tie the conclusion to the client: how the bias affects allocation, risk budget or benchmark choice.
7Answer the command word exactly: calculate gives a number, explain gives a reason with a link, recommend gives a choice with one justification.

Quickest way: Bias-direction shortcut

When to use it: Use for item set questions that ask which index or data source is better, or what an appraisal-based series overstates or understates.

Appraisal-based: smoothed and lagged. Risk and correlation too low, Sharpe too high.
Repeat sales: real prices of the same asset, but only properties that sold twice, and quality may change between sales.
Hedonic: uses all sales, but depends on the chosen variables.
Unsmoothing always raises measured volatility when the smoothing weight is below 1.
For return math, add income return and capital return, then check you used beginning value as the denominator.

Common mistakes in Appraisals, Indexes and Performance Measurement

Saying appraisal smoothing raises volatility.
Students mix up the reported series with the true market.
Fix: Smoothing lowers reported volatility and correlation. Unsmoothing raises them.
Treating repeat sales and hedonic indexes as the same thing.
Both are called transaction-based.
Fix: Repeat sales compares prices of the same property over time. Hedonic uses a regression across different properties with their characteristics.
Saying transaction-based indexes have no bias.
Real prices feel objective.
Fix: Name the bias: repeat sales has sample selection and quality-change problems; hedonic has model specification and omitted-variable risk. Both suffer when few sales occur.
Dividing by ending value when computing return.
Students use the latest figure on the page.
Fix: Divide by beginning value for single-period return.
Forgetting that appraisal series lag the market.
Students focus only on smoothness.
Fix: Mention both lag (turning points appear late) and smoothing (peaks and troughs flattened).
Skipping the portfolio consequence.
The question seems purely technical.
Fix: End with the effect: understated risk can lead to over-allocation to real estate, so adjust inputs before optimizing.

Worked examples

Example 1

A property had a beginning value of 80.0 million and an ending value of 83.2 million over the year. NOI for the year was 5.6 million. Calculate the income return, capital return and total return.

Show the solution

Income return = 5.6 ÷ 80.0 = 0.07 = 7.0%.
Capital return = (83.2 − 80.0) ÷ 80.0 = 3.2 ÷ 80.0 = 0.04 = 4.0%.
Total return = 7.0% + 4.0% = 11.0%.
Check: (5.6 + 3.2) ÷ 80.0 = 8.8 ÷ 80.0 = 0.11.

Answer: Income return 7.0%, capital return 4.0%, total return 11.0%.

Example 2

An appraisal-based series has smoothing weight α = 0.5. The reported return for the current quarter is 3.0% and for the prior quarter was 2.0%. The reported standard deviation is 4.0%. Assume true returns are serially uncorrelated. Estimate the unsmoothed current-quarter return and the approximate true standard deviation, and state what this means for a portfolio allocation.

Show the solution

Unsmoothed return = [3.0% − (1 − 0.5) × 2.0%] ÷ 0.5.
(1 − 0.5) × 2.0% = 1.0%.
3.0% − 1.0% = 2.0%.
2.0% ÷ 0.5 = 4.0%.
Approximate true standard deviation = 4.0% × √((2 − 0.5) ÷ 0.5) = 4.0% × √3.
√3 ≈ 1.732, so 4.0% × 1.732 ≈ 6.9%.
Interpretation: reported risk is approximately 58% of the estimated true risk (4.0 ÷ 6.9 ≈ 58%). This ratio is also approximate. Under the model it equals √(α ÷ (2 − α)) = √(1/3) ≈ 57.7%. An optimizer using reported data would overweight real estate.

Answer: Unsmoothed return is 4.0%. Approximate true standard deviation is about 6.9%, so reported data understate risk (reported is about 58% of true, approximately) and can lead to over-allocation to real estate.

Exam tips

Know the direction of every bias: smoothing lowers volatility and correlation and raises the Sharpe ratio.
When asked to compare indexes, give one strength and one weakness for each, in a short line.
Show the formula and the substitution in calculations, even though a bare correct number earns credit.
On recommend questions, give the choice first, then one reason tied to the client's objective or constraint.
Read command words in bold: calculate, explain and justify need different answers, and only the number of responses requested is evaluated.

Appraisals, 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.

Appraisals, Indexes and Performance Measurement: frequently asked questions

What is appraisal smoothing in real estate?

It is the tendency of appraisal-based values to change gradually because appraisers rely on past data and anchor to earlier values. This makes reported returns less volatile and less correlated with other assets than true market returns.

What is the difference between an appraisal-based and a transaction-based index?

An appraisal-based index uses estimated values of properties, so it lags and smooths. A transaction-based index uses actual sale prices, so it reflects the market more quickly, but it depends on how many sales occur.

What is the difference between a hedonic and a repeat sales index?

A repeat sales index follows price changes of the same property sold more than once. A hedonic index uses regression on property characteristics across many sales to isolate the price change over time.

Does unsmoothing raise or lower volatility?

It raises it, provided the smoothing weight is below 1. The unsmoothed series shows larger swings and higher correlation with other assets, giving a more realistic risk estimate.