In synthetic real estate, we trade property derivatives that are valued based on a real estate index. The index value represents real property values based on market forces for a country or specific region. It is not possible to gage individual property value from an index, but it is possible to get a sense of regional property values in relation to subsequent periods - a benchmark. In effect, synthetic real estate does not rely on a single property but all properties represented by a real estate index.
Today there are real estate indices that cover a broad range of commercial and residential real estate in various geographical regions and countries. In the United States, there are three main indices. The National Council of Real Estate Investment Fiduciaries Property Index (NPI) for commercial real estate, and for residential real estate there is Radar Logic's RPX and the S&P Case-Shiller indices.
The need for an index was felt by commercial real estate investors back in the 1970's. Stocks and other financial investments had broad time series of data which to study. Real estate did not have a reliable index of property value, probably because of the difficulty inherent in property investment. Buildings are mostly heterogeneous, and "trade" infrequently. Making an index would require that all properties would need to be standardized - no easy task. Undaunted, the real estate investors set about to create an index. Their results would culminate in the launching of the Frank Russell Company Property Index (FRC Property Index) in 1981. In 1982, the NCREIF was established to allow greater industry involvement in the index; they would now jointly collect and publish the index with the Frank Russell Company. Membership in NCREIF grew to include all aspects of the commercial real estate industry - from investors to academia. With growth in membership came more buildings to collect data. Soon, the index's name was changed to the Russell-NCREIF Property Index. Finally, by 1994 the NCREIF was a fully functional association able to collect and compile the index independently. They assumed responsibility for the index and would rename it the NCREIF Property Index; know now as the NPI. Today, the NPI is the main U.S. commercial real estate index.
Data used for compiling the NPI are provided by Data Contributing Members of the NCREIF. The data for qualifying properties are submitted quarterly in compliance with the requirements of the real estate information standards. An independent appraisal is required for each property every three years. To be considered as a qualifying property, the property must be operational with at least 60% occupancy. All properties must be owned or controlled by a qualified tax-exempt institutional investor or agent. Only apartments, hotels, industrial, office, and retail buildings are eligible. At the beginning of the index in 1977, only 233 properties were included with a total value of $580 million. At the end of 2009, the index measures the results of 6,186 properties with a total value of $243.8 trillion.
The NPI represents the quarterly total returns of a large representative pool of income producing investment grade properties. Three components make up the index - income return, capital value, and total value. The Income Return is the quarterly net income divided by the estimated expenses, represented in the formula:
IR = (NOI) / [BMV + (½)CI - (½)PS - (1/3)NOI)
BMV is the beginning market value, CI is capital improvements, and PS is partial sales. Notice it is not an actual income value, but a ratio that takes into account quarterly costs such as partial sales of property (i.e., selling an out parcel of land) and capital improvements in the denominator. Capital value is similarly handled as a ratio to measure changes in property value. In the equation below, EMV represents the end market value. The denominator takes consideration of improvements and partial sales, per the following:
CV = [(EMV - BMV) + PS - CI] / [BMV + (1/2)CI - (1/2)PS - (1/3)NOI]
For the total value we add the income return and capital value.
The equations have been set up to simulate that the property was bought and sold during the quarter being evaluated; they are calculating the internal rate of return (IRR) for the quarter. Capital improvements and partial sales are all estimated to take place mid-quarter. Subtracting 1/3 of the NOI is done in the denominator because the income is received monthly. The reduction adjusts the solution to account for the monthly income payments.
All contributing properties have their income returns, capital value, and total value computed. A weight factor is then calculated based on the market value of the property against other properties in the index. The higher the market value, the greater the properties influence on the index. The weights are used to compile and compute a value for the entire data set. The new values are then added to the prior quarter index value. Initially in the fourth quarter of 1977, the index was set at 100. Since then the above calculations have been made and the index adjusted (See Figure 1).
A particularly interesting pattern exists in the index data. If we plot the income and capital value ratios each quarter, the result is not as smooth as the index would lead us to believe, (See Figure 2). Remember that the index value was set at 100, and then the returns are added to that number. With only the returns, we get a better feel for how the constitute components are behaving. Our next chart shows the quarterly changes in the income return and capital values. The plot is not cumulative; it shows the change in the values for each period. On the chart, we see that the income ratio is not very volatile. Each quarter the income return seems to hover around two percent. The capital value, on the other hand, is highly volatile. Capital values reflect the appraisers and markets opinion of property value. As we can see, they seem to change their minds frequently. What can we derive from this?
When future growth in income is expected, the market values the income properties higher and vice versa. The higher appraised values are shown on the chart when the capital value line is positive. In addition, although the index is weighted for a broad sample of properties, it also shows that incomes grow at an average rate of two percent a year. Interestingly enough, inflation also averages a growth rate of around two percent. Therefore, we see that of the investment grade properties that comprise the index, growth tends to match inflation. It would seem that future expectations might play a large part in capital values since income growth does not seem to change much.
Property derivatives, such as swaps and forwards, are available on the NPI for the total return index. Trading is done over the counter with financial institutions utilizing standardized agreements between the parties. Total contract amounts, also known as the notional amount, are large with trades generally above $5 million. Due to counter party risk, the financial institutions are very cautious to make sure the counter party can live up to the contractual obligations - good and bad. With high priced contracts and credit requirements, NCREIF total return property index trades are made primarily by larger companies and institutions.
Other types of derivatives, such as future contracts are being planned. Future contracts would allow smaller investors to take up synthetic positions for commercial property, greatly expanding their investment opportunities.
With efforts going back to 2003, Radar Logic launched the Radar Logic Price Index (RPX) in 2007 for property derivative trading. The index is a composite index that covers residential real estate sales based on current transactions in twenty-five major metropolitan statistical areas (MSA's). All residential properties and legitimate sales are considered to include condominiums, new homes, and foreclosures.
|Los Angeles||Miami||Milwaukee||Minneapolis||New York|
|Philadelphia||Phoenix||Sacramento||San Diego||San Francisco|
|San Jose||Seattle||St. Louis||Tampa||Washington, DC|
The RPX indices reflect current home values on a dollar per square foot basis. Each day transaction data is taken from the appropriate municipality. A small bit of a time lag is in the data since we are seeing data when it is published by the government, not on the actual transaction day. Next, a distribution is calculated from the data to establish the day's index level. The distribution takes the last consecutive three hundred and sixty-five days of data plus the current day. From the day's transaction data and the updated price distribution, the index is computed for the MSA. Utilizing this methodology, the index reveals repeating price patterns through the year that correlates with known "sales seasons". Additionally, the distribution curve is continually updating which reflects the current market.
Looking at the chart (See Figure 4 below), we can see that the property values in Boston and Chicago have a similar saw tooth pattern. Upon further examination, we find a seasonal factor at play. The peaks occur in the summer or warmer months, while the valleys are during the winter. Miami on the other hand, does not have winter weather. No such pattern can be seen. Clearly, weather has an impact on home sales values. It would seem winter weather reduces the number of buyers which reduces demand for homes. Could sellers be willing to take a lower price during winter months rather than hope for a buyer and higher price in warmer months?
The RPX composite index is a composite weighted index of the twenty-five constituent MSA's. After each MSA index is calculated, they are weighted based on per square foot price, building area, and number of transactions. Then the index is computed as the weighted average of the MSA's daily levels.
RPX trading is also done using swap agreements. The swaps are also similar in nature to forward contracts. Trades are done through brokers over the counter since there is not a market place to conduct trading. Contract notional values are still high with $2 million being considered a small deal. Again, this puts RPX derivatives out of range for the small investor - at least in the near term.
The S&P Case-Shiller Home Price Indices are the results of many years of labor by Karl Case, Robert Shiller, and Allan Weiss. The indices track the value of single-family housing in twenty different U.S. metropolitan statistical areas (MSA). Two composite indices are also produced which track home values in ten MSA regions and twenty MSA regions. The major goal of the indices is to serve as reliable indicators of home prices. To meet that goal price changes are measured between similar quality homes. Because no two homes are ever truly alike, a repeat sales method was developed to measure price changes on a home which has been sold multiple times.
|Boston||Chicago||Denver||Las Vegas||Los Angeles|
|Miami||New York||San Diego||San Francisco||Washington, DC|
The indices are published monthly with a two-month lag time for the data being reported. For example, in December they publish the report for the October index levels. To construct the index, each month data is collected from government databases for each MSA. Same home sales know as "sales pairs" are sought out. A sales pair is two sales of the same home. The last sale in the sales pair would occur (actually be recorded at the municipality) in the month that we are collecting index data for. The prior sale could be any time in the past.
To be included in the dataset the sales pair has to meet several requirements. First, the property has to be a single family home. Condominiums, small multifamily and other residential property types are excluded. The property needs to have sold once already which excludes newly constructed homes and pre-development home sales. All transactions are arms-length; no sales between family members or "sweet heart deals" are included. Finally, the transactions must be further than six months apart to filter out "flips" and redevelopment deals.
Why are so many transactions excluded? The indices are trying to measure changes in market price levels. In an effort to compare apples to apples, the data is limited to single-family homes that constitute the majority of the transaction data. Further filtering is done to strive to keep a constant quality for comparison. For instance, if a home undergoes significant renovations or a pool addition the price change in the sale would not be a good indicator for property value. The same house has essentially changed between sales - comparing apples to oranges. Any increase in value would probably be due to the homes structural changes.
After the sales pairs have been acquired and filtered, they are split into price tiers of low, medium, and high priced homes (See Figure 5 below). The tiers price levels are set by evenly dividing the number of sales pairs in each price level - if there are thirty sales pairs then the ten lowest priced are put in the low tier and so on. Only the first sale is considered for tiers. Since we are measuring the change in property value, the first value is our starting point. The second sale price will go to help define the change in property value. Over time, a property may be considered in different tiers merely due to the initial price and distribution of sales pairs. Different tiers of the market vary in price change and volume of sales as supply and demand move about. In fact, the lower tiered properties seem to appreciate in value better than higher value properties. The higher tiered homes tend to track the index due to the data weighing which occurs in the next step. Considering each tier separately helps the indices track housing value movement more precisely, helping to achieve the goal of a consistent benchmark.
Additional factors, know as weights, are now applied to our tiered groups. A mispricing weight is applied to sales pairs to help dampen the effect of very large movements in prices. Large price movements may be due to buyers paying too much, or a home that was severely neglected and bought at a large discount (a good flip candidate). Weights for the sales pair pricing starts at one and will always be greater than zero. The amount of down weighting will depend on the spread between the sales prices. Sale pair price changes are compared to the market price change. The greater the difference between the sales and market price changes, the greater the degree of down weighting for the sales pair. Typically, only 10-15% of all sales pairs in a region are weighted down due to price changes.
The next weight is considered for the time interval between the sales for each sales pair - an interval weight. It is reasonable to assume that the longer the time between sales periods the more likely that a renovation has been done to the house, or that the neighborhood has undergone changes. To account for this interval weights are applied to the sales pairs. The interval weights are determined from statistical models that measure the variance between transactions with respect to the time interval. For example, a sales pair with a time interval of 10 years between sales would be weighed down around forty-five percent when compared to a home that had sold within a year.
From our filtered, sorted, and weighted sales pairs the index is computed. The index is computed monthly using the current and prior two-month periods for the calculations. A three-month moving data set, a moving average. Averaging offsets accuracy issues from the sales reporting delay to the counties. Additionally, it provides a large enough data set to extract better results. Finally, the composite indices are composed of the weighted constitute MSA indices based upon their total values in relation to the total group value. Back in the year 2000, all indices were set at a base value of 100. This time milestone also marks the point at which the computation methodology for the index was changed from a regression type model to a chain weighting procedure - what we have gone over.
Looking at the chart with Boston MSA data (See Figure 7 below), we can see that the S&P/Case-Shiller Indices are much smoother than Radar Logic's RPX. The averaging of three periods tends to smooth out the index. Comparing the RPX to Case-Shiller is not possible since Case-Shiller is an index of price levels - not an actual price, however, the values do seem to follow a similar path. In fact, the correlation between the two data sets is 0.97.
For trading, the S&P Case-Shiller has a broad range of property derivatives types to trade. Actual futures contracts are available through the CME Group. Options are available on futures, but are rare and hard to come by. Currently, the S&P/Case-Shiller is the main index for trading in U.S. home values.